To be clear, I was referring to the article itself.
But yes the performance is also a surprise for me after going through the article. I expected a nice performance uplift, perhaps a bit faster than Ryzen 5000 but with a higher power draw as a trade off for being on the same Intel 14nm node still. Definitely did NOT expect it to be slower.
To be clear, I was replying to you to get a reply at the top of the comments. Completely self-serving manipulation of AnandTech's comment section.
Hopefully, terroradagio is right and an update will improve performance and efficiency slightly. Also, I wonder how the 11900K will look going ~200-300 MHz above this.
I have a feeling that the additional power draw required to get that extra 2-300 MHz is going to be well out of proportion to the performance gained by it. This chip is already pushed past the edge just to not regress (usually) over the previous generation.
Considering this chip is already drawing way too much power the 11900's are probably binned like crazy and run a much lower voltage. Even then it'll probably still use more power than the 11700k. I think they should've just accepted that 14nm was not going to work for this backport. There's a market for cometlake in gamers but what market is there for this? Niche scientific computing?
You're still looking at like an 80W difference and a requirement for the best air cooling available if you don't want temps to be out of control. I might be willing to tolerate the extra power draw if I got more performance out of it, but that's not the case.
No, power consumption is exactly as bad as people are "crying" about, this chips is a joke, hotter, slower and more expensive than even the quite Sub-Par cometlake chips on the market, nevermind the superb Zen 3 offerings, which are starting to see increased availability with the single CCD 6 and 8 core 5600X and 5800X
To even be remotely appealing the 10700K need to be cheap, sub $350 cheap, and thats just not gonna happen
I'm still struggling to find a 5800X at a sensible price and now waiting to see how the USB connectivity issues with 500 series motherboards is resolved.
350 Dollar you want a 10700k and comparing it with a 5800X? Why? The 5800X don t have a gpu, and the 10700ks was under 320 euros allways! Now the 10700K not KS cost 316 euros! Hot a 14nm Chip this is the advantage its only 60 c and a 5800X comes easy over 85 c and more.
More expansiv a 5800X cost 440 euros! A 5600X cost 350 euros, more then the 10700k! Slower its a K CPU and can boost it up. When you OC the 5800X you have less fps.
Why you think it needs more power? Becouse the 40 sec boost? That is all and under AVX512 is it the only cpu that have it. Games and other stuff its the same power, And 5800X dosen t stop at 105 Watt it goes like 118 watts. Thats are very close too 125 watts and more is only a limit time!
@Jadi - that last bit of your post makes no sense. The 10700K and 11700K objectively require more power than the 5800X to do the same amount of work; the 5800X only hits 118 Watts under the same conditions that the other two approach 200 Watts. If you're comparing power draw during gaming, the 5800X uses less power than those two - roughly 25W less than the 10700K.
Comparing temperatures is daft too. 85 degrees is way under the CPU's junction temperature, so it's not a problem to runa cPU at that temp - and it's objectively easier to keep an AMD CPU at that temperature than to keep an Intel chip at 60 degrees, because your cooler isn't dissipating as much heat.
10700k is $320 on amazon and 9900k is $250 at microcenter.
At this point 5900x is best for serious users and 9900k/10700k is for gamers. 5600x and 5800x are overpriced for what it is, but 5900x is a sweet deal at $550
With gaming it's not the power consumption that's the problem here, it is the gaming performance in these benchmarks against comet lake obviously due to a higher latency. Zen 3's efforts have been put into perspective here, those engineers did quite a commendable job.
Agreed! The Zen architecture is top shelf. But it took far too long. My last AMD Cpu was an Athlon 1800 because at that time AMD was the value leader and Intel was the performance king for a price. At the time came where I needed performance over value. So I have been Intel since then. Time to start looking towards Zen 🤔
If I may be so free: "Took way too long": I agree. I just keep in mind that it took them way too long, because Intel managed to create some very shady deals with retailers, effectively pushing AMD out of the market and to the brink of bankruptcy. So, for years AMD has been struggling to stay alive.
Agreed, buying ATi didn't help either. But without those shady deals they might have pulled it off. From 2006 to -say- 2015 AMD was only struggling to stay alive. I believe they had some huge cash injection at the time, together with getting one of the best CPU designers in the world (what's his name?) back in the house.
So, in my opinion, AMD is forgiven for their 10-year "heart-attack".
I also believe it was because the K10, or Phenom, while being faster than the Athlon 64, was not aggressive enough. It had already been designed before Core 2 Duo came out, and by then, Intel was ahead of them. Who knows, they might have narrowed the gap if they had persisted with the K10 design, widening everything till it matched/surpassed Nehalem, but they shot too far, going for Bulldozer, which was supposed to turn the tables round but sank them further in the ditch.
And yet AMD was having trouble supplying enough chips to the limited market they did have. Much like today. OEMs are not going to sign up with a company that cant consistently supply CPUs.
Dont forget the billiosn they poured into making GF and pissing off TSMC in the process.
I predicted ~300W peak power about a year or so ago when I first heard they were making the big mistake of bringing AVX-512 to mainstream consumer processors. Why on earth? There was a valid, wise, very, very, very good reason Intel had reserved AVX-512 to just their 14-nm HEDT processors and that had always been the furnace-like heat and nuclear-like power consumption of it. Even with the best logical design improvements from a new microarchitecture, it is still an extremely intensive logical pill of a task to swallow. Now, we see that reason in full, unadulterated display. You bring a massively complex instruction set extension to a higher process node where you have far lengthier physical networks (meaning essentially longer wires, increased resistance, higher power, and maximum heat) and, of course, you are going to have a steaming pile. Remember this review is only looking at the number two product, the Core i7-11700K which has lower clocks and lower power draw. The Core i9-11900K will likely need a 360- or 420-mm AIO just to not thermal throttle like mad. My 5950X with its meager 240mm AIO (Corsair H100i RGB Platinum) that runs at the quiet mode setting is laughing its butt off right about now. When Ian Cutress had to use an obnoxiously loud 170 CFM fan (I have used 100 CFM Deltas and those already annoy most PC enthusiasts) on a massive 4-pound, full copper heatsink to tame the 11700K's 290W, I shudder to think. Will the 11900K be outdoing the FX-9590's record-making peak power draw of 350W (see here: https://www.anandtech.com/show/8316/amds-5-ghz-tur... ? Ian easily could have gotten the 11900K also at retail, but I think he is holding back on that because he already knows the 11900K is going to be a throttling disaster and only the 11700K is an ACTUALLY USABLE PROCESSOR. *mike drop*
I think they did they realistically could given the contraints they have. We are not going to see intels true potential until they get thier manufacturing process fixed or they swallow the dead rat and go to another foundry. This will be enough to keep them in the game for another year.
Main defect of the article: AMD best of the best (low availability) versus a common high volume medium level Intel SKU. No matter the price that will change on retails or OEMS.
I don't understand AMD, they pump hard the pedal still they can not do much to gain market share. New processes are medium volume and with too much customers. The NEW AMD will be a company capable to deliver a pile of good dies to all channels, without limitation on volume. Unfortunately Lisa follow the wrong street, in this manner Intel will always dominate the market. Bet 2021 will be Lisa last year at AMD.
How is Ryzen 5800x AMD's "best of the best"? It clearly is not but just about somewhere in the middle. If you want AMD's best consumer CPU, you gonna look at the 5950x. From the perspective of AMD, completely outsourcing manufacturing was the only way to reap the benefits of latest and greatest process nodes. Now, that everybody else, incl. the automobile industry, the consoles, Apple, and even Intel are booking production capacities at TMSC, has certainly contributed to Reaching capacity limits and thus to AMD CPU shortages. It is predicted that shortages will be mostly (hopefully) sorted out by this summer. But yes, now that AMD are earning quite a lot of money they should buy some TSMC stock and try to partner up, getting more production capacity in the future.
did you just forgetting why AMD spin-off it's fab on the first place?
we don't know exactly how much AMD put out their desktop chips from TSMC plant, and as far as we know, the market are still growing in size. while AMD market share are still a long way from reaching 50%, they still have their processor sold out all over the place.
honestly I don't get what you're trying to say here.
Actually we do know, they made ~1M Zen 3 chips in Q4. 140M PCs were shipped. Based on their market share, about 3% of the chips AMD shipped in Q4 were Zen 3.
I think adored tv did a few articles, videos on this. As I understand it thier main issues are they use a monolithic die (everything on on a single die) that gets harder to make as the process shrinks, IE more dies have defects... AMD got around this by going chiplet, lots of small dies which meant more of them are good. Until Intel chnages to chiplet or they find a way to improve the manufacturing process to lower the defect rate then they will struggle.
Class action for what, excessive power bills? :) Imagine using this during a heatwave in an airless room. Since it doubles as a heater you would need to have the AC on all the time. If you have no AC there will be a competition between who dies first, you or the processor? ^.^
I would be very cautious testing or believing results with the Z590 platforms long before Rocket Lakes official release. Ive tested 3 of those boards with my Comet Lake (Asus, MSI and ASRock) and they all had pretty bad BIOS versions still, with PCIe/IO performance being low in SSD 4K benchmarks and getting weird frame time stutters from time to time (only noticeable when actually playing or looking at a realtime graph). Not to mention Intels drivers are still bad as well. On none of them even very basic features like the sleep state worked! Comparing this review with user benchmarks in German forums shows huge differences, so theres not much to add to this.
That said, I have to laugh when Americans or people from other countries with cheap power complain about the power draw. And seeing fanboys downplay the performance of AVX and ignoring that it was always power hungry, even 6 years ago, is another obvious thing. Without it RKL actually runs pretty cool for being ancient 14nm.
And of course I love the geniuses who still think that you cant use the iGPU without it being connected to a monitor, or dont know about the new power saving/GPU switching feature. Not that this article didnt fail at pretty much everything, incl. explaining things like that.
"That said, I have to laugh when Americans or people from other countries with cheap power complain about the power draw." Electricity is expensive where I live, but that's not why I want low power consumption. The more power it consumes, the louder the cooling solution will be. That's why my last system was Intel (Ivy Bridge) and my current system is AMD (Zen 3).
It's not about the cost of electricity. High power draw typically translates to a lot of heat. My PC is on the upper floor and heat accumulates in my room and it gets extremely hot while gaming in the summer, even with AC on.
As you stick to the same process node and continue to crank up the frequency, it gets hotter and hotter and hotter. Skylake didn't run even close to the temps that these new CPUs run at.
And yes, even Zen3 produces a lot of heat when under load.
How did you conclude that there exist thermal hotspots? Is it an educated guess or did you actually measure a temperature profile? If it was a measurement, was it a thermal image of the socket area of the rear PCB, multiple thermal probes? BTW, your argument does sound logical.
Twice the power for being slower? If you are referring to the 290W power consumption with AVX-512 test, I'm desolate to inform you that a 7nm CPU with twice the core would not reach those performances and perf/W in that test.
If you are talking about the 140-160W usage at other "normal" tests, I'm desolate to inform you that a 16 core 7nm CPU does not consumes 80W.
So stop vomiting meaningless numbers. This is a 14nm CPU and for the process it is build on it is doing miracles. If Intel could ever use an advanced PP like those 7nm by TSMC Zen would still be the underdog.
For the future just hope the TMSC 5nm are good, early, low cost and really high yielding, because if Intel comes with a decent 7nm I think AMD will not look all that advanced (6 years to surpass a 6 years old architecture and all by the use of a more advanced PP that unfortunately doesn't allow for great deliveries).
"So stop vomiting meaningless numbers. This is a 14nm CPU and for the process it is build on it is doing miracles. If Intel could ever use an advanced PP like those 7nm by TSMC Zen would still be the underdog."
Not necessarily. Intel apparently is still behind in IPC/single thread performance, as evidenced by that Cinebench results, so whilst 7nm would let it run with less power (theoretically), it still would lose to its main competitor, the 5800X.
You have missed that with 14nm die area you cannot improve the architecture that much. You are still thinking that CPU designed for 7nm, with all the advantages that they would bring, would still be like Skylake which is a 6 years old architecture A PP like TSMC 7nm would bring a completely new architecture that would blow Zen away.
Zen is good because it is based on such a better PP that those Intel has now, but it still struggles at beating Skylake. And to do that it, that is by using such an advanced but production limited PP, it has sacrificed high stock delivery right in the period where demand is much higher than supply. Intel can fill the remaining market with whatever it has, being it 9xxx, 10xx or now 11xx generations.
Yes, Intel probably would beat AMD with an imaginary all-new architecture on TSMC's 7nm process. Similarly, I would be a hell of a basketball player if I were a foot taller and had any motor skills.
Here in reality, Intel has worse performance and worse efficiency. As a consumer, that's what matters to me. What Intel could do with a bunch of "ifs" is irrelevant. I haven't owned an AMD CPU since my Athlon 64 was replaced by a Core 2 Duo. But there's no way my new build this year isn't going to be AMD.
Yes, what counts is the results, you are right. But by that I can't cry for a miracle when I see Zen 3 results as with a much more advanced PP it just can win over Skylake for a few % and all the real advantages it has is smaller power consumption due to the much better PP vs this one 6 years old.
If you look at the real power consumption, that is not the one with AVX-512 tests where RKL disintegrates Zen for perf/W despite the high power requirements, you'll see that this chip is not that power hungry (though being more power hungry than Zen) and that does make me think that with a better PP this same architecture would be another thing completely, as are the 10nm Tiger Lake which however suffer the not so good power consumption at higher frequency required by desktop SKUs.
As we are not that distant from finally having something decent that is not the 14nm PP, I will really not call my thought "imaginary". AMD will not be able to pass to 5nm so soon, and seen what this architecture can do, despote the 14nm PP, I think that the future is going to be more interesting that what you hope it ti be (that is, AMD keeps on figure it has better CPUs while not having them in the shelves but what counts for you are.. yes the results... and for these Intel is outselling AMD 5:1).
"A PP like TSMC 7nm would bring a completely new architecture that would blow Zen away."
Based on what, exactly? We've seen die shrinks before without amazing architectural advances from both Intel and AMD. You have an awful lot of confidence in something that doesn't exist.
Zen2 was already faster than Skylake and its derivatives clock for clock by about 7%. While Comet Lake had higher single threaded performance than Zen2, it did so by throwing efficiency and power draw out the window and going for absolute performance. That made it such that Comet Lake could compete in ST applications but it still lost on MT applications against the same thread counted AMD CPUs. Going for absolute performance has been a double edged sword for Intel as the newer architectures hadn't been able to clock as high. Despite the higher IPCs of the newer architectures, absolute performance was no better than a wash due to 20% lower clock speeds.
Zen3 now has absolute performance dominance over any Skylake architecture CPU. It doesn't "just" beat the older CPU, as in like 2% faster. It is upwards of 20% faster clock for clock and 10%+ faster in absolute ST performance.
"Based on the fact that with a 7nm PP AMD still struggles at beating Skylake architecture which is 6 years old and was born on... oh yes, 14nm."
Struggles? The slightly older AMD chip (5800X) beats the newest and greatest out of Intel, whilst consuming less power, AND hitting lower peak turbo speeds.
That is complete domination. You could make the same argument about how Intel hadn't even made any significant gains over Sandy Bridge until Skylake, and that was *2* die shrinks.
I'm not sure if you're being wilfully obtuse or ignorant.. the only reason Skylake is even remotely in the game is that intel's 14nm is refined enough to allow them to push raw clock speeds to the moon. Do you not recall how awful Ice Lake was because it couldn't clock? TGL is starting to clock a bit better but it's still pretty damn close. This is on 10nm "superfin" which is ~= TSMC N7(P).
So Intel don't have some magic engineering pixie dust that would propel them beyond AMD if they were on the same node.
Intel already have process equivalent to 7nm – 10nm SF. And they already designed a new architecture on it: Tiger Lake. And Zen 3 is perfectly competitive with Tiger Lake.
Ultimately the latest desktop processors from Intel doesn’t perform well against AMD that’s what it is. They chose to release it on 14 nm as their 10nm was still work in progress. The numbers have meaning and not your conjecture about Intel using TSMC advanced node- it’ll be compared whenever that happens, with numbers.
But we have to live in reality that they don't even have 10nm ready for desktop. Fantasies about creating an alternate reality where their core architecture exists on a smaller node for desktop are just that, fantasies. The reality is AMD clearly has the far better product right now aside from niche edge cases.
I still agree with the conclusion though that given current circumstances, get what you can get if you need to upgrade or build new. But the reality there seems like the 5800X is available at MSRP in-stock at multiple storefronts.
A chip that is just released, the best Intel currently has to offer for the mainstream consumer, can't match a chip that has been out for months. While using more power.
Thats not a good look for Intel. I hope the 11900K (or whatever they're going to call it) at least matches the 5900X in games.
This is the first time in a long time, with generations of chips current, that I cannot think of a single reason to recommend Intel's latest and greatest over AMD.
The point being, the 11700k doesn't even catch the 5800X, which has been out for a few months already. Given that this was supposed to be Intel's "response to Zen 3", its pretty disappointing.
Availability of Ryzen 5000 except 59xx parts is already a non-issue. You can get 5600X with a few days delay at worst, and 5800X is in abundant stock pretty much everywhere.
The key is price, especially the platform price because Intel MBs are generally more expensive. On top of that you absolutely need a larger cooler, and most likely also need a beefier PSU for the Intel CPUs, so the CPU price for the intel parts have to be substantially lower than a performance equivalent AMD part to be competitive. And given the history of intel that seems very unlikely to happen.
blppt - my concern is that AMD may have a superior IPC, but the real fruit comes from the manufacturing process. Intel is still (somewhat) competitive at 14nm and that in itself is quite unbelievable. Imagine where this chip would be on 7nm or 10nm, at 6GHz+ and more cores with 2-3x the cache.
That said, this victory may be short lived because AMD is basically taking advantage of the embarrassing execution Intel has repeated, much like they did 20 years ago with the P4 (albeit that was an architecture failure, not a manufacturing process failure)
Intel's latest 10nm process delivers *slower* clocks than its 14nm one. So, no, 6 GHz is not on the table. I imagine that when the transition to 7nm, Intel will be able to achieve moderately faster clock speeds than with 14nm.
10nm SF is good enough for 5 GHz. 10nm ESF can clock higher, so Intel's latest (but unreleased) process should match 14nm. I would not expect 7nm to clock higher than 14nm because it is becoming very clear that 5Ghz+ is just a waste of power and transistors, so i would not expect 7nm architectures to be designed to clock higher. We either are getting lots of IPC or just over 5GHz.
That makes this even sadder. This port was actually worthless. An i5 quad core might be manageable by a cooler that fits in an average case, but that's about it.
Would've been better off releasing Tiger Lake 35W processors on an LGA package.
The 6C/12T i5-11500 should be "fine," as it has the same 32CU Xe iGPU as the higher end parts. The part below it, the i5-11400, is also 6C/12T but has a cut down 24CU Xe core.
The Xe upgrade is a nice change, at about a 33% uplift over the previous "Gen11" iGPU, but it's still just what I'd call "passable" for light gaming. Anything above 1080p and you'll want a discrete GPU for the best experience, and that goes for both teams.
I understand what you're saying but I still think the 6-core i5 is going to draw a boatload of power.
Obviously for gaming a dGPU is preferable. In my experience with AMD's Ryzen 2400G, though, it isn't actually too stable with medium resolution monitor configurations (think 1080p + 1440p) and isn't all that well supported by drivers.
On the Intel side, uplift over Gen11 is cool but uplift over Gen9 is where it gets noticeable, and important. Gen9 was fine when it came out but that was quite some time ago. Shows its age when too many apps want GPU acceleration with that same multi-monitor setup I described.
Sorry, didn't complete my thought. Anyway, point is that a decent % of Intel's sales are for business desktops (maybe not right now, but, ya know, offices might open up sometime this calendar year, before Alder Lake ships in volume). Rocket Lake i5 would be perfect, if its real-world power consumption wasn't out the roof. Good enough graphics for multi-monitor, fine performance elsewhere.
I guess that makes me wonder what performance would be if limited to a 65W avg/95W peak thermal output. That's about what those small Dell towers can handle.
I honestly don't expect the i5 models to have the same outlandish power usage characteristics as the reviewed i7, but will definitely be reading any day one reviews that may pop up. The existing i5-10400 is a pretty compelling product (and priced well), and if the 11400 or 11500 can manage to fit into the same ~100W PL2 envelope I think it'll find a home in a lot of desktops. For OEMs, they're sometimes board limited to the 65W PL1 via a BIOS option, and I'd expect that to continue with the 11th gen versions.
If they price the 11400 close to the 10400, it'll be a solid choice.
Not really. AMD changed the way they report boost frequency numbers. Before it was "up to xxxxMHz boost", now it's the more wordy "given adequate cooling the boost frequency is _at least_ xxxxMHz". This change was driven by the stick they got for the 3000-series only very rarely hitting the listed boost frequency. Now you can generally assume to get 50-100MHz _above_ the listed boost frequency if you have a half-decent cooler.
The power consumption you comment shows while using AVX512 on hand optimized test. Your idea about Alder Lake to solve this is not the correct one as Alder Lake itself will not implement AVX512 ISA at all. IMHO very bad decision by Intel again.
Not likely, but it'll at least beat the 5800X and probably go even on efficiency. The real upside is in the server space and laptop space. I expect Alder Lake to do excellently in both of those segments.
Alder Lake's Golden Cove cores should have a decent IPC improvement over Rocket Lake, so 8 of those cores should be able to match more than 8 Zen 3 cores. Then throw in the 8 Gracemont Atom cores which will be better than Tremont. 8+8 should top 5900X but not 5950X in multi-threaded, and beat Zen 3 in gaming.
There's caveats, perhaps related to DDR5 or schedulers, but I will be surprised if the top Alder Lake chip can't beat the 5900X.
It might be able to beat the 5900x, but by the time you add in Intel's overpriced motherboards (have you looked at Z590's recently?!) and the premium of DDR5, you're going to be at 5950x+ pricing.
Traditionally Big.Little designs don't work that way. They either are running the big 8 cores or they are running the little eight cores but not at the same time. The type of workload determines which is run when. Personally don't think it makes a lot of sense In desktop.
And whats the point of the little chip big chip design, when TMSC will very shortly produce in mass the 5nm and in 2023 the production will move to 3nm.
Alderlake is based on the worse case scenario and has been introduced to buy Intel time until it resolves the shortfalls of their 10nm production.
You're really confused if you think Atom doesn't help Intel here. Tremont performance per watt and performance per die area is really quite excellent. Also worth remembering that nearly every Atom you've ever seen has had mediocre memory and cooling paired with it. I don't expect Intel to "win" off of this move but it'll help for as long as Intel doesn't have chiplet ready.
Even if it does - and Intel's current record on 10nm suggests it won't - by that time AMD will have had over a year of Zen 3 reining unopposed, and Zen 4 well on the way.
It is incredible bad form and bad taste to release a review before anyone else and before Intel has provided the new microcode update to resolve the early issues. All because Anandtech wants to get out early.
And your defense by way of saying well we got it at retail and therefore this doesn't matter is a joke. Terrible publication.
I'd be saying this if they did with an AMD part too. This is something I would expect from someone unknown guy doing reviews, not Anandtech. Extremely disappointing they are going for the click bait.
We used to have performance previews all the time back in the day to let people have an idea of what is coming and if it is worth it to wait for the next generation. Consider this to be a performance preview, meaning final performance will be higher. In general workloads rocket lake is 10-15 % faster while being reasonable with power. In gaming it's trash, but that could be a consequence of the non-finalized nature of the product.
By the time someone can reasonably buy this product it will be April/May. If I need a pc now, the performance uplift isn't that high to make me have buyers remorse. If I can wait, I can get a comet lake chip at probably a great discount, or if rocket lake's gaming is fixed, get that instead.
Performance previews like this are excellent journalism because the public is informed of things that are otherwise concealed. My only gripe with the article is calling it a "review" instead of "early review" or "preview" because someone may stumble on this article and mistakenly believe this is how the product will be shipped.
This is a REVIEW! Not early review or preview or any excuse you're making. You've lost touch with reality if you hope for a higher performance on official release date. If Intel haven't released these CPUs (before their announced release date), no way in hell Anandtech or anyone would be able to purchase them.
I'm amused by how many people have stumbled upon this and are mistakenly assuming that performance will be significantly better when it's shipped. This was a CPU bought at retail running on a retail motherboard. If they do noticeably improve performance (and it's a big if) consumers will have to install updates themselves to get it.
This is not an engineering sample. If hundreds were being sold by a retailer then that is what Intel was intending to sell to its clients. If you are not satisfied with the performance don't blame it on Anandtech
Unless you have something like the X570 and Zen2 being released at the same time, quite often it is a Beta BIOS that gives initial support for new chips. Eventually with further validation that Beta BIOS becomes official. Perfect example of this is in October 2019 I was installing vSphere 6.7 onto some servers. With the initial BIOS release I was getting the Purple Screen of Death. The only BIOS that was available was a Beta BIOS and on that vSphere installed just fine. Push forward a month later and the official supported BIOS by VMware for that server was the Beta BIOS for vSphere 6.7. It took another 6 months before a non-beta BIOS was available for that server as well. Hence just because it is Beta doesn't mean it is bad. Things in Beta testing generally have full features and performance just hasn't been validated long enough to become the official release.
"Latest beta BIOS from vendor, was told that they don't know when the next BIOS update will be and this contained everything to date. So unless you've got special information.
Note that this is always the risk of doing reviews even on launch day. At some point you have to lock in a BIOS version for published results. Vendors who send BIOSes 24 hours before embargo lift are told to go away."
I mean the leaks suggest that z590 motherboards having some problems resulting in performance regressions. So some poor sod who bought into z490 got to enjoy a fast cpu and can upgrade to an even faster one. AMD is obviously the best for non-avx-512 workloads, but where I am I can't find one for a reasonable price, so Intel is the only viable option. Perhaps the real travesty here is the lack of capacity in TSMC's 7nm node preventing us from buying excellent cpus and gpus at reasonable prices.
the question is, how many things actually use these special avx instructions ?? a handful ? unless you know you can use it, no point in them. seems intel creates these, just so it can win a benchmark.
AnandTech has always honored NDAs, and continues to honor this one. We adhere to the requirements of every agreement we sign, even when doing so is not in our best financial interests. We do this because we're honest people, and just as pragmatically, we need hardware vendors to be able to trust us.
The flip side to that, however, is that retail hardware always has (and always will be) fair game. This was a processor sold by a major European retailer, tested in a motherboard based on a chipset that has been selling at retail for the past couple of months.
Although Intel may not be happy with that retailer over their lapse, at the end of the day this is final silicon running on final silicon. We have done every bit of due diligence both to ensure the accuracy of our results, and to inform the necessary parties in advance about what we intend to do, in case they wish to raise any issues with us.
So we stand by this review both from a technical perspective and an ethical perspective. All of this material was handled in a fair manner that was entirely above the board and legal in all steps of the process.
Using a processor that isn't suppose to be sold is sketchy, plain and simple. And you know full well this. It is not to the benefit of anyone who may be interested in this part. You are thinking about yourselves, full tilt. And that is your choice. You have no idea what could happen with this update they are talking about. Will it be magic? Probably not. But will it have fixes for other things in your review? Perhaps. So therefore you are putting out an inaccurate piece for the purpose of getting out early that may very well be inaccurate in parts at the end of the month. And quite clearly from the comments here, feeding many hungry AMD fanboys.
I'm not mad at you, just disappointed. Enjoy the attention.
”Using a processor that isn't suppose to be sold is sketchy”
The operative word you’re missing is “yet”. The batch of processors this one is from is Intel release silicon intended for end customer hands. And there are more out there. This article represents exactly the kind of performance that at least some day one customers should expect. Now maybe there will be updates to the software stack that will improve performance over time, but that doesn’t make this review any less valid.
Pre-release content is nothing new for AnandTech, and it's interesting to see how passionately some people feel about the topic. Might be something worth exploring in a future article, as I'd wager that there are a lot of readers who weren't around to see things like the original Sandy Bridge pre-review which Ian referenced in another reply.
I think the criticism on this review is quite justified. You are reviewing a device that is not ready to be sold yet. Yes, you could buy it at retail but just because, like you, some other "smart guy" made the wrong (would say fraudolent) move to not respect a date.
What we have here is a "preview" of the CPU performances that may (or not, but you don't know now) change when the CPU will really available for the rest of the mortals on the globe.
I would like to think that you will do a new review of the CPU once the motherboards will be updated and make evidence if, how and by how much something has changed since this preview with what are early samples that results being compatible with the device.
However, while you were at it, you could also try PCI4 connected memory storage to see how good Intel implementation of the technology is.
"You are reviewing a device that is not ready to be sold yet." How is it Ian or Ryan's fault that Intel released a CPU in this condition? If Intel isn't doing any QA on their end for proper use out of the box then the problem is on their end.
And yes, they did release it in that condition because unless they intend to recall all the packages and open them up and replace the CPU inside, it is the product that people are going to be getting and using when they open it up and slapping it in their computer. I used to work retail and getting new product on major releases is usually about 3-4 weeks ahead.
Following your logic then they shouldn't do a review unless they are willing to also update all of the AMD CPU's as well to include their performance and bug fixes which would turn bench-marking into a never ending nightmare because of updates.
You have misses that Intel has not released the product yet and the samples you can find on the market come from a seller that broke the NDA and start selling them before the official release time. They are actually are in beta support with BIOS, microcode, drivers and such on not final motherboards as well.
Once you understand that you'll understand why this is a preview or a beta test, not a full review of the product.
Oh CiccioB, give it a break. That exact CPU was packaged by Intel for retail, it was meant to be sold as is, just a few weeks from now. Yet, that *exact* piece of hardware. You keep implying "it was not ready for retail" as if Intel was gonna start etching this CPU some more to turn it from a hot grill into a cool as a cucumber lightning fast CPU. Intel may tweak it a bit and have a new revision but it's not like they're just getting the line moving now, they've been building stock for a while so retailers are ready to sell this silicon.
But let's be honest, this very CPU that AT put to the test would have been something that an end user would have bought days or weeks from now. A real customer would have used it as is.
The only straw you could grab is that the BIOS might be tweaked until launch. And while it's true, it also probably doesn't work in your favor. I doubt retail MoBos will have CPUs running at over 100C so likely they will limit power more aggressively, and there's very little microcode optimization that one can do to squeeze that much performance.
This is a retail CPU that was sold some days or weeks early. But still retail, and still exactly what consumers will get. Performance won't get better. Power will but only at the price of performance. Stop shilling, unlike many other articles, this one actually made it clear that the BIOS was not final. Any person who doesn't understand that the BIOS can't magically fix this CPU probably don't bother with reading AT anyway.
Thank you very much for this comment. My thoughts exactly. To help change some people's perspectives regarding this review. Look at it as a review for those who bought those RL cpus who are about to buy them from that retailer. Otherwise, for the rest who can't handle such an early review, just look away and wait for the launch date and the reviews which will commence.
I mean testing a whatever PCI4 SSD to see if they are working correctly. My suspects is that the BIOS used on that motherboard was so early that there was not PCI4 support at all and that's the reason there are not those test, which would have been a normal thing to add seen PCI4 is one of the new feature brought by these new CPUs, newer than the AVX-512 instruction set.
PCIe Gen4 support is certainly a welcome improvement over Intel's previous desktop processors, but it can't be considered much of a benefit over AMD's alternatives. We will be investigating whether there are any measurable differences in PCIe 4 storage performance between Intel and AMD hosts, but given how limited the benefits of PCIe Gen4 over Gen3 are for NVMe storage, it's pretty clear that differences between PCIe Gen4 hosts will be insignificant.
@Billy - they're demanding you validate Ryan Shrout's claimed benefits of Rocket Lake over Zen 3 in a synthetic storage benchmark. Who could possibly imagine why 🧐
I don't see how a BIOS update will do much for performance. The motherboard is already running with unlimited turbo so the CPU is pegged at 4.6GHz for MT tests. That is the rated all core turbo frequency.
Ofcourse bios updates will improve. Just like They have improve amd performance... but not buy much. 1% I prove is a big in these changes... bios upgrades Are more to clean up bugs.
He reviewed on available now retail parts, even if a microcode update iss issued, there will still be people using this cpu on same microcode as in this review, as its down to the user to manually update it.
I have always hated specially coordinated review programs where everyone agrees to publish shame time (wtf?), and the reviewers are working with vendor to make sure review doesnt upset them, wild west reviews like this need to be more frequent.
I understand AnandTech has honored their applicable NDA. And that you informed Intel of your intentions or whatever. And that you didn't break any laws. And you also considered the early release OK because the chip was (what would be sold here in the US) retail. But I think I agree w/ User terroradagio and some others in that, Anandtech shouldn't have released their review early because they happened upon a favorable, early deal - (which itself may have been contrary to an Intel company policy w/ the retailer), not available to any other reviewer or consumer. It's taking advantage of a slip in how the system was supposed to work. You don't want to see it as wrong because it's almost like time doesn't really matter. In the end, you're still buying the product, doing the work, publishing and maintaining the website and revisiting the numbers and updating the motherboard and more and more work. You do all this hard work, and you're highly respected, (and for good reason), so for these good reasons, and more, I think this clouds your decision on this matter. I just feel that all tech sites should respect the same release review date! To not do so reminds me of the less ethical journalism methods used by some photographers, who then sell them legally to the newspapers. But integrity goes deep - more than 1 level. The benefit Anandtech COULD take is the one that they have apparently become so used to, that it is assumed. The ability to buy the chip before consumers, test it, write their review, and click the mouse to post it 1 second after the NDA says they could. (My point here is, some reviewers are either not able to buy them early, or not given the chips, don't have the connections to buy them, and have to wait to buy them like any other consumer, to test them, review them and then publish.) In other words, you are already at an advantage over some reviewers by your early access to the chip - and have weeks more than them to test it. Publishing it 3 weeks earlier than your standard NDA (that may not apply), before nearly anyone else, is (in my opinion) an unfair advantage. You are a well established website and reviewer - so I'm not saying you did it for the views. I just feel it's not right. I get it - you must have a different ethical view. Thank you for the review otherwise.
pentiuman, and would you be saying the same thing if another site did this, or were also able to get one of these cpus to test ? or maybe, like others have suggested, some just dont like to see intel in such a disappointing light ?
Every other reviewer out there had the chance to take advantage of this "slip", so it's not unethical.
Unethical would be taking advantage of insider contacts to produce an officially-sanctioned "preview" prior to release of a product and formal reviews that provides a misleading picture of the product's performance, like DF did with Nvidia and the RTX 3080.
I would like to know what you will say if Anantech will do the same with AMD Ryzen 4, that is reviewing it on an early motherboard with a beta BIOS and not yet tuned microcode and it will result not being as fast as you would expect (or hoped it to be) one month head of the actual release date. And present it as an official review of the product.
I would bet you (and your "friends") would go and cry out for a payed article by Intel to make AMD product look worse that it really is "like the good old times when it payed everywhere on Earth to not make AMD sell its products".
It would actually be fine because Zen 4 is already showing 30% performance uplift clock-for-clock, and that's a full year ahead of launch.
Be honest with yourself. Achieving even a 5% uplift with microcode optimizations ahead of launch is a pipe dream. The review successfully shows the kind of performance you would expect. Is there room for improvement over time? Of course, but that applies to any product.
I'm not speaking about the improvements in IPC. IPC is not everything to evaluate a product. If that were true ARM chip would be the winner since early '90s. I'm speaking about the fact that in many tests shwn here this architecture shows worse results than the previous one. That would mean there's something really broken in the architecture or in the SW they execute.
About Zen 4, don't old your breath because 5nm for HP are not that close. Even Intel Ocean Cove is said to be the really new revolutionary architecture that is finally going to show what the new 7nm PP could have really brought if it was available today as we speak. These are speculations, while this chip will be out in less than a month and waiting for the final tuning would have just made a better service to anyone that really wants to know how really it behaves. Not how it looks like on a unknown motherboard with a not updated BIOS and not the final (or even the first version) of the microcode.
This performance from RKL is unsurprising if you had been paying attention. We've known for months that Intel hasn't had great results with RKL and that's why they're pushing for Alder Lake ASAP. Once again, you cannot honestly expect substantial performance uplifts a month before launch. It's possible the ring/uncore frequency was low for this review, however that will only make a significant difference in games, or other latency sensitive scenarios.
As for Zen 4, I know for fact from reputable sources that it's coming around the middle of 2022, with large uplifts in performance.
" I'm speaking about the fact that in many tests shwn here this architecture shows worse results than the previous one. That would mean there's something really broken in the architecture or in the SW they execute. " no your crying cause intel still lost. and that well. rocket lake, isnt the performer it was made out to be. AND it looks like some of the performance regressions, were explained/accounted for, in the review, which you obviously did not read
Originally ocean cove was leaked to be a revolutionary architecture with a massive ipc uplift that can serve as the backbone for future architectures like Conroe did in the original Core2 line. Leaks have later said that Intel cancelled the revolutionary nature of the product and ocean cove is simply going to be another microarchitecture like sunny or golden cove. By that point AMD should have zen 5 and we the consumer can enjoy healthy competition.
As to the point on how it really behaves. We can see from the frequency graphs that rocket lake is turbo boosting as it should. A new BIOS could change power limits which would change behavior, but when rocket lake is already given infinite turbo time, such changes are likely to lead to performance regressions. The only other possibility is that the slight maturation of the bios leads to a small performance uplift. This WOULD BE IPC, but such improvement would be small at best given that z590 (and the similar sunny cove) has been out for a while. At best gaming performance may not suck as hard (or be fixed), but overall the performance improvements should be less than 5%. Not enough to change any conclusion.
My only gripe is the usage of "review" over something that indicates it's a pre-release product. However, given that Intel themselves didn't have any comments on the article, the final performance is likely to be so similar that this is basically the review of the final-release product.
They've already done that with a bunch of AMD products - like the OEM-only 4000 series APU and the weird Xbox One S APU desktop board.
Speaking as a tech enthusiast, if they get hold of Zen 4 before release and can do a preview that doesn't break NDA, I would be over the moon. I love to get an idea of how a release will shape up, as long as there are caveats that it may not be final performance - which is exactly what we got here.
I'd Lóve it! Beta BIOS might be a slight disadvantage, but we've seen it with the release of ZEN1. At the time everyone blamed the BIOS for memory compatibility, etc. etc. etc.
In the end, not much improvement was found, once stable BIOSes were out. Bugs were fixed, but ZEN1 was still not beating Intel.
It would still match the expected out of box performance. Stop being so salty over Intel sucking the big one, RKL is a total dud performance wise, a microcode tweak is not going to increase IPC. It already boosts to where it should be and draws LMFAO power.
terroradagio, your argument would have had more weight and be less hypocritical if you had not read the article that was posted. You lost the moral high ground when you did.
I will wait, because I will read the reviews that waited like everyone else and for the microcode update that very well may fix some of the issues. Anandtech has just given the middle finger to a bunch of other sites and channels who are doing the right thing and waiting to see what happens.
I didn't say that. I don't know what it will do. And neither do you. And that is why you wait to see before the official launch. Using a CPU that wasn't supposed to be sold is a total backdoor and deceptive way of handling this. Totally unprofessional.
Dude this isn't a new CPU core or a new GPU core. They've both already been released. There's no magic microcode fix. There is only an upcoming die shrink for desktop. That is all.
It is entirely possible a new update could fix issues. For example, in this review, the latency issues they were seeing. Look at AMD. AMD has put things out many times that have later been fixed with updates.
Those updates allowed for better turbo or memory frequency. Microcode updates won't fix cache latency issues if the physical SRAM is already slower. Same as it won't fix intercore latency as that deals with the mesh fabric.
microcode updates are extremely rarely anything to do with performance, they are usually to fix erattas. Since there was no instability in this review there is nothing for a microcode update to fix.
If you think a microcode update is going to give any kind of performance boost or power usage drop you going to be disappointed.
@terroradagio - "Using a CPU that wasn't supposed to be sold is a total backdoor and deceptive way of handling this. "
You're really not making sense. They bought a retail boxed CPU from a retailer; they didn't test an engineering sample. The same box would have ended up in someone's hands when purchased by them, and the retailer's stock of RKL CPUs will also be sold to customers. To say Anandtech used "a CPU that wasn't supposed to be sold" is a complete lie. A retailer sells products; they don't stock products just for themselves.
If you're going to argue something, at least argue something with more validity.
Not that I want to defend Intel, but to you and all others that have difficulties at understanding numbers (you just look at them and report the value without thinking about their meaning): The 290W power consumption was achieved by using AVX-512 instructions. In the test that uses them Intel is <b>6 times</b> faster at double the power consumption than AMD CPU. So under the point of perf/W with AVX-512 compatible workload (responsible for those high power consumption) there no doubts that Intel is the winner with a very large margin.
So better to concentrate on other power consumption terms to make this chip appear the fail it actually is with respect to the new architecture that does not really improve on almost anything.
Frankly I think some of us (at least, I) got confused by the revision of CPU architecture going into this design. Tiger Lake's Cove version might've changed the outcome, even if the timeline didn't work. Maybe would've been way too big though.
I'm not a CPU architectural engineer so I just can guess that this compromise of architecture and process is the best Intel could come up also taking into account production limitations. Probably it is not the best thing Intel could have done in absolute terms, but it for sure will be available at big quantities (unlike "super fast advance mega efficient hyper many core" AMD CPUs, and worse APUs) and that could be enough to fill the market until Alder lake is presented.
You know, this generation is just a fill gap as it is not a 10nm generation and can't get all those Cove's improvements on the now obsolete (but still high performing and delivering) 14nm process.
Zen 3 in the form of the directly competing R7 5800X is widely available at MSRP. How many times must you be told this before it gets into your thick ass skull??? O_o
Why are you guys making this an Intel vs AMD thing, when I am not? Get a life. This is about professionalism. Go take your AMD worshipping and Intel hating elsewhere.
I am an intel fanboy, but I dont agree with you at all either, intel scewed up and they have to deal with that, they wont fix anything the motherboards are made the cpus are retail and being sold. Anand did me a solid. I literally was about to pre buy all my stuff for rocket lake now I am not. If intel does pull some kind of magic they can change the reviews. No harm if something changes, but I don't think it will. I don't want AMD either so I am without a home right now also not really interested in big little cores either unless it pulls off some magic. This is also just a hardware review site not some major corporation not sure why you hold them to such a high standard.
big-little cores will help improve efficiency so the mobile parts may become competitive again but for desktop parts it will almost certainly decrease performance rather than increasing it. There is literally zero chance that the scheduler will always get it right which processes are assigned to which cores, and moving processes between different core types will always incur a performance hit.
Uhhh... why not? Deliberately not buying the best products available on the market because of the company that made them doesn't make ANY SENSE unless the company being referred to tends to be on the wrong side of basic business ethics/morality. You are cutting off your nose to spite your face here for no other reason than some misplaced stupid fanboy loyalty... -_-
Trying to make a moralistic critique out of your own personal objections is a desperate move.
If Intel want to do pre-orders before they've even announced the product, yet they're going to release stock to a retailer, then personally I'm only too happy for one of the best CPU testing websites out there to buy one and put the screws to it.
What a waste of time. 5xxx is the better Design. 3xxx series is when Intel should’ve noticed they were in trouble. Lucky for them I don’t need the best, just the most mature hassle free CPU. But I’m not buying anything until the server parts 480/1200 Xeons drop prices. If AMD can deliver on the 5xxx APU before that, I’m going to jump ship.
I'm not sure anybody saw the massive single thread or core performance jump from the 3 series to the 5 series. I remember being astonished at the geekbench and CB synthetics, thinking there was something bugged somewhere.
3xxx Zen was good, but 5xxx series is outstanding.
Zen 2 was always market as a a zen++ that for some reason had a massive enough performance uplift to be comparable to a new microarchitecture. Zen 3 on the other hand was marketed as a newly designed microarchitecture with the largest ipc improvements (from AMD) since zen. AMD delivered exactly what was promised, a significantly faster and more efficient product on 7nm.
The reason why you would have been astonished by the synthetics is because we haven't had a typical (15-30%) microarchitecture performance uplift since sandy bridge. Haswell was only ~11% faster than ivy bridge. The thing that we didnt see coming was zen 3 being called 5000 instead of 4000 and the price hikes.
Yes. And we also got spoiled in the process. Now these are the kind of jumps we want with every release because AMD showed us that they were possible. The thrill of these performance jumps AMD has brought to the table are nothing short of exciting to the entire semiconductor realm. Things Intel should have been doing.
i read else where that the " reason " intel didnt go above quad cores, or increase performance more then 10% gen on gen before Zen, was cause if they did, then amd would of been put out of business.
i would like to know is, when did/has intel ever done anything that was good for anyone other then intel ?
Well, if AMD were to go under, then Intel would be subject to increased scrutiny of being a near-monopoly. Thats probably why they would have done such a thing (if this was true).
Its similar to what Microsoft did way back in the day, propping up troubled Apple with cash infusions.
That sounds like a rationalisation more than a reason - they had the clear motive of protecting the revenue from their "HEDT" and server processor lines. It was telling how quickly they dropped their trousers on those products after Ryzen launched.
That would imply some incredibly hefty performance reserves to be unlocked when Intel felt threatened. Nah, all Intel seems to have done is to keep shrinking chip sizes at same 4C. This maybe was picked to help AMD, or just to help their bottom line. All clocks, IPC and whatever other improvements they had available were put in end products - otherwise these would be unlocked in Ice and Tiger lake, or this Rocket Lake abomination.
The biggest reason, and thisis one thing people dont want to admit, is there was simply no reason to. Their 6 core HDET parts were consistently slower then their 4 core peasant parts in consumer applications. So spending the money on making a new die just to have 6 cores on peasant class would have been a waste of money.
Remarkably underwhelming and disappointing. If the only distinction between 11900K and 10700K is left to binned core frequencies, you’re then left with the upgrades to PCIe 4 (which really is a designed delay from Comet Lake) and backported IPC’s at a tragic TDP. Yikes.
Ian can you guys start posting the actual wattage usage during each test beside the 125w rating? Would be nice seeing how much each cpu used during each specific test, if possible.
Adding in power detection during a test will often slow it down a bit, or require running the tests twice (they already take 24hr) and then parsing the data (probably more time than it's worth). Which would all lend to a slower review cycle. No easy way to automate it, unfortunately.
Maybe true but hand-picking a few benchmarks would be nice. The power draw section is opaque from our vantage because we don't know what performance looks like at max power draw.
A wall meter can be automated, easily. Even system level power consumption will be more useful that the TDP figures published by the manufacturers these days.
"...the only parts readily available on retail shelves right now are from Intel." Maybe in North America, but not in Europe. The only CPU missing here is the 5950X, the other models are in stock.
This.. is rough. As someone who has been team blue for a long time, and who knew this launch was going to be bad: this is bad. I'm sure things will get better with a bios update or two, but 14nm has just been stretched too thin. It's begging to be put down.
I mean, part of the problem is simple though: AVX is a waste of die space. An onboard GPU appears to be a waste of die space -- AMD sure doesn't need to sell one on its desktops. And a modular approach on desktop does wonders for yields.
This launch should've been 14nm, 10 cores, more spaced out in the die (if possible) and no iGPU on the die at all. Saves the Xe backport money and produces the same effect, since these real-world power numbers are unsustainable in anything other than a powerful workstation.
AMD is expected to add an iGPU chiplet starting with Zen 4. It might be added for machine learning acceleration in addition to entry-level graphics.
Intel will actually have RKL chips without iGPU, for what that's worth. i7-11700KF and i9-11900KF. I don't think it will improve the thermal performance though, just lower the price a few dollars.
Having just been unable to use my PC for 10 weeks whilst waiting for a Zotac 3080 RMA (pray nothing fails in this silicon shortage..) I’m all for every CPU having a basic GPU chiplet in the future...
I think you, as many others, over estimate the need for a dedicated GPU. Intel sells more GPUs (and that means they are all iGPUs up to now) than AMD and Nvidia together, meaning that most users are happy with the iGPU and don't want to spend more for another component. Especially in this period where a 4 year old GPU costs twice what it costed at its launch.
Not putting a iGPU for how small it can be would be a failure, not the opposite. AMD is struggling in trying to put an iGPU into its desktop CPUs and will have to resort to a dedicated chiplet for that (with performances and power consumptions stil to be seen) and Intel would do the wrong step in going backward directions to the "full on package integration (and in die if possible)". They do not sell discrete GPUs (not yet though, and not decent ones for a while when they will) and not providing a integrated GPU will just mean giving the chance to its competitors to make more money when they could easily avoid that and still use more power.
Because in all those benchmarks none has evidenced that AMD's set up needs a discrete GPUs (if you want to see something on your monitor) that needs much more power than Intel integrated one, so you should take into account the sum of AMD CPU+dGPU vs Intel CPU only to see what are the real differences between the two.
and you over estimate the need for an iGPU on a high end cpu, that most will probably DISABLE, and use a discrete one. " Intel sells more GPUs (and that means they are all iGPUs up to now) than AMD and Nvidia together, " of course they do, cause they sell one with EVERY desktop CPU they sell. "Because in all those benchmarks none has evidenced that AMD's set up needs a discrete GPUs (if you want to see something on your monitor) that needs much more power than Intel integrated one, so you should take into account the sum of AMD CPU+dGPU vs Intel CPU only to see what are the real differences between the two. " um no, cause as i said above, most will probably disable the igpu on the higher end cpus, to use a discrete one. would YOU buy this cpu, and use the iGPU to play ANY game that was released in the last say 2-3 years ? lets see you play CP2077 or and of the games that were tested in this article.
True, Intel is the biggest gpu seller in the earth because of integrated gpu in cpus... And Also true that most people use cpus in the Office Computers, so They don`t need more speed that intregrated gpu offer. Gamers Are really small minority in the world. But for gamers neither Intel nor amd makes GPUs fast enough. Maybe in the few years integrated GPUs will become gamers choise, but not yet. But all in all gamers don`t count in the big picture.
If you are the type of person to need an igpu, you don't need more than 4 cores, or you need a massively parallel processor (many cores for compiling, simulations, etc). AMD not having an igpu is not a fail because if I need that much cpu, I likely need a fast gpu. Intel's 96 EU Xe does not make the cut, let alone these 32 EU parts.
If Intel gave an igpu fast enough for at least 1080p medium, your argument may have some merit because I can still game in spite of not being able to find a dgpu. Most non-gaming or compute-heavy work tasks people can already do with their existing computers/phones. Meaning the inclusion of an igpu is meaningless because I can already web browse on another device.
No BIOS update is going to be able to "fix" the physically much larger cores & caches creating higher core-for-core AND core-to-memory latency. That's just an inherent part of the new design.
Well Gaming is not important to most people... but power usage does matter. So not bad product, but shows that Intel needs the next step Sooner than later.
I only posted "gaming" because especially back when the 9650 was in its heyday, single thread/core/IPC performance was by far the most important thing for gaming. Other apps benefit from such things as well---especially back then when coders were still learning to properly code for more than 2 cores.
> However, due to high demand and prioritizing commercial and enterprise contracts, the only parts readily available on retail shelves right now are from Intel.
Wonder if this will continue to be the case. At least in the US, there seems to be plenty of retail Ryzen 5000 availability for about 3 weeks now.
What planet are you living on? The only chips reliably in stock are the 5600x and sometimes the 5800x, and $100+ over MSRP. The ryzen 9s are nowhere to be found.
The 5800X is in stock right now on AMD's webstore, for anyone reading this at date of publication.
I agree with them that the situation is improving, but also agree with you in that the 5900X is very tough to come by, with the 5950X effectively ceasing to exist.
As usual, you have to watch Microcenter stocks daily and hope you get to the store in time to pick one up (they do not allow you to reserve it online).
Newegg has been hopeless for me with the new gpus and cpus.
As someone 3,500+ miles from the nearest Microcenter, I press my face up against the glass longingly every time someone mentions their in-store pricing and availability.
I briefly attempted to purchase a 5950X from Best Buy, but their Zen 3 CPUs were listed both as "Online only" as well as "Cannot be shipped." I then tried ship to store, and was met with "Unavailable for ship to store within 250 miles."
I feel for you, man. I'm blessed in the sense that there are like 5 microcenters within a 100 mile radius of me. Its still basically luck though---I would have no chance of getting there in time to get a 6900XT or 5900X/5950X unless they had double digit stock, because I wouldn't be able to get there until work is over, and the stock dives during the day.
As somebody noted before, they still have nice stock of the 5800X.
I bought a 5800X (for myself) and 5600X (for my dad) at MSRP from my local Microcenter, with literal piles more remaining on the shelves (and more in the "back"). It's only Ryzen 9 that's hard to find.
I live in GA, and both Microcenters are well stocked of 5600X/5800X at MSRP. Not only that, but Walmart, Newegg, Amazon, and others have had 5600X/5800X stock come up daily - I kept the notifications on despite getting what I needed. Ryzen 9 has had a lot less stock - that I agree with.
"When AVX-512 comes to play, every-one else goes home" pretty much nails it. If your code runs AVX-512 and is limited by single core speed (plus max bandwidth to a RTX3090) then no other processor comes close.
AVX512 in client class hardware could perhaps get more sales for these parts at the expense of lower sales of higher margin hardware that would normally be required. An extremely poor financial decision to include AVX512 in client class hardware, especially considering it very well may be at the expense of two regular CPU dies which would have narrowed the performance gap to it's competition.
Only if it's 100% AVX-512 though. This CPU still loses to AMD in tests where there is a sprinkling of AVX-512, which is how things will always look in client workloads. Most people running simulations will be running them on workstations that support ECC RAM.
From my understanding, low end xeons take the consumer dies and enable ECC. So the dies could have meaning if intel is trying to increase the volume of avx-512 computers in the market. If the goal was for the die to be used to retake the gaming crown or remain competitive with AMD, it was an absolute fail. The only reason it was included was probably because sunny cove was designed around it's inclusion, and removing it from cypress cove would have delayed the launch to the point that Intel may as well just wait for Alder Lake.
"This made the advantage of AVX-512 suitably only for strong high-performance server code. But now Intel has enabled AVX-512 across its product line, from notebook to enterprise, with the running AI code faster, and enabling a new use cases."
As usual multiple spelling mistakes...
Is Anandtech ever going to proof read their articles before releasing them?
Those aren't exactly spelling mistakes. Those typos amount to grammar mistakes, but using correctly-spelled words. That makes them a lot harder for an automated program to accurately catch. With humans proofreading, it's very hard to catch your own typos, and each new person who proofreads might catch 80-90% of the remaining mistakes. The only ways around that are to have several more people do proofreading passes, or to set the article aside for a few days before going over it again with less familiarity.
It is not just the grammar. It is inept and shoddy to be using a cooler from 2008, which has not been in production for years and refer to it as 'the best air cooler on the market'.
Actually, it's a down right untruth. The Thermalright True Cooper was not considered the best cooler in 2008 (see reviews on Overclockers.com) so it most definitely not the best one in 2021.
Why not use a Noctua D15? Or even at least use exactly the same Noctua on both platforms.
This is methodology that Anandtech should be ashamed to publish.
Did you mean to reply to a different comment? Because it looks like you're not even trying to say anything that relates to my comment or the one above it.
Because it IS the best air cooler on the market. The 100% cooper cooler they used absolutely CRUSHES an NH-D15. If they can't do it with the pure copper, a Noctua had no chance in HELL!!!
Importance of this depends what exactly you mean by "on the market". Cooler WAS on the market - ie it isn't something custom made. If even that cooler was insufficient, anything else currently on the market would be even worse (assuming it is actually the best air cooler for this, something I have no idea about)
Right, should test the CPU with the same components. So as Intel does not need a dGPU, it should not be used on AMD as well. And use the AMD platform without any video out, as it is sold. Sound fair, doesn't it?
Thank you for pointing out that grammatical error. I've gone ahead and corrected it.
"Is Anandtech ever going to proof read their articles before releasing them?"
This article actually did get a full proof-reading pass by me. But even then, I missed that error.
But to answer the question at hand, for the typical article the answer is "no, it will not get proof read". We let our copy editor go almost 15 years ago during the Great Recession. And while I'd still like to have one, it simply doesn't make economic sense. The modern, Google-driven news cycle does not reward copy editing for online news. It's better to post more articles, and to post them sooner.
Thanks, Ryan. You know, these people who keep on pointing out tiiiiinnnyyy errors are getting tedious. I sigh every time I see another one of these comments castigating grammar or spelling. It's better to have excellent material, like Anandtech has, with the occasional typographical slip, than lacklustre material + "perfection" (like some other sites).
Been planning a rocket lake build for over a year. Already bought everything else including motherboard, was only waiting on CPU. Thanks for ruining my day. What the hell am I supposed to do now?
Devil's advocate: It's not a home run, certainly not anything revolutionary, but it's still a fine chip.
If you were already interested in the i9-10900K, this is essentially the same thing with a few feature improvements. It's not a great value proposition (assuming it lands at ~$450), but you still have:
· Improvements to single core performance due to IPC uplift · AVX 512 (if you should need it for anything) · Gen4 support for the primary PCI Express x16 slot (or x8/x8 for some boards) · Gen4 support for the dedicated NVMe M.2 slot · 8GB/s (vs 4GB/s) DMI link bandwidth, for storage or peripherals connected through the chipset · 32CU Xe iGPU, for ~33% improvement vs 10900K's Gen11 IGP. Full features, including AV1 decode · 20Gbit USB 3 · TB3/4 support (depending on board model)
If you're wanting an Intel system, it's still arguably a solid option. It offers useful features that the 10th gen CPUs lack, and Gen4 alone may be worth it if you intend to use fast storage ("RTX I/O" or DirectStorage down the line) or want to be able to upgrade to a faster GPU in the future.
Is it a good value? Not particularly, but the competing 5800X isn't a bargain at $450, either.
You won't find yourself shouting from the rooftops, but it'll do the job and let you skip the first iteration of Intel and AMD's DDR5 platforms. By then, something more interesting will definitely have come along, and the early adopter tax should have eased up.
What you're supposed to do now is base future purchasing decisions off benchmarks and not hype, empty promises and marketing, especially when it comes to a company who has given themselves a reputation for under delivering over the last half decade or so.
A technical annoyance maybe so eBay the motherboard quickly to some sweaty palmed Intel fanboy who doesnt read Anand and reuse everything else in a more modern Ryzen build?
Otherwise maybe email Intel to see if theres any discount coupon available for 2 kilo coolers? Or maybe go head with that 4 fan twin 240mm water build you always wanted?
Be happy that you can get a CPU early (If Mindfactory has some in stock again) and enjoy! Including the various problems of being one of the first on the platform. Take it as an adventure!
Would it be possible going forward to list in the review on whether the current CPU supports both AV1 encoding & decoding? (To my understanding, RKL only support decoding).
Correct, AV1 decoding only. There was an early Intel slide that made the rounds which listed support for AV1 encode, but it was quickly acknowledged as a typo by Intel. Decode only.
My guess is that you're looking for faster AV1 encoding. It might take a while before it reaches hardware, and even then, quality and size will likely suffer. The best option at present is to use libaom, through FFmpeg for example, and set the "cpu-used" parameter to 8 and see how that goes. If fast, keep on dropping by one. Intel's SVT-AV1 is another option: fast but lacking in quality. Also, keep an eye on VVC and x266: it's in the same tier as AV1 but, from my testing with the Fraunhofer encoder, slightly better, sharper, and faster.
OK, so once more into the gap with 14nm... looking about as good as could be expected for a node this old. Then there are all the AMD faobois.. To whom I will ask, where is my 5970x? Hmmm??? And to either camp, why can't I have 32 or more (40 seems about right...) PCIe 4.0 lanes on a desktop CPU? From my perspective, neither camp is executing particularly well. Intel at least has Gelsinger back, we'll see what he can do in about a year (or two). AMD, Dunno... If they can't execute at 7nm, what makes anyone think 5nm will be easier? Then there is poor TSMC, having to worry about China's takeover. And mark my words, China WILL take over.
Both Intel and AMD failed utterly to avoid turning "mainstream" computers into toys when they moved PCIe from the chipset to the processor.
Intel moved their non-toy processors into the HEDT space, before it even had a name (I had to buy Sandy Bridge-E rather than Ivy Bridge to get a usable computer back in 2012), at ever increasing prices.
AMD at least brought that buy-in price down a little bit with Threadripper, then turned it way back up with the Zen 2 incarnation. I'm hoping the Zen 3 TR chips start at 12 or 16 cores, so an upgrade is reasonable. I don't need 24 cores, but I do need more lanes.
I put 40 lanes as the bare minimum for a non-toy PC.
24 is fine for the vast majority of people, 28 would certainly be nice. I don't tend to take comments seriously when they refer to computers with 16 cores and the ability to run as much IO as 90% of users need as "toys", though.
It's a good option, but the $1,000 motherboard does put a bit of a damper on things. I was also hoping for a 16C/32T Threadripper 3000 part, but when it became obvious that they didn't intend to offer one I picked up a 3950X instead. While it has in no way disappointed me, the extra lanes and memory bandwidth of the Threadripper platform would have been appreciated. A $1,000 16C version would have been an absolute no-brainer vs the 10980XE and still leave a market space for the $750 3950X.
The lack of 8, 12 or 16 core Threadrippers at the end of 2019 is essentially the only reason X299 still exists; it baffles me why they didn't pursue it at the time.
Are you sure you need 40 PCIe 4.0 lanes though. I'm willing to be you probably just need more PCIe lanes or PCIe 2.0 lanes at most. For that you can get bifurcation cards that splits or doubles PCIe lanes. AmeriRack is an example of a company that makes these.
The answer to "why can't I have 32 or more PCIe 4.0 lanes" is pretty obvious. It's the socket size and pin count. Duh.
So for AMD on the AM4 socket there is absolutely no way to increase the PCIe lanes without a new socket. Which is what the sTRX and sWRX sockets are for. 4,094 pins for the latest Threadripper Pro which lets them have 8 memory channels and 128 PCIe lanes. As compared to AM4's 1,331 pins.
AM4 on the x570 chipset is still an improvement on PCIe lanes because the PCIe4 speed allows doubling the chipset bandwidth. So it can drive twice the number of PCIe 3.0 devices as before.
I predicted ~300W peak power about a year or so ago when I had first heard they were making the big mistake of bringing AVX-512 to 14nm mainstream consumer processors. Why on earth? There was a valid, wise, very, very, very good reason Intel had reserved AVX-512 to just their 14-nm HEDT processors and that had always been the furnace-like heat and nuclear-like power consumption of it. Even with the best logical design improvements from a new microarchitecture, it is still an extremely intensive logical pill of a task to swallow. Now, we see that reason in full, unadulterated display. You bring a massively complex instruction set extension to a higher process node where you have far lengthier physical networks (meaning essentially longer wires, increased resistance, higher power, and maximum heat) and, of course, you are going to have a steaming pile. Remember this review is only looking at the number two product, the Core i7-11700K which has lower clocks and lower power draw. The Core i9-11900K will likely need a 360- or 420-mm AIO just to not thermal throttle like mad. My 5950X with its meager 240mm AIO (Corsair H100i RGB Platinum) that runs at the quiet mode setting is laughing its butt off right about now. When Ian Cutress had to use an obnoxiously loud 170 CFM fan (I have used 100 CFM Deltas and those already annoy most PC enthusiasts) on a massive 4-pound, full copper heatsink to tame the 11700K's 290W, I shudder to think. Will the 11900K be outdoing the FX-9590's record-making peak power draw of 350W? Ian easily could have gotten the 11900K also at retail, but I think he is holding back on that because he already knows the 11900K is going to be a throttling disaster and only the 11700K is an actually usable processor. *mike drop*
Wow... 5800X looks fantastic here compared to the competition.
I'm happy with my 3600, but it's really nice to know that the 5000 series is available as a future upgrade.
Intel will come back eventually... because $$$$$$$... but I'm honestly surprised that AMD's chips have so thoroughly beaten Intel's on in basically all areas for two generations. This hasn't happened since the Athlon 64 vs Pentium 4\D days. Whether Intel's next big performance jump will be a complete game changer like Core2 though... time will tell.
Typical applications these days are not optimized for AVX512 or even AVX. That is the main problem for Intel. As soon as you do benchmarks using software compiled with Intel C++ and their libraries, AMD is not even on the map. Despite huge possible gains (10x), most big companies never invested.
So, what you are suggesting is that because companies are not being 100 percent focused on Intel performance, Intel loses in the majority of situations now?
Most companies are obviously not keen on investing in performance. When first compilers came out in 1960s, they brushed away with assembler, which produced 3x faster code, because it was cheaper to write software with them. We have a similar situation now, but in reverse. Intel adds new instructions, which require "more development cost" in order to achieve higher speed. AMD instead focused on speeding up existing code, which does not require more development cost. On Anandtech you have one (!) benchmark which is hand optimized (3DPM) for AVX512, but there is ZERO reports, on out of the box performance of Intel C++ (never mind optimizing C++ for SIMD). The whole situation is further skewed, by the so called power inefficiency of AVX512, but completely forgetting, that when used correctly, these instructions result in 4x faster code per core than AMD AVX2. Even Intel AVX2 is 2x faster than AMD AVX2, when done properly. Yes, Intel can draw 300W, which is horrible, but it can run 4x faster than best AMD with that power draw. The final power/performance ratio for what you could call "quality software" is in fact still two to one in favor of Intel. But when running GCC compiled code, written by just "anybody", this is not the case and Intel struggles. If you buy Intel, you also need a qualified development team and no prejudice for using Intel compilers and libraries to make TCO stand.
Here are the number of cycles for 2000x2000 matrix multiplication with 1 thread on libopenblas-r0.3.5, which is done properly.
1158M Skylake (6600K) 2198M Zen (1600X) 1451M Zen2 (3700X)
So, Skylake is indeed almost twice as fast compared to Zen, but not compared to Zen2 (assuming similar clock rates). It will be interesting to see Zen3 and Sunny/Cypress Cove.
Matrix multiply is already maxed out with AVX2 and Intel MKL shows no gains with AVX512 capable CPU. Using one thread still requires blocked matrix multiply to have data in cache. On Skylake X-series 4GHz, with 8 cores, I get 65ms for multiplying of two matrices, double precision 2000x2000 in size. With a single thread, it needs 290ms.
4GHz*290ms=1160M cycles, same as the consumer Skylake above. I would have expected AVX-512 to gain a lot for matrix multiplication: wider units, and more registers for register blocking.
Matrix Multiply depends on cache L1 to CPU core bandwidth, which is about 1.9TBytes/s and this is relatively fixed in the last years/generations. To get more speed with AVX2 and especially AVX512 you need an algorithm, which has more than one or two CPU instructions (multiply + add for dot product) per one memory read/write. This is a similar situation as with GPUs, which require algorithms with 50 cycles of instructions or so per one memory read to really show their muscles. During those 50 cycles, the GPU thread can work with registers or wait for memory to respond.
For 2000x2000 DGEMM, libopenblas on AVX performs 1G loads, 0.285G stores, and 2G AVX FMAs. E.g., a Skylake can perform 2 AVX256 loads and 2 AVX FMAs per cycle, so this is bounded by the FMAs. The Sunny Cove (and, I expect, it's 14nm offspring Cypress Cove) can do 2 twice-as-wide loads and 2 stores, as well as 3 twice-as-wide FMAs; and AVX512 has more registers, allowing to reduce the number of loads per FMA. So I expect a speedup by a factor of two from the wider units, and some additional speedup from the additional resources available.
It does not happen. Intel started introducing AVX512 some five years ago in to their toolset and for large matrix multiply the presence of AVX512 with 2FMAs has zero effect. There is difference however for other algorithms, which in fact are 2x faster. For example vectorized math functions: Sin, Cos, etc.. where the memory bandwidth requirement is much less.
Intel wins bigly... mainly because ICC nerfs the shit out of performance on non Intel PCs? Although both are large corporations, at least AMD is much less scummy.
So this is even worse than 10nm sunny cove because it has worse L3 latency which effect gaming performance
Intel should have just released 8 cores tiger lake for desktop. Even the mobile version is beating rocket lake in single-thread. 125w tiger lake would easily destroy that thing
I don't know if they could put 125W through Tiger Lake and have it survive - the heat at the most dense parts would be intense. I think that's a big part of the problem. The other part is that they are *still* struggling with either yield or fab capacity on 10SF (possibly both), which is why TL-H45 still isn't here. If they had to split that across desktop and notebook lines then they'd have a similar supply problem to AMD, and for Intel that would be a much larger issue.
I suspect AMD CPUs are becoming available because no one can build a computer due to the lack of video cards.
The one benefit this CPU has over the AMD counterparts is the iGPU, which at least allows a functioning system. I would love to see how this new iGPU performs.
If I want a functioning system I can use my existing pc, laptop, or phone. The only reason I would upgrade is for a better gaming experience. My i5 2400 just isnt cutting it in games any more, and a 5600x is really enticing. Just waiting for a 5600 or for the x to drop below 300. If I were to be building a PC from scratch, I still can't game with decent settings on intel, so it just isn't that big of an upgrade.
To me your argument only makes sense if a person doesnt have any computer or phone, is knowledgeable about PC hardware, physically walks into a store to buy computer parts, and a dgpu is unavailable. It is possible if say someone's house burned down, but it's very unlikely. Especially because buying a phone is probably more important than a computer, meaning the person can already have a basic functioning system.
Well, if you're waiting for the 5600X to come in at $299, then it's there for you. $299 is the list price, and there are retailers who stick to the list price when it comes back in stock. I got a 5800X from Microcenter for list, and they had the 5600X for list as well.
Not very well. It's only a third of the Tiger Lake iGPU, which itself performs serviceably well in optimised games when it has enough TDP headroom and LPDDR4X. This will be unspectacular, albeit still better than the junk bolted onto Sky/Kaby/Coffee/Comet Lake.
There are 2 problems for 8 core tiger lake to be released to desktop. One is a lack of available volume because Intel's one advantage over AMD is availability (I assume that Alder Lake is getting much of the volume of 10nm especially with its new iteration ESF). Two, tiger lake probably struggles to scale performance with power beyond that point. If Intel's 4 core tiger lake can have a tdp of 35w and scale up to 65w, then intels 8 core tiger lake can probably have a tdp of 65w and scale to 125w. TDP no longer implies the maximum power draw of a processor, and rocket lake pulling 250-300 watts can probably clock higher. Since IPC is the same, rocket lake wins on performance. Intel obviously no longer cares about efficiency because they know that they can't win on that metric.
Ian, are these two statements for the same thing or intentionally worded different: "Intel is promoting that the new Cypress Cove core offers ‘up to a +19%’ instruction per clock (IPC) generational improvement," which you mention on page 1; and "Intel is promoting a +19% average performance gain," which you mention on the last page?
If so, those statements don't mean the same thing. One statement refers to a best case scenario and the other refers to an average. This puts them at odds with each unless Intel made two separate claims both involving 19% with one "up to" and and the other "average." If so, that is pretty slimy, misleading business practice on their part because that would mean the latter has failed to live up to their advertised promises.
Fantastic work. An absolute coup for your page, and especially important in an era where manufacturers are trying to secure presales on products that haven't even been tested yet; especially when Intel seem to be so keen on being misleading (19% IPC) or even straight-up lying (better performance in games!) about their upcoming products.
On a personal note, I *finally* the data I was looking for to put a damper on the fanboy agitation that had been hailing the Return of the Gaming King. Perhaps the 11900K will do better at whatever silly price it launches at.
CPU aside, the integrated GPU should be reviewed too.. it's hard to buy reasonably priced GPU these days.. might be the new Xe- GPU can serve gamer at low setting better than the predecessor.
Better than the predecessor? Sure! Well enough to be useful for a gamer? No, not at all. You could buy a second or third-hand GTX 750Ti on eBay and get better performance.
How can you be sure if there's no test? That's why those numbers are needed for a conclusion to be made. If Tiger Lake can do reasonable gaming why can't this do the same?
Precisely this. You'd also need to budget for a board and RAM that will support running DDR4 above 4266Mhz to get those 32EUs running at similar performance levels to the ones in the mobile SKUs.
TL;DR: even if clocks are higher on desktop, Xe on Rocket Lake can only have (at best) ~50% the performance of Xe on Tiger Lake - so, slower than the Vega 8 on the 4750G.
Intel had best hope that some last minute chipset /driver/Win10 updates perhaps change these results; gaming FPS being less compared to 9900K (from 2+ years ago!) and 10900K ago is downright pathetic....
I feel like this review might have just killed interest in 11th gen Intel desktop parts. Maybe there was a reason the retailers were trying to sell these before the official date.
Definitely nice to have a review out long before an actual release to give people time to think things over before buying.... Well except for those who have already bought it in the EU.
5800X and 5600X CPUs are reasonable to find these days, it's only 5900X/5950X that are extremely difficult. There's no reason at all to buy this stinker Intel CPU.
I think 'they' (AMD) is running through their AMD Ryzen 9 3900X/3950X Matisse inventory --- currently $449/$679 at my local MC (versus $549/$799). There is still incredible value in the Zen2 chips ...
The improvements seem less impressive when you realize this is the top dog of the product line, being compared to the second-best of Intel's last gen and third-best of AMD's. Granted, we don't know the price so the value may still be a good improvement over what we had. There is just some perspective missing here as 8 cores is only mid-range for the competition.
Still, AVX-512 can be a game-chnager for some(if they have the cooling to match).
Ian used a 4 pound full copper beast level heatsink and blaring 170 CFM fan. I have used 100 CFM Deltas, the kind you find in servers, and those sound like jet engines taking off at the runway (hence why they sometimes jokingly called Delta airlines) and they annoy the heck out of PC enthusiasts. Good luck with that.
If there isn't room on the CPU itself at 14nm, the answer would appear to be shifting whatever processes/functions that could be moved to the motherboard side of the CPU interface where there is more room. For example if the motherboard has the PCIe 4 connectors for discrete graphics card why can't it also include a chip for iGPU rather than having it on CPU.
I wonder how they sell F type CPUs then (iGPU disabled)...
They could easily fit two more cores into 32EU iGPU space, but they they'd sell fewer Xeons.
The fact that AMD sold cores on die area Intel is giving away for free (they basically charge $5 for an iGPU) has been the main appeal of Zen (except APUs) from the start.
"die area Intel is giving away for free (they basically charge $5 for an iGPU)"
Trying to fathom the cost of Intel's iGPU by seeing what they charge for a SKU with it disabled (but still physically there in silicon) is a fool's errand.
Intel don't give anything away for free - their margins didn't go down after they put their iGPU on-die.
Whatever the advantages of the new microarchitecture, Intel managed to negate them all, by trying to sell overclocked products. Once upon a time when you overclocked a chip Intel would void your warranty. Sometimes it would directly ban overclocking or distinguish its products between these that can be overclocked and these that can't be.
I would buy a true 125W TDP chip, with a new microarchitecture, even knowing it is slower, with a reasonable price of course. I would not buy overclocked chips at any price. If Intel wants to command prices comparable with AMD, they need to offer chips with comparable performance per watt power.
It would be less embarrassing for them to sell a 70% slower product with the right power than an equal regarding speed chip with twice the power draw. In my eyes of course. There are people who only care about speed , no matter the power or the money.
You can have a fast porsche, or a fast ferrari, or a fast koenigsegg. And you can have a fast dragster. Not the same thing.
Define overclocked. It really just means clocking higher than the manufacturers specifications. If Intel is defining the clock speed it's not overclocking.
While AMD CPUs may be difficult to find at suggested retail prices, it's FAR better to buy the AMD CPU than a POS back ported overheating Intel CPU if you can find one.
For the record I have documented that several sources including Newegg have been selling the Ryzen 5 5600x @ $299. as soon as they receive stock which appears to be weekly or more frequently. Newegg advertises the $384 CPU price and (5) offers. When you click on the item you see the $299. Newegg shipped item when they are actually in stock for a short period of time. THAT is when people should buy and not support scalpers with inflated prices. It's worth checking Newegg frequently as they are not advertising the $299. price much with so many scalpers on their network.
In addition for anyone near a Micro Center they are only selling in-store but they have had a continuous supply of AMD 5000 CPUs for some time and at the correct price. This should be the first choice for anyone located near a Micro Center store because there are no games or price gouging involved.
So what it the point of Rocket Lake? The fact that it is only available with a maximum of 8 cores makes it pretty undesirable for most workstation tasks, compared to competition from AMD and even themselves. Also the fact that they seemingly sacrificed their architectural advantage for games to a point where it gets outperformed comet lake makes it seem like a chip that is completely dead on arrival with no viable market whatsoever. As it is it does everything worse than both zen 3 and comet lake while running a lot hotter and consuming more power. I mean great that intel managed to backport it to 14nm, but what the hell was the point.
They have to do an annual refresh or release to appeal to their investors. That's literally it. Honestly, Skylake+++++++++++++++ would have been a better choice because at least it would have better gaming performance than this hot mess. Now, you have idiots like Usman Pirzada at fake news WCCTech playing arm-chair apologist, saying that the peak power here is a lie since it is AVX-512 and other garbage. Well, newsflash: AVX-512 is one of a few reasons why Intel got improvements in IPC, so if you want an AVX2 to AVX2 load test, also test the 11700K with AVX-512 disabled. Intel loses some of their minor gains in synthetic benchmarks, where they are still losing to the Ryzen 5000.
Leaving aside the 14 nm handicap, I reckon that Intel's engineers took Sunny Cove a bit too far, widening the out-of-order window excessively, adding more decoders, etc. One will notice that Zen 3, while raising performance considerably, was pretty conservative in its design.
The reason here is the 14nm handicap. You cannot bring long lengths of transistor networks without having to counteract latency, noise, and a host of other problems that come from stretching a microarchitecture intended for shorter physical paths. As an extreme example, that is why Zen 3 will never be backported to a 1 micron process. ;)
Would be interesting to see Sunny Cove fabricated on TSMC's 7 nm. I always wanted to see such a comparison, with Zen 3, taking the process out of the picture.
Intel apologists out in full force. What a joke. I can't believe I invested in the stock of Intel recently thinking they would turn it around. Can't wait for the GPU dept to underwhelm next.
Tech journalism at its peak. It is in such a way that even a tech illiterate can easily understand what is being explained. Thanks for the effort Ian and Anandtech.
I previously had categorized RL as mostly optimised for gaming. But now even with the peak power draw when using AVX 512 instructions leaves me without an immediate use recommendation.
Intel has been forced to release a Bulldozer. All AMD and TSMC need to do now is steadily increase supply in 2021. If Alder Lake is anything less than stupendous, Intel is (insert negative term).
I don't know if you realize this Ian, but your article may have contributed to stabilizing AMD stock yesterday. There had been so much "Intel is returning to dominance!" gobletygook in the world investment community, and this article will do a lot to bring everyone back to earth and the reality of AMD technological and product superiority (for at least 2021).
Absolutely not, there is absolutely no connection with the stock price of AMD and this review. The stock stabilized because AMD stock value is undervalued at 75$. The real stock price should be closer to 105$, but the selloff of semiconductors companies last week was general. It happens because some people think the economy is going to restart the good old ways with vaccination... they are all wrong.
lol I said it was bullshit that userbenchmark had this listed at #1 for processors and said it should almost be considered illegal and what they are doing is manipulating public opinion and posted this review stating check for yourself.
Of course I was promptly blacklisted and they banned my IP Range for speaking the truth.
* should note I said all this in the comment section for the processor itself, the comment of course is deleted and barely was there for 5 minutes before they blacklisted my IP range and removed it *
I stopped reading user benchmark ever since last year (the only year I had started reading on the site) because it became very apparent that they were biased, it is easy to see in their product descriptions.
Most of their content would be worthless even if their tests weren't biased as hell - there's no way to tell how people's systems are configured, what background software they're running, etc.
"Additional improvements over Comet Lake include AVX512 units, support for 20 PCIe 4.0 lanes,"
Along with the avx512, Rocket Lake is also including the dlboost int8 operations ... so quantized ai inference performance can be pushed to around 4x vs Comet Lake, minus any avx512 related clock limiting.
Is it the CPU or is it the motherboard and/or early BIOS that’s causing the performance regression? Perhaps consider performing tests on different z490 and z590 boards to see if it is the chip that has regressed or if some motherboards have better performance.
Also, many people use AIOs so what is the temperature differential when using a 280 or 360 rad?
Not fair to call this a “review” especially since it’s not been announced yet, and the bioses are still in beta. For example, on my gigabyte vision d z490, the beta f20 bios with rocket lake support is a flaming mess. With boot loop issues and memory training issues. Completely unusable with Comet Lake. Sleep issues as well (computer wakes up immediately after sleep). The earlier non-rocket lake BIOS versions perform much better.
So I will wait to reserve judgement on performance until after these early beta BIOS bugs are worked out.
I'm actually more interested in the H45 notebook parts with embedded TB4 on 10nm. I use a notebook with an eGPU. The higher core counts and direct-to-cpu TB will be a big deal for me. As right now I'm relegated to 4c 28w or marginally H35 right now to get an efficient TB implementation for eGPU gaming. (No I don't want the Asus Flow X13 and it's 52dbA XG mobile)
What a load of crap. Hot, late, slow. The biggest joke of all is this site actually recommends you buy one “because you can get one”. I didn’t think this place could sink any lower. I was wrong.
One thing I think that is of interest to people is what security holes have been filled with these newer Intel CPUs, if any, and were these benchmarks run with OS and firmware security mitigations in place? My guess would be: no. It's not really an issue for AMD as that story is well known--1 or possibly 2 mitigations in place--both were handled in firmware for Zen2, probably neither is a problem for Zen 3. Intel, otoh, has been cursed with dozens of security holes patched with Windows microcode patching & firmware patching for quite a while. An update on the situation would have been thoughtful. I don't think Intel believes it is ready to compete just yet.
Thanks for the excellent work, Ian. While it hasn't displaced Ryzen, I commend Intel for the effort that Rocket Lake is. Zen 3's architecture is a lot lighter, which suggests that Intel may have to curtail theirs in the next round. Starting from scratch might be of use but will take a few iterations before catching up with Zen.
Well I finally got my 5950x a couple days ago but I did order it December 15th on special order so it took awhile I think I payed $850ish for it with tax. I was thinking 11th gen intel was gonna be better than this hell seems like 10th gen the smarter bet and going with a 10900k or one of the close versions if you wanted intel. But overall amd is killing intel in every section of the market. I couldn’t imagine this about 5 years ago especially in the laptop market where 7nm efficiency really helps. Now if they can get it together with their gpu’s, the 6000 series is great and all but if you gave most people the choice i still think they would pick a 3080 over a 6800xt. Ray tracing performance is nice but it’s the dlss really helping nvidia. I have a strix 2080ti with a 9900ks right now as my main gaming rig and a 3900x radeon vii for my work/gaming rig. I’m gonna put the ti with the 5950x but I wanted to find a 3080/3090 for the build but I have gave up for now. Hell I need to sell a bunch of my gpu’s while the market is crazy. I have a strix 2080, nitro 5700xt se, msi gaming x 1080, and a couple more thats just collecting dust.
Interesting results, but I wasn't surprised by the heat and power consumption to be honest. I was expecting Intel to reclaim single core performance, but at least I don't see it in this independent review and also a credible one. I think I kind of understand why Intel had to rush Alder Lake out on the same year now. While Intel's 14nm has bought them a long time, it is really on its last leg, particularly when AMD and ARM are utilizing more advanced nodes.
Wow, if this thing is *that* hot (as was expected) then the 11900K will probably double as an egg fryer, no matter how tightly it was binned. Those 292 W and 104 °C AVX-512 peaks were scary, really. I wonder how Zen 4 will fare with AVX-512 code. Will TSMC's 5nm process node help it keep power and thermals at reasonable levels?
Under the absolute worst scenario, you might see AVX-512 on Zen 4 to hit the power level of AVX-2 on Intel. And that’s worst case, mind. Best case, the same power level as Zen 3 with AVX-2. Realistically, somewhere between those two extremes on that spectrum, leaning more towards Zen 3’s maximum power load.
Two things: Latency increase is most likely down to size of CPU. Second, why are you using average FPS in games???? It is the worst part of reviews here, because it completely obscures any useful info.
Remember "Inside a second" by Techreport. And with modern CPUs and high variability of clocks, that methodology is bare minimum for even basic comparison between CPUs.
"Second, why are you using average FPS in games?" Earth shattering concept reveal: faster processors often produce more min/average/max fps in games...; by seeing the 11700K producing lower results that the last two generations, folks are...disappointed, to say the least.
You have 95 percentile results here, presented as FPS. This is standard fare these days: You get a number X ms where 95% of frames have lower or equal time vs previous frame than that. Then you convert those X ms frame time numbers to the more typical FPS (=1000/X).
Yeah, there are no percentile graphs that illustrate some issues with CPUs or GPUs, but outside of extremely few cases those were mostly boring linear increase in frame time to ~99% then a sizeable spike for the last ~1% and that's it. Dragging a line between average (50%) and 95% and extrapolating to 99% would give you essentially all the same info.
Comparing gaming benchmarks to my i7-5775c (reminder base 3.3 turbo 3.7) with DDR3 at to my surprise the i7-5775c wins a few benchmarks and loses to some. I get it more cores and better IO but why would someone that does not do heavy professional workloads buy this (and to some extent anything that AMD bought out).
i7-11700k | i7-5775c (m-7) Strange Brigade DX12 - 1080p Ultra - Average FPS 219.6 | 225.9 Civilization VI - 4K Min - Average FPS 94.3 | 113.1 Gears Tactics - 4K Low - Average FPS 49.2 | 53
sure enough, these are handpicked and the 11700k wins some benchmarks with a big margin, but we have here 4 real cores less and 1.3 GHz less turbo and a mere DDR3 memory interface. I wonder if Intel went the wrong way by ending the L4 monster caches. At least they would have retained the gaming crown if they would introduce an "extreme" version with a decent-sized L4 low latency cache.
Hi Ian; I appreciate your work and also I have a suggestion hoping to be considered. I think it's better not to mention processors by their formal or nominal power in benches. for us, it can be much more justifiable when we know how much a processor consumes power in each test separately. I know there's a separate test for power consumption but, as you know, in such an old-fashioned yet common way, the efficacy of each processor for each test is not clear. and I'm afraid, it undermines your work's articulacy. regards =)
Charlie Demerjian was right long time ago, when he stated this was going to be a fiasco. Seems like a desperate move from intel, more than lift off, this rocket is hitting the ground.
Thanks Ian! I actually applaud the decision to go ahead and put a legally purchased CPU through the paces before the official embargo date for the official, free-for-testing samples; you were very up front how and where you got your test candidate. And, once an authorized reseller sells them, they're "public" and fair game. These performance tests are also a unique opportunity to see by just how much the official sample Rocket Lake with the then-valid BIOS firmware might differ from this store-bought one. Actually, if Intel would want to be smart about this, they'd send you a coupon to order a RL and MoBo from any authorized retailer, to avoid accusations of a "review edition" hand-selected to be the best binning possible.
Lastly, I wish you (AT) and other review sites would occasionally cross-check the results of their review samples (sent by the manufacturer) to the same unit bought through retail. I know that's not always feasible (costs $$$) , but doing so every once in while might help keep the manufacturers honest and assure us, their customers.
Partially for timeliness reasons. Partially because we've already seen Xe in Tiger Lake. Rocket Lake's Xe-LP implementation is much slower than the mobile chips, since it has only one-third the number of EUs. So while we'll collect that data eventually, it's nothing terribly exciting.
The boost clock frequency in that WCCF article for the 11700K is especially interesting; according to the table in it, the boost clock goes up to 470 GHz (: Now, that's a clock speed worth writing about!
WCCFTech has been Intel’s lap dog ever since Zen 2 hit the scene with article after article with a spin in Intel’s favor and I would not be the least bit surprised if they are just like UserBenchmark. Their notorious Usman Pirzada in particular harped on Ian for showing peak power for Rocket Lake, claiming it was disingenuous and misleading since AVX-512 is the reason and that puts the processors on unequal footing. Well, Mr. Pirzada, you forget. AVX-512 also gives a performance uplift so you better also ask to note all the benchmarks where it is engaged so we are playing fair here. If you truly want to level the playing field and throw out AVX-512, you have to throw out a key item that has contributed to an uplift in performance and turn it off in both benchmarks and power tests alike. Either both ways or no ways, sonny Jim.
"So coming to the benchmarks, only three titles were tested which include Crysis Remastered, Star Wars: Jedi Fallen Order, and Cyberpunk 2077. The test settings and resolution are not mentioned."
It's bad enough that they only showed tests from games that Anandtech don't have in their test suite - so we can't compare directly - but it's even worse that we have no data about the settings used, so nobody else could possibly compare directly, either.
Honestly, at this point I consider myself to have a good indicator of final performance, and will draw final conclusions when the embargo lifts.
Someone literally wrote ‘I commend Intel for this release’. Commend? Perhaps the word originally sought was condemn?
This product exists due to inadequate competition. It is a gift from monopolization. While the mantra of the corporation is ‘sell less for more’, it’s adequate competition that’s supposed to be its saving grace. That hasn’t been the case in many areas of tech for a long time.
Not only is AMD not enough competition, it took this long for the company to finally beat Intel badly. If Intel had managed its business better it would still be competitive.
And, to top things off, just like in GPUs (dire lack of competition) it’s extremely profitable to fail. Nvidia is failing to meet gamer demand, for various reasons that come down to inadequate competition. Despite that, it is taking record money. Intel is profitable despite failing in the consumer desktop CPU market.
Allison Kilkenny joked that America is special because one can ‘fail upward’. But, really, tech has a huge problem with inadequate competition — a global one.
People see brief glimmers of competition and mistake it for adequate competition. This is not the first time AMD had the better CPU design and we all know how that turned out.
Being held captive by one and 1/2 companies (the typical tech competition ratio) isn’t at all close to approaching ideal.
If huge profitability while failing very badly to meet demand is the recipe for a good state of business...
I believe that was me. "While it hasn't displaced Ryzen, I commend Intel for the effort that Rocket Lake is."
I should have worded it better, I agree, but what I meant was, instead of just releasing another Skylake refresh, they did the unthinkable, in an attempt to release something: porting Sunny Cove to their legacy node. "We will fail, but we'll try anyway, futile though our efforts may be." Their being the underdog at present makes me feel a bit sorry for them, but then again, when we consider Intel's bank account, we realise they don't need our sympathy.
For my part, I am not an Intel shill or proponent. In fact, from my teenage years I've had a distaste towards them, and as I see it, Rocket Lake is a disappointment: still behind Zen 3, using far more power, slower graphics than Tiger Lake. There's practically no good in it. For the elusive Sunny Cove, I had expected a lot more. Despite that, I do commend them for making an attempt, something in life that's more important than winning.
As for the business/economic side of the matter, I don't understand these things too much to comment, but from that side, Intel has always been wretched; and all these companies are only out to make profit. We've even learned that those who offer something for nothing---it comes at a cost (Google, Facebook, etc.).
Intel are very much not the underdog - they're still the 800 lb gorilla, they've just got themselves trapped inside a cage of their own making. Tiger Lake is an example of the fact that they're still capable of first-tier core designs; they're just not (currently) capable of bringing them to market on a competitive process.
They are 800-lbs on a fixed rations in self-imposed solitary confinement. Watch them slim down in a jiffy to as skinny as a rail if Alder Lake fails to impress. It was a slippery slope with Bulldozer. The only issue with Intel is the sheer amount of corporate waste. If they go down, others will be ready to swoop in and pick up the layoff broken pieces—pieces, once surrendered, that never again be recovered without a fight. Intel has never been fully challenged but they just might fall into Bulldozer-like obscurity this time around.
Hard to believe how the tables have turned in a matter of four years, and worse for Intel, this isn't the slack, "we've done a good job, let's rest" AMD of the K8 days. This time round, they're relentlessly executing, with a vigour I don't remember them possessing. Whether it's Lisa or just the pain they went through in the Bulldozer days, I don't know, but it's magnificent to watch.
Cage of their own making: I like that phrase. Well, I think for Intel, the best path is to design a microarchitecture from scratch, which I'm sure they're doing. Possibly Intel Israel. Sunny Cove, descending all the way from P6, is not holding up any more.
What an awesome deep-dive technical expose/review of this processor. No stone left un-turned - incredibly detailed investigation and easily understood explanations. With this level of detail every reader can make up their own mind about the pros/cons of this product - there's no need for me to offer an opinion. Many thanks for this well written treasure trove of technical information and the graphics are wonderful. Kudos to the entire team and keep up the Good Work (TM) !!
Wow how the tables have turned. I remember for AMD's Bulldozer review, which was supposed to be AMD's answer to Core, Bulldozer was similar to Rocket Lake...it consumed MORE power and was SLOWER than the competition. Well, here's to Intel making a comeback with Alder Lake, but 2021 clearly belongs to AMD.
Alder Lake, just confirmed in a lineup leak to use a heavy amount of little or low power cores, first has to ensure that they get heterogenous core architecture working in the Windows scheduler. The biggest roadblock Microsoft claimed to have already fixed it with Lakefield but even after they did and additionally Cinebench, per Intel’s special emergency request, to recompile a special update to R20 after that, Lakefield was an absolute failure in performance compared to a Y-class (or M notated) Core product. Personally, I have little faith in Microsoft and I also do not expect software companies recompiling every last piece of software made in the last decade either just to accommodate a radical new design all for Intel’s sake. Developers would much rather assume a simple, vanilla core architecture than an unbalanced one with little and big cores and therein lies a huge risk factor for Intel. I foresee a huge misfire here because Microsoft isn’t exactly a good company to bet on refixing what they claimed they have already fixed with Lakefield.
I'm sceptical of this whole fast/slow core approach myself and believe it's a mistake. They will learn the hard way. Perhaps it works for Apple, I don't know, but Intel and AMD are better of spending their efforts in cutting down power in their ordinary cores dynamically.
It works for Apple because they have a iron-tight grasp on the full stack which requires developers to use their latest compilation and development tools to get their apps to the App Store and ultimately in users’ hands. You can run pretty much anything under the sun for 2 decades plus on Windows and without a recompile, you will never utilize their cores correctly. The Windows scheduler as it stands is garbage tier compared to Linux with a plain vanilla homogenous core design so I expect very little from Microsoft unless they actually redesign their scheduler from the ground up. Don’t hold your breath on that one.
Only if their single thread performance sucks (which from all I've seen, it won't)---Bulldozer's problem was being behind in manufacturing process and requiring a highly-threaded application to shine.
Well, the problem then is how smartly it assigns threads to the right cores and the Windows scheduler still has serious problems with that to this very day. If it doesn’t do that right, it could appear like an overclocked Core i3 in comparison to Ryzen and that is where the Bulldozer analogy is fitting. AMD banked on their unique topology getting OS and development optimization down the road—FineWine to coin a phrase—but that never materialized.
Not really---the Bulldozer design had far more problems than just the inadequate Windows scheduler---while "optimized" linux kernels gave better performance, the true issue with Bulldozer was that you had 8 relatively weak cores versus 4 strong cores from Intel, and at the time, coders were still struggling to optimize for anything over 2 cores.
In a sense, the designers of Bulldozer REALLY misread the timeline of highly multithreaded coding taking over the market. Heck, even if it was released now alongside the equivalent Core 2 quads, it would still stink for the majority of users, because no game gets any significant advantage of 4+ physical cores even today---and most games still value high single thread/core performance.
'the true issue with Bulldozer was that you had 8 relatively weak cores versus 4 strong cores from Intel, and at the time, coders were still struggling to optimize for anything over 2 cores. In a sense, the designers of Bulldozer REALLY misread the timeline of highly multithreaded coding taking over the market.'
My guess is that AMD designed Bulldozer for the enterprise market and didn't want to invest in an additional design more suited to the consumer desktop space. Instead, its additional design priority was the console scam (Jaguar). While that was a good move for AMD it wasn't beneficial for consumers, as consoles are a parasitic redundancy.
One thing many ignore is that Piledriver supercomputers occupied quite high spots in the world performance lists. Mostly that was due to the majority of their work being done by the GPUs, though. Even the original Bulldozer, in Opteron branding, was used in some.
The cheapness of Piledriver chips was also probably a factor in the adoption of the design for supercomputers. Turn down the voltage/wattage so that you're in the efficient part of the improved 32nm SOI node and rely almost completely on heavily threaded code when not running GPU-specific code... and voila — you have an alternative to the monopoly-priced Intel stuff.
But, on the desktop, Piledriver was a bad joke. That's because of its very poor single-thread performance mainly. Not everything can be multi-threaded and even if it is that can mean a speed regression sometimes. The slowness of the L3, the lack of enough operations caching... the design wasn't even all that optimized for multi-thread performance — especially FPU stuff. The cores were very deeply pipelined, designed to use very high clocks. They were not efficient with avoiding bubbles and such. I read that AMD relied too heavily on automated tools due to cost sensitivity.
My vague understanding of the design is that it was narrow and deep like the Pentium 4. Why AMD tried NetBurst 2.0 is beyond me. Even for the enterprise market it's a bad move because power efficiency is important there, especially with servers (rather than supercomputers which, I think, were more tolerant of high power usage – in terms of acceptable design requirements). Even turning down the clocks/voltage to get the best efficiency from the node doesn't fix the issue of the pipelining inefficiency (although hand-tuned code used for some enterprise/scientific stuff would mask that weakness more than general-purpose consumer-grade apps would).
Intel know very well that is mistake, but alder lake's big cores are not much better in term of power consumption and 10nm quality wafers so their only choice is to keep the production of max 8 core mainstream. This is bad for them, because even 3 years after AMD show 16 core mainstream they can't and as our very well known Intel they will cheat by introduce their 16 core cpu with 8 fake cores. Even more, when they introduce their "super duper" 12900k with 16 cores they will set price higher than every amd main stream and this is triple win, cheap 8 core cpu for manufacturing, better binned with higher clocks and on price of $800-1000. If you think intel trying to do something new and innovative you are wrong.
You may well turn out to be right about 16 cores having 8 junk ones, and knowing Intel, that's how they operate, with smoke and mirrors when they can't compete properly.
The irony here is how Intel used to give ARM smack for having inferior single core performance while they were surpassed in multicore by the likes of Qualcomm. I believe—paraphrasing—what they would say is not all cores are created equal. Well, it looks like Intel is trying to look like they are maintaining parity when they are really just giving us mostly crappy cores that can’t perform well at all.
Also, as others have pointed out before, the nomenclature is just there to obfuscate the whole picture. Not knowing anything much about Alder Lake, I did some searching and saw that it's Golden Cove + Gracemont. Wondering what exactly GC was, I searched a little but couldn't find the answer, so I'll guess it's just Sunny Cove with a new name stuck on.
the more I read about Intel's 10nm (check relevant articles on SemiWiki and Semi accurate), the more I feel ADL is designed as a workaround for the power/thermal limitations of their 10nm process, instead of being some sort of revelation for next gen performance.
Note how TGL remains Intel's only viable product on 10nm at the moment, with Ice Lake SP now two years late. I think Intel knows their 10nm may never be ready for desktop parts, so ADL is a way to have a desktop product on 10nm except not really (it's more akin to a mobile part).
It will probably do fine for gaming, but highly doubtful it will be a meaningful competition to Zen 4 for the prosumer space.
Bulldozer/Piledriver were a dumb design because they relied HEAVILY on highly-threaded applications to achieve their performance. Almost none of which existed in the late 2000s when they launched. Single thread/core performance was absolutely pathetic compared to Intel's offerings at the time (Sandy Bridge and on).
And yet here we are today with 8 cores in game consoles and AMD's Mantle API being the basis of both DirectX 12 and Vulkan. Multicore was always the future and it was obvious even in 2003.
"And yet here we are today with 8 cores in game consoles and AMD's Mantle API being the basis of both DirectX 12 and Vulkan. Multicore was always the future and it was obvious even in 200"
You're missing the point. Even TODAY, few (if any) games gain an advantage of more than 4 physical cores + 4 Virtual. It is still, to this day, far more advantageous for games to have 4 strong physical cores than 8 weak ones. The latter was Bulldozer.
bored. just here to say this is unsurprising though this strongly reminds me of the time where AMD is releasing new, well designed CPUs but two process node generations behind intel. I think AMD was 32nm and 28nm while Intel is 22 and 14nm. most comments were really harsh with AMD but I reasoned that it is simply due to the manufacturing superiority of Intel
Still, within that mess, AMD did a pretty good job raising Bulldozer's IPC and cutting down its power each generation. But the foundation being fatally flawed, it was hopeless. I believe it taught them a lot about cutting power and so on, and when they poured that into Zen, we saw the result. Bulldozer was a fantastic training ground, if one looks at it humorously.
Firstly, Bulldozer had worse IPC than Phenom. No engineers with brains release a CPU to replace the entire line while giving it worse IPC. The trap of going for high clocks was a lesson shown to the entire industry via Netburst. AMD's engineers knew all about it, yet someone at the company decided to try Netburst 2.0.
Secondly, AMD was so sloppy and lazy that Piledriver shipped with a performance regression in AVX. It was worse to use AVX than to not use it. How incredibly incompetent can the company have been? It doesn't take a high IQ to understand that one doesn't ship broken AVX.
AMD then refused to replace Piledriver until Zen came out. It tinkered half-heartedly with APU rubbish and focused on pushing junk like Jaguar.
While it's true that the extreme failure of AMD (the construction core line) is due, to a large degree, to Intel abusing its monopoly to starve AMD of customers and cash — cash it needed to do R&D, one does not release a new chip with worse IPC and then very shortly after break AVX and refuse to stop feeding that junk to customers for many years. Just tinkering with Phenom would have been better (Phenom 3).
As for the foundation claim... we have no idea how well the CMT concept could have worked out with competent engineering. Remember, they literally broke AVX in the Piledriver revision that was supposed to fix Bulldozer enough to make it sellable. Operations caching could have been stronger. The L3 cache was almost as slow as main memory. The RAM controller was weak, just like Phenom's. Etc.
We paid for Intel's monopoly and we're still paying today. Only its monopoly and the lack of adequate competition is enabling the company to be so profitable despite failing so badly. Relying on two companies (or one 1/2, when it comes to R&D money ratio and other factors) to deliver adequate competition doesn't work.
Google and Microsoft = Google owns the clearnet. Apparently, they have some sort of cooperation agreement which helps to explain why Bing has such a tiny index and such a poor-quality search.
TSMC and Samsung = Can't meet demand.
AMD and Nvidia = Nvidia keeps breaking profit records while utterly failing to meet demand. Both companies refuse to stop making their cards attractive for mining and have for a long long time. AMD refused to adequately compete beyond the lower midrange (Polaris forever, or you can buy a 'console'!) for a long time, leaving us to pay through the nose for Nvidia's prices. AMD literally competes against the PC market by pushing the console scam. Consoles are gaming PCs in disguise and they're parasitic in multiple ways, including in terms of wafer allocations. AMD's many many years of refusal to compete with Nvidia beyond the Polaris price point caused so much pent-up demand and now the company can enjoy the artificially high price points from that. It let Nvidia keep raising prices to get consumers used to that. Now that it has finally been forced to improve the 'consoles' beyond the garbage-tier Jaguar CPU it has to offer a bit more value to the PC gaming market. And so, after all these years, we have something decent that one can't buy. I can go on about this so-called competition but why bother. People will go to the most extravagant lengths to excuse the problem of lack of adequate competition — like the person who recently said it's easier to create Google's empire from scratch than it is to make a competitive GPU and sell it as a third GPU company.
There are plenty of other areas in tech with inadequate competition, too.
"AMD then refused to replace Piledriver until Zen came out. It tinkered half-heartedly with APU rubbish and focused on pushing junk like Jaguar."
To be fair, AMD had put a LOT of time, money and effort into Bulldozer/Piledriver, and were never a company with bottomless wells of cash to toss an architecture out immediately. Plus, Zen took a long time to design and finalize---thankfully, they made literally ALL the right moves in designing it, including hiring the brilliant Jim Keller.
I think if Zen had been another BD like failure, that would have been the almost the end of AMD in the cpu market (leaving them basically as ATI was) The consoles likely would have gone with Intel or ARM for their next iteration. AMD once again spent tons of money that they don't have as disposable income in designing Zen. Two failures in a row would have been disastrous.
Heck, the consoles might go with their own custom ARM design for PS6/Xbox(whatever) anyways.
AMD did not put a lot of resources into fixing Bulldozer.
It shipped Piledriver with broken AVX and never bothered to replace Piledriver on the desktop until Zen.
Inexcusable. It shipped Steamroller and Excavator in cost-cut mode, cutting cores, cutting clocks, cutting the socket standards, and cutting cache. It used a dense library to save money by keeping the die small and used the inferior 28nm bulk process.
'They did try to at least 'ride it out' until Zen could get done, and that required smoothing out the rough edges, so they did devote some resources.'
Wow... watch the goal posts move.
Riding out = doing nothing. Piledriver was not improved. The entire higher-performance & supercomputer market was unchanged from Piledriver to Zen. All AMD did was ship cheap knock-off APU rubbish and console trash.
The fact that AMD succeeded with Zen is probably mostly a testament to one largely ignored feature of monopoly power: the monopolist can become so slow and inefficient that a nearly dead competitor can come back to best it. That's not symptomatic of a well-run economic system. It's a trainwreck.
AMD should have been wealthy enough to do proper R&D and bulldozer would have never happened in the first place. But, Intel was a huge abusive monopolist and everyone went right along, content to feed the problem. After AMD did Bulldozer and Piledriver the company should have been dead. If there had been adequate competition it would have been. So, ironically, AMD can thank Intel for being its only competition, for resting on its laurels because of its extreme monopolization.
Oxford Guy. I don't remember the exact details and am running largely from memory here. Yes, I agree, Bulldozer had far lower IPC than Phenom, but, according to their belief, was supposed to restore them to the top and knock Intel down. In practice, it failed miserably and was worse even than Netburst. Credit must be given, however, for their raising Bulldozer's IPC a lot each generation (something like 20-30% if I remember right), and curtailing power. It also addressed weaknesses in K10 and surpassed K10's IPC eventually. Anyway, working against such a hopeless design surely taught them a lot; and pouring that knowledge into a classic x86 design, Zen, took it further than Skylake after just one iteration.
AMD would have done better had they just persisted with K10, which wasn't that far behind Nehalem. But, perhaps we wouldn't have had Zen: it took AMD's going through the lowest depths, passing through the fire as it were, to become what they are today, leaving Intel baffled. I agree, they were truly idiotic in the last decade but no more. May it stay that way!
Concerning CMT, I don't know much about it to comment, but think Bulldozer's principal weakness came from sharing execution units---the FP units I believe and others---between modules. Zen kept each core separate and gave it full (and weighty) resources, along with a micro-op cache and other improvements. As for Jaguar, it may be junk from a desktop point of view, yes, but was excellent in its domain and left Atom in the dust.
'Credit must be given, however, for their raising Bulldozer's IPC a lot each generation (something like 20-30% if I remember right), and curtailing power.'
Piledriver was a small IPC improvement and regressed in AVX. Piledriver's AVX was so extremely poor that it was faster to not use it. Piledriver was a massive power hog. The 32nm SOI process node, according to 'TheStilt' was improved over time which is probably the main source of power efficiency improvement in Piledriver versus Bulldozer. I do not recall the IPC improvement of Piledriver over Bulldozer but it was nothing close to 20% I think. Instead, it merely made it possible to raise clocks further, along with the aforementioned node improvement. And, 'TheStilt' said the node got better after Piledriver's first generation. The 'E' parts, for instance, were quite a lot improved in leakage — but the whole line (other than the 9000 series which he said should have been sent to the scrapper) improved in leakage. What didn't improve, sadly, is the bad Piledriver design. AMD never bothered to fix it.
While Piledriver, when clocked high (like 4.7 GHz) could be relevant against Sandy in multi-thread (including well-threaded games like Desert of Kharak) it was extremely pitiful in single-thread. And, it sucked down boatloads of power to get to 4.7, even with the best-leakage chips.
And, going back to your 20–30% claim. Steamroller, which was considered a serious disappointment, featured only 4 of the CMT quasi cores, not 8. Excavator cut things in cache land even further. Both were cost-cutting parts, not performance improvements. Piledriver killed both of them simply by turning up the clocks high. The multi-thread performance of Steamroller and Excavator was not competitive because of the lack of cache, lack of cores, and lack of clock. Single-thread was a bit improved but, again, the only thing one could really do was blast current through Piledriver. It was a disgusting situation due to the single-threaded performance, which was unacceptable in 2012 and an abomination for the later years AMD kept peddling Piledriver in.
The only credit AMD deserves for the construction core period is not going out of business, despite trying so hard to do that.
Oxford Guy, while I respect your view, I do not agree with it, and still stand by my statement that AMD deserves credit for improving Bulldozer and executing yearly. Agreed, my 20-30% claim was not sober but I just meant it as a recollection and did qualify my statement.
I don't think it's fair to put AMD down for embarking on Bulldozer. When they set out, quite likely they thought it was going to go further than the aging Phenom/K10 design, and the fact is, while falling behind in IPC compared with K10, it improved on a lot of points and laid the foundation. Its chief weakness was the idea of sharing resources, like the fetch, decode, and FP units, as well as going for a deeper pipeline. (The difference from Netburst is that Bulldozer was decently wide.)
Piledriver refined the foundation, raising IPC and adding a perceptron branch predictor, still used in Zen by the way, and I believe finally surpassed K10's IPC (and that of Llano). While being made on the same 32 nm process, it dropped power by switching to hard-edge flip flops, which took some work to put in. They used that lowered power to raise clock speeds, bringing power to the same level as Bulldozer. And Trinity, the Piledriver APU, surpassed Llano. I need to learn more about Steamroller and Excavator before I comment, but note in passing that SR improved the architecture again, giving each integer core its own fetch/decode units, among other things; and Excavator switched to GPU libraries in laying out the circuitry, dropping power and area, the tradeoff being lower frequency.
I don't need to look at reviews agaih. I know how bad the IPC was in Bulldozer, Piledriver, Steamroller, and Excavator. Single-thread in Cinebench R15, for instance, was really low even at 5.2 GHz in Piledriver. It takes chilled water to get it to bench at that clock.
Lack of competition, high prices, lack of integrity. I agree it's one big mess, but there's so little we can do, except boycotting their products. As it stands, the best advice is likely: find a product at a decent price, buy it, be happy, and let these rotten companies do what they want.
'find a product at a decent price, buy it, be happy'
Buy a product you can't buy so you can prop up monopolies that cause the problem of shortage + bad pricing + low choice (features to choose from/i.e. innovation, limited).
I know is a strange request, but i would like to know the iris integrated graphics benchmarks since i'm using my old 2080 ti for mining and i'm playing Minecraft and simple games with my integrated uhd630 of my 9700k, and unfortunately 5900x does not have integrated graphics, so i would like to know 11700k and 11900k perf with that, i have seen mobile benchmarks but as you know, is not the same thing, would like to see quality gaming benchmark as always, from you. thanks
Would also be interested in this. I sold my 2070 Super - I owned it for a year, and sold it for what I paid (so free card for a year). The idea was to buy a 30X0 card with that money. That didn't happen, so lately I've just been playing Minecraft and older games on an old GTX 460. I'm curious about how the Xe graphics compares - with current prices on Ebay, the graphics along can add about $60 worth of value to the cpu.
ASUS just released another BIOS update with Rocket Lake enhancements. Probably more to come closer to the release too. This is why you don't post your review 3 weeks early.
Like maybe an AVX-512 down clocking offset? Either Intel released their Rocket Engine a quarter too early or no amount of BIOS tweaking can do what you think it can do, at this, or any, point in time.
From this review "Looking at our data, the all-core turbo under AVX-512 is 4.6 GHz, sometimes dipping to 4.5 GHz. Ouch. ... Our temperature graph looks quite drastic. Within a second of running AVX-512 code, we are in the high 90ºC, or in some cases, 100ºC. Our temperatures peak at 104ºC ... "
So already thermal throttling at Intel's promised 4.6 all core frequency using AVX-512. Makes you wonder what it takes to significantly OC this CPU. Which, you know, has barely been mentioned here in the comments section, OC'ing the damn thing, north or south of 300W or ~300W ... https://i.imgur.com/8BEsGVo.png
I guess you missed also the spot where normal AVX used less power than the 9900k. The vast majority don't care about AVX-512. It is just there so Intel can say it is. People who buy Rocket Lake will be interested because of gaming and there will probably be more stock than 7nm products from AMD.
wow. really ? one test ( of a few) where intel was faster, and used less power ? big deal. over all rocket lake, looks to be a joke. " People who buy Rocket Lake will be interested because of gaming " wrong, i know a few peope who are not even looking at intel, and are just waiting for zen 3 to be available, and this is for gaming and non gaming usage.
Yes, a 5.0GHz (all core boost clock) at 231.49W for the i9-9900KS versus a 4.6GHz (all core boost clock) at 224.56W for the i7-11700K. Conclusion? The i7-11700K runs 20-25W higher at the same all core boost frequency (4.6-5.0GHz). The i7-11700K wins at test duration though (by a similar margin as the inverse of the power ratio). The CPU energy used is about the same for both.
1) Why is apple silicon or ARM equivalents not part of the benchmarks? 2) Why are so many CPU benchmarks needed, especially if they don't tell anything significant about them. 3) I'm not a huge gamer, but I also don't understand the point of so many gaming benchmarks for a CPU review.
Perhaps I'm the wrong audience member here, but it does seem a whole lot of charts that roughly say the same thing!
1: AT benches on Windows, and right now x86 vs ARM is kinda Apples-to-oranges on that platform, especially when one starts mixing in emulation and AVX.
Give it time. More comparisons will come. But you'll probably see comparisons on Linux/Mac, and open source software in general, sooner.
2: People uses CPUs for different things. Some of these benches are relevant to those people.
At the same time, my use cases weren't really covered here, so... I get what you're saying.
3. Yeah, it seems rather silly to me, especially when Anand test GPU limited AAA games.
Where you really need a big CPU is in simulation/sandbox games, especially in servers for such things, and sluggish early access stuff. But no one ever benches those :/
So with the 19% IPC claim and losing to 9900KS and 10700K what is the point of releasing this chip, from Intel. I never got much from AT benches a lot and preferred the Hardware unboxed, Gamernexus guys and others. But if this is the final performance figures, then this is really a DOA product from Intel. How can they allow this ? I never saw Intel in such a position..maybe X299 got rekted when Zen 2 dropped but this is mainstream segment.
Damn it. AMD processors have the idiotic stock related issues, add that WHEA and USB shitstorm. Intel has bullshit performance over past gen except a Gen4 addition and extra lanes from chipset. GPUs are out of damn stock as well.
2020 and 2021 both are completely fucked up for PC HW purchases.
The Pentium 4, in particular the Prescott architecture was a dud back in the early 2000's. That era spawned the antitrust lawsuit against Intel for illegally blocking AMD sales since the Intel products weren't competitive.
The fact that Intel can even remain in a close second place, using a 14 nm process is impressive. Imagine what they could do with TSMC's 7 nm process! It would almost certainly outperform AMD by a significant margin.
architectures are designed for specific nodes - RKL's problems are exactly due to porting an arch onto a node it wasn't designed for.
the fact is Intel is not a partner for TSMC and their archs are not designed for TSMC processes. if Intel were to outsource CPU production to TSMC, they will either have to make a new arch or make make tweaks to existing ones - a multi-million $ endeavor with risms of issues like your just read with RKL.
^Here we see in his natural environment your common everyday dude who fails at reading comprehension. I guess you didn't read the part about the serious gaming performance losses and latency regressions gen-over-gen, the 10% performance gap in single-threaded or 10-20%+ performance gaps in multithread, or the inexcusably high peak power draw? Talk about deluded...
Though their efforts may be futile, I am glad Intel attempted to do something out of the ordinary; not a Skylake refresh, but a backport that is found to have worse performance. And yet, it is an attempt for Intel to learn their lessons for generations to come.
Backporting is not a lesson; it is a last ditch effort or a fallback when all else fails on the manufacturing side. Half full, half empty cup viewing aside, they wasted even more valuable engineering manhours into a failed backport when it should have been invested into developing new architectures. A best use would have been developing the next release. The problem is Intel had to make Rocket Lake good enough in synthetic benchmarks to appease their investors. That, however, still does not address the elephants in the room of 10-20% single threaded performance gaps or—the one that takes the cake—the latency regressions that makes gaming worse, Intel’s historic crown jewel. Much like movies that fail at release and live on box office bombs that their producers later opine should have been cut early on in development, Intel should have cut this idea early on. If you are looking for a lesson that Intel should have learned here, there it is: avoid another Rocket Lake backporting disaster and just warm over your current microarchitecture with one more middling refresh one last time.
Ian, I just want to say thank you for the incredible review. Just ignore the haters on social media and in the comments who get their panties in a bunch. If the product is garbage, say it like it is, like you did and quite well I might add. You were incredibly diplomatic about it and even openly and honestly showed when and where Intel did win on the rare occasion in the benchmarks. It is so silly how people make these CPU companies (who don't know them from Sam Hill) their religion, as if erecting a Gordon Moore or Lisa Su shrine would avail them anything. Silly geese.
Bulldozer came from a money-starved little company, a company that couldn’t capitalize on having the superior design because of Intel’s monopolization.
This chip is from the big rich company that stomped on AMD with dirty tricks.
So, no. As stupid as Bulldozer was, this CPU seems to be quite a bit more unjustified.
First off, let's take a quick looksie at the Cinebench R20 results: https://www.hardwareluxx.de/index.php/artikel/hard... When switching from BIOS version 0402 to 0603, the 11700K's single-threaded performance actually DROPS from a score of 609 to 600. And its multicore performance is still less than the 10900K and the 5800X. Switching gears, the games are no less unflattering: https://www.hardwareluxx.de/index.php/artikel/hard... The 11700K there, regardless of which of the two BIOS releases it uses, often loses to the 10900K and Ryzen 5000 series. It loses to the Ryzen 5000 series and 10900K in THREE out of the four games: The Division 2, Metro Exodus, The Shadow of the Tomb Raider.
As expected new reviews with newer BIOS versions improve performance significantly and puts the numbers to expected levels as well. This was a quick shot of a review and I fear it has tainted Cutress reputation a lot, especially because he defended it that much, even on video.
Raising performance by pushing power even higher may improve things like FPS in gaming tests but it obscures the big picture.
Want to talk about reputation? Remember the giant fridge-sized chiller Intel surreptitiously used to give a benchmark demo? Or, remember the ‘GenuineIntel’ fiasco? Or, remember the cute trick of putting a black box CPU inside the one people pay for, so that only special customers get the option of avoiding that particular spyware?
Somehow I think the writers here are going to be very hard pressed to challenge Intel in the cheatiness department, even without mentioning Intel’s history of abusing its monopoly power via OEM deals and the like.
Beaver M " new reviews with newer BIOS versions improve performance significantly " from what i can tell from the graphs, for the most part, while performance may have improved over previous gen, it looks like it still looses to zen 3, but definitely not significantly, and still using more power then zen 3 overall, not that much of an improvement. a little upset that intel didnt get any performance crowns back, maybe, and that rocket lake still looks to be a dud ?
I fail to see the performance improvements in this other review. the games chosen were different (only 3 games? really?), and still lost to 5800X in 2/3.
is it just "better" relative to last gen performance? the other review only tested a single resolution (and again, 3 whole games!)
First off, let's take a quick looksie at the Cinebench R20 results: https://www.hardwareluxx.de/index.php/artikel/hard... When switching from BIOS version 0402 to 0603, the 11700K's single-threaded performance actually DROPS from a score of 609 to 600. And its multicore performance is still less than the 10900K and the 5800X. Switching gears, the games are no less unflattering: https://www.hardwareluxx.de/index.php/artikel/hard... The 11700K there, regardless of which of the two BIOS releases it uses, often loses to the 10900K and Ryzen 5000 series. It loses to the Ryzen 5000 series and 10900K in THREE out of the four games: The Division 2, Metro Exodus, The Shadow of the Tomb Raider.
The 10700K costs a pittance at the minute, and after the final bios and microcode 11700K will be around 15 to 20% faster than it. Both a bargain when the only other option is dealing with the 'quirks' (to be kind) of an AMD system
" The 10700K costs a pittance at the minute, and after the final bios and microcode 11700K will be around 15 to 20% faster than it " i will believe that, when i see it, honestly, that is hopeful thinking. " Both a bargain when the only other option is dealing with the 'quirks' (to be kind) of an AMD system " and intel has had its own quirks over the years.
They’re just doing this to give the impression they’re relevant. It’s safe to assume they’ll retake some market share but AMD took servers, laptop and desktop sectors by storm while Intel keeps moving old designs out the door.
AMD 5750G, if it exists will render Intel designs useless this summer, while Intel struggles to get Alder Lake up and running.
Motherboard manufacturers will get tired of chasing new sockets after AM5 comes out.
What happened to these guys? It’s embarrassing and I’m an i7 fan boy..
My guess would be: complacency, underestimating the enemy, putting eggs in too many dead-end baskets, and management that made a mess of excellent engineering talent.
Motherboard manufacturers will get tired of chasing new sockets after AM5 comes out.
When reading this I think some of you just ended their school yesterday (with poor results) and just came here to say the first thing they think it is pro AMD. Just to give a (poor) contribution to what they think is an easy (for everyone) task as beating a dead horse (Intel). I may shock you if I say that that "chasing designs" effort is the secret trick for motherboard producer too... surprise surprise.. MAKE MONEY! So they do not get tired to do anything if this means selling more motherboards, and this just happens if you have to change your motherboards every couple of generations. And I may shock you even more if I say you that those that make an upgrade using the same motherboard is just a so small number that no motherboard producer is really interested in supporting. Usually when you change your CPU you just do not want only it to go a little faster but you want also the new technological improvements that meantime have been created, from faster bus, new and more connectors (M2 vs SATA), faster USBs, Thunderbolts, better memory support and such. And this doesn't come if you do not also change the motherboard. And to have and propose a better motherboard to sell, guess what? Yes, motherboard producer have to play the "chasing designs" game.
i7 10700K actually beating the 5800x in many game benchmarks. I don't have a preference between AMD and Intel generally, but the i7 10700K is a great gaming processor, Intel did seem to make an embarrassing move with the 11th Gen, but for the cost, the 10700k a top, possibly THE top gaming CPU.
When an official review comes in, with all the details, things may look a little better. But even now, I see one thing that's being overlooked. Since these chips have AVX-512, where that can be used, that will double their performance compared to processors that only have AVX-256. Except, of course, for the necessary slowdown for thermal reasons. So on workloads that involve a lot of AVX-512, they should really shine instead of being as terribly lackluster as they appear when that isn't taken into account.
Yeah, might have to agree with you on this, from what I've seen the new instructions are amazing. They might be a real game changer but one thing is certain, it's the way of the future.
(100% speed performance = i9-9900k) the values are from userbenchmark dot com, and can be checked your selves. you might also want to go and check how these values come together b4 making any statements!
in comperison with these values you can clearly see that the i7-11700k is the top runner! not only in gaming but also in number crushing. tho i would still more than recomend the ryzen9 5900x/5950x or a threadripper if you realy depend in numbercrushing like renderings and such.
also worth mentioning is the I7-11400F if youre on a budget. best gaming cpu for that price.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
541 Comments
Back to Article
Zoeff - Friday, March 5, 2021 - link
Now this is a welcome surprise!nandnandnand - Friday, March 5, 2021 - link
This is actually Intel's best product launch ever. Because you can buy it and get in on the class action lawsuit later.Zoeff - Friday, March 5, 2021 - link
To be clear, I was referring to the article itself.But yes the performance is also a surprise for me after going through the article. I expected a nice performance uplift, perhaps a bit faster than Ryzen 5000 but with a higher power draw as a trade off for being on the same Intel 14nm node still. Definitely did NOT expect it to be slower.
nandnandnand - Friday, March 5, 2021 - link
To be clear, I was replying to you to get a reply at the top of the comments. Completely self-serving manipulation of AnandTech's comment section.Hopefully, terroradagio is right and an update will improve performance and efficiency slightly. Also, I wonder how the 11900K will look going ~200-300 MHz above this.
barich - Friday, March 5, 2021 - link
I have a feeling that the additional power draw required to get that extra 2-300 MHz is going to be well out of proportion to the performance gained by it. This chip is already pushed past the edge just to not regress (usually) over the previous generation.whatthe123 - Friday, March 5, 2021 - link
Considering this chip is already drawing way too much power the 11900's are probably binned like crazy and run a much lower voltage. Even then it'll probably still use more power than the 11700k. I think they should've just accepted that 14nm was not going to work for this backport. There's a market for cometlake in gamers but what market is there for this? Niche scientific computing?terroradagio - Friday, March 5, 2021 - link
Games don't require the type of power people here are crying about. Honestly, do people just sit at their computers and run AVX benchmarks all day?barich - Friday, March 5, 2021 - link
You're still looking at like an 80W difference and a requirement for the best air cooling available if you don't want temps to be out of control. I might be willing to tolerate the extra power draw if I got more performance out of it, but that's not the case.eva02langley - Saturday, March 6, 2021 - link
You mean LIQUID COOLING at this point. You need a good AIO with such power requirements.JimmyTheFish - Friday, March 5, 2021 - link
No, power consumption is exactly as bad as people are "crying" about, this chips is a joke, hotter, slower and more expensive than even the quite Sub-Par cometlake chips on the market, nevermind the superb Zen 3 offerings, which are starting to see increased availability with the single CCD 6 and 8 core 5600X and 5800XTo even be remotely appealing the 10700K need to be cheap, sub $350 cheap, and thats just not gonna happen
JimmyTheFish - Friday, March 5, 2021 - link
*11700KDamn this stupid naming scheme.
EasterEEL - Saturday, March 6, 2021 - link
I'm still struggling to find a 5800X at a sensible price and now waiting to see how the USB connectivity issues with 500 series motherboards is resolved.Cooe - Saturday, March 6, 2021 - link
...R7 5800X was literally on sale at MSRP all day yesterday on Amazon. Probably still is tbh.Freeb!rd - Saturday, March 6, 2021 - link
A quick search shows it available at Newegg & B&H both at MSRP; wait, my bad B&H added 99cents to $449.99.Kallan007 - Saturday, March 6, 2021 - link
I don't think you are even trying; Amazon has AMD Ryzen 7 5800X in stock at Price: $449.99 ; https://www.amazon.com/dp/B0815XFSGK?tag=nismain-2...Gich - Saturday, March 6, 2021 - link
Almost always available on AMD own shop.Marlin1975 - Saturday, March 6, 2021 - link
https://smile.amazon.com/AMD-Ryzen-5800X-16-Thread...Selling at MSRP, not some marked up price all day long.
Cooe - Saturday, March 6, 2021 - link
Stop lying. The R7 5800X is EVERYWHERE at MSRP.Qasar - Sunday, March 7, 2021 - link
no its not, 2 comp stores here, are still sold out of zen 3Bfree4me - Sunday, March 7, 2021 - link
MicroCenter had it priced at $285 just last week.jacropolis - Sunday, March 7, 2021 - link
The 10700K is regularly $279 at Micro Center and isn't a bad buy for that price at all.Jadi - Sunday, March 7, 2021 - link
350 Dollar you want a 10700k and comparing it with a 5800X?Why?
The 5800X don t have a gpu, and the 10700ks was under 320 euros allways!
Now the 10700K not KS cost 316 euros!
Hot a 14nm Chip this is the advantage its only 60 c and a 5800X comes easy over 85 c and more.
More expansiv a 5800X cost 440 euros! A 5600X cost 350 euros, more then the 10700k!
Slower its a K CPU and can boost it up. When you OC the 5800X you have less fps.
Why you think it needs more power? Becouse the 40 sec boost? That is all and under AVX512 is it the only cpu that have it.
Games and other stuff its the same power, And 5800X dosen t stop at 105 Watt it goes like 118 watts. Thats are very close too 125 watts and more is only a limit time!
Spunjji - Monday, March 8, 2021 - link
@Jadi - that last bit of your post makes no sense. The 10700K and 11700K objectively require more power than the 5800X to do the same amount of work; the 5800X only hits 118 Watts under the same conditions that the other two approach 200 Watts. If you're comparing power draw during gaming, the 5800X uses less power than those two - roughly 25W less than the 10700K.Comparing temperatures is daft too. 85 degrees is way under the CPU's junction temperature, so it's not a problem to runa cPU at that temp - and it's objectively easier to keep an AMD CPU at that temperature than to keep an Intel chip at 60 degrees, because your cooler isn't dissipating as much heat.
TrueJessNarmo - Tuesday, March 9, 2021 - link
10700k is $320 on amazon and 9900k is $250 at microcenter.At this point 5900x is best for serious users and 9900k/10700k is for gamers. 5600x and 5800x are overpriced for what it is, but 5900x is a sweet deal at $550
rfxcasey - Wednesday, March 17, 2021 - link
I got my 10700K for 330 dollars new, I've seen them go for as low as 300.Makste - Saturday, March 6, 2021 - link
With gaming it's not the power consumption that's the problem here, it is the gaming performance in these benchmarks against comet lake obviously due to a higher latency.Zen 3's efforts have been put into perspective here, those engineers did quite a commendable job.
Bfree4me - Sunday, March 7, 2021 - link
Agreed! The Zen architecture is top shelf. But it took far too long. My last AMD Cpu was an Athlon 1800 because at that time AMD was the value leader and Intel was the performance king for a price. At the time came where I needed performance over value.So I have been Intel since then. Time to start looking towards Zen 🤔
Timoo - Monday, March 8, 2021 - link
If I may be so free: "Took way too long": I agree.I just keep in mind that it took them way too long, because Intel managed to create some very shady deals with retailers, effectively pushing AMD out of the market and to the brink of bankruptcy. So, for years AMD has been struggling to stay alive.
Agreed, buying ATi didn't help either. But without those shady deals they might have pulled it off. From 2006 to -say- 2015 AMD was only struggling to stay alive. I believe they had some huge cash injection at the time, together with getting one of the best CPU designers in the world (what's his name?) back in the house.
So, in my opinion, AMD is forgiven for their 10-year "heart-attack".
GeoffreyA - Monday, March 8, 2021 - link
I also believe it was because the K10, or Phenom, while being faster than the Athlon 64, was not aggressive enough. It had already been designed before Core 2 Duo came out, and by then, Intel was ahead of them. Who knows, they might have narrowed the gap if they had persisted with the K10 design, widening everything till it matched/surpassed Nehalem, but they shot too far, going for Bulldozer, which was supposed to turn the tables round but sank them further in the ditch.TheinsanegamerN - Monday, March 8, 2021 - link
And yet AMD was having trouble supplying enough chips to the limited market they did have. Much like today. OEMs are not going to sign up with a company that cant consistently supply CPUs.Dont forget the billiosn they poured into making GF and pissing off TSMC in the process.
Geef - Tuesday, March 9, 2021 - link
What, you don't?Hifihedgehog - Saturday, March 6, 2021 - link
I predicted ~300W peak power about a year or so ago when I first heard they were making the big mistake of bringing AVX-512 to mainstream consumer processors. Why on earth? There was a valid, wise, very, very, very good reason Intel had reserved AVX-512 to just their 14-nm HEDT processors and that had always been the furnace-like heat and nuclear-like power consumption of it. Even with the best logical design improvements from a new microarchitecture, it is still an extremely intensive logical pill of a task to swallow. Now, we see that reason in full, unadulterated display. You bring a massively complex instruction set extension to a higher process node where you have far lengthier physical networks (meaning essentially longer wires, increased resistance, higher power, and maximum heat) and, of course, you are going to have a steaming pile. Remember this review is only looking at the number two product, the Core i7-11700K which has lower clocks and lower power draw. The Core i9-11900K will likely need a 360- or 420-mm AIO just to not thermal throttle like mad. My 5950X with its meager 240mm AIO (Corsair H100i RGB Platinum) that runs at the quiet mode setting is laughing its butt off right about now. When Ian Cutress had to use an obnoxiously loud 170 CFM fan (I have used 100 CFM Deltas and those already annoy most PC enthusiasts) on a massive 4-pound, full copper heatsink to tame the 11700K's 290W, I shudder to think. Will the 11900K be outdoing the FX-9590's record-making peak power draw of 350W (see here: https://www.anandtech.com/show/8316/amds-5-ghz-tur... ? Ian easily could have gotten the 11900K also at retail, but I think he is holding back on that because he already knows the 11900K is going to be a throttling disaster and only the 11700K is an ACTUALLY USABLE PROCESSOR. *mike drop*Hifihedgehog - Saturday, March 6, 2021 - link
I am referring to 14nm mainstream consumer processors if that wasn't abundantly clear.Santoval - Saturday, March 6, 2021 - link
"To be clear, I was replying to you to get a reply at the top of the comments".You got me.. :)
Samus - Saturday, March 6, 2021 - link
In all fairness, it's actually amazing what Intel has achieved here with, essentially, a 6 year old manufacturing process!Spunjji - Monday, March 8, 2021 - link
Are we reading the same review? It looks like they *regressed* in performance vs their 5.5-year-old architecture on the same process.dihartnell - Thursday, March 11, 2021 - link
I think they did they realistically could given the contraints they have. We are not going to see intels true potential until they get thier manufacturing process fixed or they swallow the dead rat and go to another foundry. This will be enough to keep them in the game for another year.Hifihedgehog - Friday, March 5, 2021 - link
Summary image:https://i.redd.it/4i2eu882qbl61.png
tl;dr: Worse in games, moderately better in synthetics, slower and far more power hungry than Ryzen 5000
Bik - Saturday, March 6, 2021 - link
ThanksGondalf - Saturday, March 6, 2021 - link
Main defect of the article: AMD best of the best (low availability) versus a common high volume medium level Intel SKU.No matter the price that will change on retails or OEMS.
I don't understand AMD, they pump hard the pedal still they can not do much to gain market share. New processes are medium volume and with too much customers.
The NEW AMD will be a company capable to deliver a pile of good dies to all channels, without limitation on volume. Unfortunately Lisa follow the wrong street, in this manner Intel will always dominate the market.
Bet 2021 will be Lisa last year at AMD.
ottonis - Saturday, March 6, 2021 - link
How is Ryzen 5800x AMD's "best of the best"?It clearly is not but just about somewhere in the middle. If you want AMD's best consumer CPU, you gonna look at the 5950x.
From the perspective of AMD, completely outsourcing manufacturing was the only way to reap the benefits of latest and greatest process nodes.
Now, that everybody else, incl. the automobile industry, the consoles, Apple, and even Intel are booking production capacities at TMSC, has certainly contributed to Reaching capacity limits and thus to AMD CPU shortages.
It is predicted that shortages will be mostly (hopefully) sorted out by this summer.
But yes, now that AMD are earning quite a lot of money they should buy some TSMC stock and try to partner up, getting more production capacity in the future.
Fulljack - Saturday, March 6, 2021 - link
did you just forgetting why AMD spin-off it's fab on the first place?we don't know exactly how much AMD put out their desktop chips from TSMC plant, and as far as we know, the market are still growing in size. while AMD market share are still a long way from reaching 50%, they still have their processor sold out all over the place.
honestly I don't get what you're trying to say here.
shady28 - Saturday, March 6, 2021 - link
Actually we do know, they made ~1M Zen 3 chips in Q4. 140M PCs were shipped. Based on their market share, about 3% of the chips AMD shipped in Q4 were Zen 3.Source:https://wccftech.com/amd-shipped-nearly-1-million-...
inighthawki - Monday, March 8, 2021 - link
5800X = "Best of the best"11700K = "medium level SKU"
Your bias is showing.
Bluetooth - Saturday, March 6, 2021 - link
Does any one know the details of Intels 10 nm node problems. Any article discussing that in technical details?dihartnell - Thursday, March 11, 2021 - link
I think adored tv did a few articles, videos on this. As I understand it thier main issues are they use a monolithic die (everything on on a single die) that gets harder to make as the process shrinks, IE more dies have defects... AMD got around this by going chiplet, lots of small dies which meant more of them are good. Until Intel chnages to chiplet or they find a way to improve the manufacturing process to lower the defect rate then they will struggle.Santoval - Saturday, March 6, 2021 - link
Class action for what, excessive power bills? :) Imagine using this during a heatwave in an airless room. Since it doubles as a heater you would need to have the AC on all the time. If you have no AC there will be a competition between who dies first, you or the processor? ^.^sabot00 - Friday, March 5, 2021 - link
Indeed! I have been reading nothing but wccftech and Tom's leaks. Absolutely amazing surprise this Friday night while searching Rocket LakeBeaver M. - Saturday, March 6, 2021 - link
I would be very cautious testing or believing results with the Z590 platforms long before Rocket Lakes official release.Ive tested 3 of those boards with my Comet Lake (Asus, MSI and ASRock) and they all had pretty bad BIOS versions still, with PCIe/IO performance being low in SSD 4K benchmarks and getting weird frame time stutters from time to time (only noticeable when actually playing or looking at a realtime graph). Not to mention Intels drivers are still bad as well.
On none of them even very basic features like the sleep state worked!
Comparing this review with user benchmarks in German forums shows huge differences, so theres not much to add to this.
That said, I have to laugh when Americans or people from other countries with cheap power complain about the power draw.
And seeing fanboys downplay the performance of AVX and ignoring that it was always power hungry, even 6 years ago, is another obvious thing. Without it RKL actually runs pretty cool for being ancient 14nm.
And of course I love the geniuses who still think that you cant use the iGPU without it being connected to a monitor, or dont know about the new power saving/GPU switching feature. Not that this article didnt fail at pretty much everything, incl. explaining things like that.
Gigaplex - Saturday, March 6, 2021 - link
"That said, I have to laugh when Americans or people from other countries with cheap power complain about the power draw."Electricity is expensive where I live, but that's not why I want low power consumption. The more power it consumes, the louder the cooling solution will be. That's why my last system was Intel (Ivy Bridge) and my current system is AMD (Zen 3).
inighthawki - Monday, March 8, 2021 - link
It's not about the cost of electricity. High power draw typically translates to a lot of heat. My PC is on the upper floor and heat accumulates in my room and it gets extremely hot while gaming in the summer, even with AC on.As you stick to the same process node and continue to crank up the frequency, it gets hotter and hotter and hotter. Skylake didn't run even close to the temps that these new CPUs run at.
And yes, even Zen3 produces a lot of heat when under load.
YB1064 - Saturday, March 6, 2021 - link
How did you conclude that there exist thermal hotspots? Is it an educated guess or did you actually measure a temperature profile? If it was a measurement, was it a thermal image of the socket area of the rear PCB, multiple thermal probes? BTW, your argument does sound logical.ThereSheGoes - Wednesday, March 10, 2021 - link
Aaannddd. these results are obviously flawed. https://www.hardwareluxx.de/index.php/artikel/hard...Nameboy - Saturday, March 20, 2021 - link
Guys!! In this article, I am going to introduce the Yamaha YFZ450R ATV Complete guide Including:Yamaha YFZ450R Price
Yamaha YFZ450R Specs
Yamaha YFZ450R Top speed
Yamaha YFZ450R Horsepower
Yamaha YFZ450R Review
Yamaha YFZ450R 2021 Price
http://allatvprice.com/
blppt - Friday, March 5, 2021 - link
Wow, couldn't even match the 5800X. AMD really knocked it out of the park with the 5xxx series.FreckledTrout - Friday, March 5, 2021 - link
Fell short while using considerably more power.Azix - Friday, March 5, 2021 - link
I wouldn't call it knocking it out of the part if a 14nm chip is right behind them.DV8_MKD - Friday, March 5, 2021 - link
Yeah, "right behind them" with 20% more power, smhinighthawki - Friday, March 5, 2021 - link
TDP is not a measurement of power draw. The 11700k peak power usage is over 2x the 5800xlmcd - Friday, March 5, 2021 - link
Yea if it was only 20% behind in power draw, that'd be a win at this point.TheinsanegamerN - Friday, March 5, 2021 - link
That 14nm chip pulls over twice the power of the 7nm 16 core chip and is consistently slower then 7nm 8 core chip.It's not so much "right behind them" but rather "barely keeping up while burning through a nuclear reactor's power output".
CiccioB - Friday, March 5, 2021 - link
Twice the power for being slower?If you are referring to the 290W power consumption with AVX-512 test, I'm desolate to inform you that a 7nm CPU with twice the core would not reach those performances and perf/W in that test.
If you are talking about the 140-160W usage at other "normal" tests, I'm desolate to inform you that a 16 core 7nm CPU does not consumes 80W.
So stop vomiting meaningless numbers. This is a 14nm CPU and for the process it is build on it is doing miracles. If Intel could ever use an advanced PP like those 7nm by TSMC Zen would still be the underdog.
For the future just hope the TMSC 5nm are good, early, low cost and really high yielding, because if Intel comes with a decent 7nm I think AMD will not look all that advanced (6 years to surpass a 6 years old architecture and all by the use of a more advanced PP that unfortunately doesn't allow for great deliveries).
blppt - Friday, March 5, 2021 - link
"So stop vomiting meaningless numbers. This is a 14nm CPU and for the process it is build on it is doing miracles. If Intel could ever use an advanced PP like those 7nm by TSMC Zen would still be the underdog."Not necessarily. Intel apparently is still behind in IPC/single thread performance, as evidenced by that Cinebench results, so whilst 7nm would let it run with less power (theoretically), it still would lose to its main competitor, the 5800X.
CiccioB - Friday, March 5, 2021 - link
You have missed that with 14nm die area you cannot improve the architecture that much.You are still thinking that CPU designed for 7nm, with all the advantages that they would bring, would still be like Skylake which is a 6 years old architecture
A PP like TSMC 7nm would bring a completely new architecture that would blow Zen away.
Zen is good because it is based on such a better PP that those Intel has now, but it still struggles at beating Skylake. And to do that it, that is by using such an advanced but production limited PP, it has sacrificed high stock delivery right in the period where demand is much higher than supply.
Intel can fill the remaining market with whatever it has, being it 9xxx, 10xx or now 11xx generations.
barich - Friday, March 5, 2021 - link
Yes, Intel probably would beat AMD with an imaginary all-new architecture on TSMC's 7nm process. Similarly, I would be a hell of a basketball player if I were a foot taller and had any motor skills.Here in reality, Intel has worse performance and worse efficiency. As a consumer, that's what matters to me. What Intel could do with a bunch of "ifs" is irrelevant. I haven't owned an AMD CPU since my Athlon 64 was replaced by a Core 2 Duo. But there's no way my new build this year isn't going to be AMD.
CiccioB - Saturday, March 6, 2021 - link
Yes, what counts is the results, you are right.But by that I can't cry for a miracle when I see Zen 3 results as with a much more advanced PP it just can win over Skylake for a few % and all the real advantages it has is smaller power consumption due to the much better PP vs this one 6 years old.
If you look at the real power consumption, that is not the one with AVX-512 tests where RKL disintegrates Zen for perf/W despite the high power requirements, you'll see that this chip is not that power hungry (though being more power hungry than Zen) and that does make me think that with a better PP this same architecture would be another thing completely, as are the 10nm Tiger Lake which however suffer the not so good power consumption at higher frequency required by desktop SKUs.
As we are not that distant from finally having something decent that is not the 14nm PP, I will really not call my thought "imaginary". AMD will not be able to pass to 5nm so soon, and seen what this architecture can do, despote the 14nm PP, I think that the future is going to be more interesting that what you hope it ti be (that is, AMD keeps on figure it has better CPUs while not having them in the shelves but what counts for you are.. yes the results... and for these Intel is outselling AMD 5:1).
blppt - Saturday, March 6, 2021 - link
"A PP like TSMC 7nm would bring a completely new architecture that would blow Zen away."Based on what, exactly? We've seen die shrinks before without amazing architectural advances from both Intel and AMD. You have an awful lot of confidence in something that doesn't exist.
CiccioB - Saturday, March 6, 2021 - link
Based on the fact that with a 7nm PP AMD still struggles at beating Skylake architecture which is 6 years old and was born on... oh yes, 14nm.schujj07 - Saturday, March 6, 2021 - link
Zen2 was already faster than Skylake and its derivatives clock for clock by about 7%. While Comet Lake had higher single threaded performance than Zen2, it did so by throwing efficiency and power draw out the window and going for absolute performance. That made it such that Comet Lake could compete in ST applications but it still lost on MT applications against the same thread counted AMD CPUs. Going for absolute performance has been a double edged sword for Intel as the newer architectures hadn't been able to clock as high. Despite the higher IPCs of the newer architectures, absolute performance was no better than a wash due to 20% lower clock speeds.Zen3 now has absolute performance dominance over any Skylake architecture CPU. It doesn't "just" beat the older CPU, as in like 2% faster. It is upwards of 20% faster clock for clock and 10%+ faster in absolute ST performance.
blppt - Saturday, March 6, 2021 - link
"Based on the fact that with a 7nm PP AMD still struggles at beating Skylake architecture which is 6 years old and was born on... oh yes, 14nm."Struggles? The slightly older AMD chip (5800X) beats the newest and greatest out of Intel, whilst consuming less power, AND hitting lower peak turbo speeds.
That is complete domination. You could make the same argument about how Intel hadn't even made any significant gains over Sandy Bridge until Skylake, and that was *2* die shrinks.
Shorty_ - Monday, March 8, 2021 - link
I'm not sure if you're being wilfully obtuse or ignorant.. the only reason Skylake is even remotely in the game is that intel's 14nm is refined enough to allow them to push raw clock speeds to the moon. Do you not recall how awful Ice Lake was because it couldn't clock? TGL is starting to clock a bit better but it's still pretty damn close. This is on 10nm "superfin" which is ~= TSMC N7(P).So Intel don't have some magic engineering pixie dust that would propel them beyond AMD if they were on the same node.
Thesubtlesnake - Saturday, March 6, 2021 - link
Intel already have process equivalent to 7nm – 10nm SF. And they already designed a new architecture on it: Tiger Lake. And Zen 3 is perfectly competitive with Tiger Lake.Teckk - Friday, March 5, 2021 - link
Ultimately the latest desktop processors from Intel doesn’t perform well against AMD that’s what it is.They chose to release it on 14 nm as their 10nm was still work in progress. The numbers have meaning and not your conjecture about Intel using TSMC advanced node- it’ll be compared whenever that happens, with numbers.
Cooe - Saturday, March 6, 2021 - link
*Zen 3hfm - Saturday, March 6, 2021 - link
But we have to live in reality that they don't even have 10nm ready for desktop. Fantasies about creating an alternate reality where their core architecture exists on a smaller node for desktop are just that, fantasies. The reality is AMD clearly has the far better product right now aside from niche edge cases.I still agree with the conclusion though that given current circumstances, get what you can get if you need to upgrade or build new. But the reality there seems like the 5800X is available at MSRP in-stock at multiple storefronts.
blppt - Friday, March 5, 2021 - link
A chip that is just released, the best Intel currently has to offer for the mainstream consumer, can't match a chip that has been out for months. While using more power.Thats not a good look for Intel. I hope the 11900K (or whatever they're going to call it) at least matches the 5900X in games.
This is the first time in a long time, with generations of chips current, that I cannot think of a single reason to recommend Intel's latest and greatest over AMD.
terroradagio - Friday, March 5, 2021 - link
The 11900k has always been what should be compared to the 5900x anyway. Not the i7-11700k.blppt - Saturday, March 6, 2021 - link
The point being, the 11700k doesn't even catch the 5800X, which has been out for a few months already. Given that this was supposed to be Intel's "response to Zen 3", its pretty disappointing.Fulljack - Saturday, March 6, 2021 - link
the only thing that could save Rocket Lake-S are availability and price. otherwise just get Ryzen 5000 processors.SaturnusDK - Saturday, March 6, 2021 - link
Availability of Ryzen 5000 except 59xx parts is already a non-issue. You can get 5600X with a few days delay at worst, and 5800X is in abundant stock pretty much everywhere.The key is price, especially the platform price because Intel MBs are generally more expensive. On top of that you absolutely need a larger cooler, and most likely also need a beefier PSU for the Intel CPUs, so the CPU price for the intel parts have to be substantially lower than a performance equivalent AMD part to be competitive. And given the history of intel that seems very unlikely to happen.
TheinsanegamerN - Monday, March 8, 2021 - link
They're the same chip, the only difference is clock speeds. Dont get your hopes up, RKL is a total dud, much like Williamette was.Samus - Saturday, March 6, 2021 - link
blppt - my concern is that AMD may have a superior IPC, but the real fruit comes from the manufacturing process. Intel is still (somewhat) competitive at 14nm and that in itself is quite unbelievable. Imagine where this chip would be on 7nm or 10nm, at 6GHz+ and more cores with 2-3x the cache.That said, this victory may be short lived because AMD is basically taking advantage of the embarrassing execution Intel has repeated, much like they did 20 years ago with the P4 (albeit that was an architecture failure, not a manufacturing process failure)
Thesubtlesnake - Saturday, March 6, 2021 - link
Intel's latest 10nm process delivers *slower* clocks than its 14nm one. So, no, 6 GHz is not on the table. I imagine that when the transition to 7nm, Intel will be able to achieve moderately faster clock speeds than with 14nm.Otritus - Sunday, March 14, 2021 - link
10nm SF is good enough for 5 GHz. 10nm ESF can clock higher, so Intel's latest (but unreleased) process should match 14nm. I would not expect 7nm to clock higher than 14nm because it is becoming very clear that 5Ghz+ is just a waste of power and transistors, so i would not expect 7nm architectures to be designed to clock higher. We either are getting lots of IPC or just over 5GHz.Slash3 - Friday, March 5, 2021 - link
Mad lad.edved - Friday, March 5, 2021 - link
Nice write-up. Thank you.lucasdclopes - Friday, March 5, 2021 - link
Power efficiency is abysmal on this one.CiccioB - Friday, March 5, 2021 - link
No, it is not. It lower than AMD's efficiency, but it not that bad for being based on such an old process.PixyMisa - Friday, March 5, 2021 - link
So it's abysmal, but that's only to be expected?Spunjji - Saturday, March 6, 2021 - link
Not bad for an old process is still abysmal by the standards of 2021. No wonder Apple dropped them like a hot rock.lmcd - Friday, March 5, 2021 - link
As an Intel fan, this is abysmal. This is literally only good for an i3 or i5 SKU, for its upgraded iGPU. That's it.Slash3 - Friday, March 5, 2021 - link
The 11th gen i3 SKUs will not feature the new Xe core - they are still based on Comet Lake.lmcd - Friday, March 5, 2021 - link
That makes this even sadder. This port was actually worthless. An i5 quad core might be manageable by a cooler that fits in an average case, but that's about it.Would've been better off releasing Tiger Lake 35W processors on an LGA package.
Slash3 - Friday, March 5, 2021 - link
The 6C/12T i5-11500 should be "fine," as it has the same 32CU Xe iGPU as the higher end parts. The part below it, the i5-11400, is also 6C/12T but has a cut down 24CU Xe core.The Xe upgrade is a nice change, at about a 33% uplift over the previous "Gen11" iGPU, but it's still just what I'd call "passable" for light gaming. Anything above 1080p and you'll want a discrete GPU for the best experience, and that goes for both teams.
lmcd - Friday, March 5, 2021 - link
I understand what you're saying but I still think the 6-core i5 is going to draw a boatload of power.Obviously for gaming a dGPU is preferable. In my experience with AMD's Ryzen 2400G, though, it isn't actually too stable with medium resolution monitor configurations (think 1080p + 1440p) and isn't all that well supported by drivers.
On the Intel side, uplift over Gen11 is cool but uplift over Gen9 is where it gets noticeable, and important. Gen9 was fine when it came out but that was quite some time ago. Shows its age when too many apps want GPU acceleration with that same multi-monitor setup I described.
lmcd - Friday, March 5, 2021 - link
Sorry, didn't complete my thought. Anyway, point is that a decent % of Intel's sales are for business desktops (maybe not right now, but, ya know, offices might open up sometime this calendar year, before Alder Lake ships in volume). Rocket Lake i5 would be perfect, if its real-world power consumption wasn't out the roof. Good enough graphics for multi-monitor, fine performance elsewhere.I guess that makes me wonder what performance would be if limited to a 65W avg/95W peak thermal output. That's about what those small Dell towers can handle.
Slash3 - Friday, March 5, 2021 - link
Yep, I'm with you.I honestly don't expect the i5 models to have the same outlandish power usage characteristics as the reviewed i7, but will definitely be reading any day one reviews that may pop up. The existing i5-10400 is a pretty compelling product (and priced well), and if the 11400 or 11500 can manage to fit into the same ~100W PL2 envelope I think it'll find a home in a lot of desktops. For OEMs, they're sometimes board limited to the 65W PL1 via a BIOS option, and I'd expect that to continue with the 11th gen versions.
If they price the 11400 close to the 10400, it'll be a solid choice.
Makaveli - Friday, March 5, 2021 - link
The Turbo for the 5800X should be 4850 that is what I see at stock. So the table is off by 50mhzSaturnusDK - Saturday, March 6, 2021 - link
Not really. AMD changed the way they report boost frequency numbers. Before it was "up to xxxxMHz boost", now it's the more wordy "given adequate cooling the boost frequency is _at least_ xxxxMHz". This change was driven by the stick they got for the 3000-series only very rarely hitting the listed boost frequency. Now you can generally assume to get 50-100MHz _above_ the listed boost frequency if you have a half-decent cooler.Marlin1975 - Friday, March 5, 2021 - link
WTF!!! 291 watts?!?!?Intel REALLY needs to step up is manufacture game. Designs seem ok but the nodes are killing them.
nandnandnand - Friday, March 5, 2021 - link
Alder Lake on 10nm will fix everything, and be out before the end of the year.kgardas - Friday, March 5, 2021 - link
The power consumption you comment shows while using AVX512 on hand optimized test. Your idea about Alder Lake to solve this is not the correct one as Alder Lake itself will not implement AVX512 ISA at all. IMHO very bad decision by Intel again.TheinsanegamerN - Friday, March 5, 2021 - link
Yeah, some weak ant sized atom cores are what intel needs to fix this problem LOLnandnandnand - Friday, March 5, 2021 - link
You don't need more than 8 big cores for gaming. With a real IPC improvement, 8+8 should be able to beat the 5900X.lmcd - Friday, March 5, 2021 - link
Not likely, but it'll at least beat the 5800X and probably go even on efficiency. The real upside is in the server space and laptop space. I expect Alder Lake to do excellently in both of those segments.nandnandnand - Friday, March 5, 2021 - link
Alder Lake's Golden Cove cores should have a decent IPC improvement over Rocket Lake, so 8 of those cores should be able to match more than 8 Zen 3 cores. Then throw in the 8 Gracemont Atom cores which will be better than Tremont. 8+8 should top 5900X but not 5950X in multi-threaded, and beat Zen 3 in gaming.There's caveats, perhaps related to DDR5 or schedulers, but I will be surprised if the top Alder Lake chip can't beat the 5900X.
DigitalFreak - Friday, March 5, 2021 - link
Sorry, but in this case 8 + 8 does not equal 16.nandnandnand - Friday, March 5, 2021 - link
5900X is 12 cores, not 16. That's what Alder Lake 8+8 has a chance of beating.SaturnusDK - Saturday, March 6, 2021 - link
The problem is that Alder Lake has been pushed to second half of 2021 at the earliest so it will not be competing against Zen3 but Zen4.Pneumothorax - Saturday, March 6, 2021 - link
It might be able to beat the 5900x, but by the time you add in Intel's overpriced motherboards (have you looked at Z590's recently?!) and the premium of DDR5, you're going to be at 5950x+ pricing.dihartnell - Thursday, March 11, 2021 - link
Traditionally Big.Little designs don't work that way. They either are running the big 8 cores or they are running the little eight cores but not at the same time. The type of workload determines which is run when. Personally don't think it makes a lot of sense In desktop.Jasonovich - Friday, March 12, 2021 - link
And whats the point of the little chip big chip design, when TMSC will very shortly produce in mass the 5nm and in 2023 the production will move to 3nm.Alderlake is based on the worse case scenario and has been introduced to buy Intel time until it resolves the shortfalls of their 10nm production.
lmcd - Friday, March 5, 2021 - link
You're really confused if you think Atom doesn't help Intel here. Tremont performance per watt and performance per die area is really quite excellent. Also worth remembering that nearly every Atom you've ever seen has had mediocre memory and cooling paired with it. I don't expect Intel to "win" off of this move but it'll help for as long as Intel doesn't have chiplet ready.The_Assimilator - Saturday, March 6, 2021 - link
Even if it does - and Intel's current record on 10nm suggests it won't - by that time AMD will have had over a year of Zen 3 reining unopposed, and Zen 4 well on the way.The_Assimilator - Saturday, March 6, 2021 - link
*reigningm53 - Friday, March 5, 2021 - link
291W is for AVX512 workload. The rest of the CPUs here won't match it's performance on AVX512 workloads no matter how much power you give them.But if you are not interested in AVX512 workloads then don't look at the AVX512 power consumption.
scineram - Saturday, March 6, 2021 - link
Already beaten in DigiCortex.RaistlinZ - Friday, March 5, 2021 - link
Ouch. :(terroradagio - Friday, March 5, 2021 - link
It is incredible bad form and bad taste to release a review before anyone else and before Intel has provided the new microcode update to resolve the early issues. All because Anandtech wants to get out early.And your defense by way of saying well we got it at retail and therefore this doesn't matter is a joke. Terrible publication.
Makaveli - Friday, March 5, 2021 - link
lol that power consumption really bothering you ?terroradagio - Friday, March 5, 2021 - link
I'd be saying this if they did with an AMD part too. This is something I would expect from someone unknown guy doing reviews, not Anandtech. Extremely disappointing they are going for the click bait.Otritus - Saturday, March 6, 2021 - link
We used to have performance previews all the time back in the day to let people have an idea of what is coming and if it is worth it to wait for the next generation. Consider this to be a performance preview, meaning final performance will be higher. In general workloads rocket lake is 10-15 % faster while being reasonable with power. In gaming it's trash, but that could be a consequence of the non-finalized nature of the product.By the time someone can reasonably buy this product it will be April/May. If I need a pc now, the performance uplift isn't that high to make me have buyers remorse. If I can wait, I can get a comet lake chip at probably a great discount, or if rocket lake's gaming is fixed, get that instead.
Performance previews like this are excellent journalism because the public is informed of things that are otherwise concealed. My only gripe with the article is calling it a "review" instead of "early review" or "preview" because someone may stumble on this article and mistakenly believe this is how the product will be shipped.
sonny73n - Saturday, March 6, 2021 - link
This is a REVIEW! Not early review or preview or any excuse you're making. You've lost touch with reality if you hope for a higher performance on official release date. If Intel haven't released these CPUs (before their announced release date), no way in hell Anandtech or anyone would be able to purchase them.Spunjji - Saturday, March 6, 2021 - link
I'm amused by how many people have stumbled upon this and are mistakenly assuming that performance will be significantly better when it's shipped. This was a CPU bought at retail running on a retail motherboard. If they do noticeably improve performance (and it's a big if) consumers will have to install updates themselves to get it.Makste - Saturday, March 6, 2021 - link
This is not an engineering sample. If hundreds were being sold by a retailer then that is what Intel was intending to sell to its clients. If you are not satisfied with the performance don't blame it on Anandtechshady28 - Saturday, March 6, 2021 - link
This is labelled review, not performance preview.Cutress is officially a hack now. I'll wait for real reviews on systems that officially and properly support the chip.
schujj07 - Saturday, March 6, 2021 - link
It was a review. This was a retail CPU running on a motherboard with a BIOS that supports RKL. Just because RKL doesn't look good isn't surprising.shady28 - Sunday, March 7, 2021 - link
On a ++BETA++ BIOS.schujj07 - Monday, March 8, 2021 - link
Unless you have something like the X570 and Zen2 being released at the same time, quite often it is a Beta BIOS that gives initial support for new chips. Eventually with further validation that Beta BIOS becomes official. Perfect example of this is in October 2019 I was installing vSphere 6.7 onto some servers. With the initial BIOS release I was getting the Purple Screen of Death. The only BIOS that was available was a Beta BIOS and on that vSphere installed just fine. Push forward a month later and the official supported BIOS by VMware for that server was the Beta BIOS for vSphere 6.7. It took another 6 months before a non-beta BIOS was available for that server as well. Hence just because it is Beta doesn't mean it is bad. Things in Beta testing generally have full features and performance just hasn't been validated long enough to become the official release.Qasar - Sunday, March 7, 2021 - link
like that will change much, if anything shady28.Makste - Saturday, March 6, 2021 - link
Well.. if it was for the click bait, hre we are 😁half_mexican - Friday, March 5, 2021 - link
This is from anandtech Dr. Ian Cutress"Latest beta BIOS from vendor, was told that they don't know when the next BIOS update will be and this contained everything to date. So unless you've got special information.
Note that this is always the risk of doing reviews even on launch day. At some point you have to lock in a BIOS version for published results. Vendors who send BIOSes 24 hours before embargo lift are told to go away."
terroradagio - Friday, March 5, 2021 - link
It has been well discussed that there is a forthcoming update coming. That is why the release and NDA is the way it is.Spunjji - Saturday, March 6, 2021 - link
I'm sure it will make a huuuuge difference. Especially for all the poor sods who bought Z490 in anticipation of a compatible upgrade. /sOtritus - Saturday, March 6, 2021 - link
I mean the leaks suggest that z590 motherboards having some problems resulting in performance regressions. So some poor sod who bought into z490 got to enjoy a fast cpu and can upgrade to an even faster one. AMD is obviously the best for non-avx-512 workloads, but where I am I can't find one for a reasonable price, so Intel is the only viable option. Perhaps the real travesty here is the lack of capacity in TSMC's 7nm node preventing us from buying excellent cpus and gpus at reasonable prices.Qasar - Saturday, March 6, 2021 - link
the question is, how many things actually use these special avx instructions ?? a handful ? unless you know you can use it, no point in them. seems intel creates these, just so it can win a benchmark.IanCutress - Friday, March 5, 2021 - link
People forget that Anand posted our Sandy Bridge review several months early.terroradagio - Friday, March 5, 2021 - link
So that makes this right? If I were Intel, I'd be revoking your early samples.Ryan Smith - Friday, March 5, 2021 - link
Yes, this makes this right.AnandTech has always honored NDAs, and continues to honor this one. We adhere to the requirements of every agreement we sign, even when doing so is not in our best financial interests. We do this because we're honest people, and just as pragmatically, we need hardware vendors to be able to trust us.
The flip side to that, however, is that retail hardware always has (and always will be) fair game. This was a processor sold by a major European retailer, tested in a motherboard based on a chipset that has been selling at retail for the past couple of months.
Although Intel may not be happy with that retailer over their lapse, at the end of the day this is final silicon running on final silicon. We have done every bit of due diligence both to ensure the accuracy of our results, and to inform the necessary parties in advance about what we intend to do, in case they wish to raise any issues with us.
So we stand by this review both from a technical perspective and an ethical perspective. All of this material was handled in a fair manner that was entirely above the board and legal in all steps of the process.
terroradagio - Friday, March 5, 2021 - link
Using a processor that isn't suppose to be sold is sketchy, plain and simple. And you know full well this. It is not to the benefit of anyone who may be interested in this part. You are thinking about yourselves, full tilt. And that is your choice. You have no idea what could happen with this update they are talking about. Will it be magic? Probably not. But will it have fixes for other things in your review? Perhaps. So therefore you are putting out an inaccurate piece for the purpose of getting out early that may very well be inaccurate in parts at the end of the month. And quite clearly from the comments here, feeding many hungry AMD fanboys.I'm not mad at you, just disappointed. Enjoy the attention.
Makaveli - Friday, March 5, 2021 - link
lol the only one looking like a fanboy is you.DigitalFreak - Friday, March 5, 2021 - link
Why so butthurt? You wouldn't be on here whining if the benchmarks were in Intel's favor. Never understood the fanboi mentality.arashi - Friday, March 5, 2021 - link
Because now his website won't get the views he thought they'd get.Spunjji - Saturday, March 6, 2021 - link
Nice projection there.MarcusMo - Sunday, March 7, 2021 - link
”Using a processor that isn't suppose to be sold is sketchy”The operative word you’re missing is “yet”. The batch of processors this one is from is Intel release silicon intended for end customer hands. And there are more out there. This article represents exactly the kind of performance that at least some day one customers should expect. Now maybe there will be updates to the software stack that will improve performance over time, but that doesn’t make this review any less valid.
Slash3 - Friday, March 5, 2021 - link
Appreciate the statement, Ryan.Pre-release content is nothing new for AnandTech, and it's interesting to see how passionately some people feel about the topic. Might be something worth exploring in a future article, as I'd wager that there are a lot of readers who weren't around to see things like the original Sandy Bridge pre-review which Ian referenced in another reply.
CiccioB - Friday, March 5, 2021 - link
I think the criticism on this review is quite justified.You are reviewing a device that is not ready to be sold yet.
Yes, you could buy it at retail but just because, like you, some other "smart guy" made the wrong (would say fraudolent) move to not respect a date.
What we have here is a "preview" of the CPU performances that may (or not, but you don't know now) change when the CPU will really available for the rest of the mortals on the globe.
I would like to think that you will do a new review of the CPU once the motherboards will be updated and make evidence if, how and by how much something has changed since this preview with what are early samples that results being compatible with the device.
However, while you were at it, you could also try PCI4 connected memory storage to see how good Intel implementation of the technology is.
MFinn3333 - Friday, March 5, 2021 - link
"You are reviewing a device that is not ready to be sold yet." How is it Ian or Ryan's fault that Intel released a CPU in this condition? If Intel isn't doing any QA on their end for proper use out of the box then the problem is on their end.And yes, they did release it in that condition because unless they intend to recall all the packages and open them up and replace the CPU inside, it is the product that people are going to be getting and using when they open it up and slapping it in their computer. I used to work retail and getting new product on major releases is usually about 3-4 weeks ahead.
Following your logic then they shouldn't do a review unless they are willing to also update all of the AMD CPU's as well to include their performance and bug fixes which would turn bench-marking into a never ending nightmare because of updates.
Intel whiffed this release, get over it.
CiccioB - Saturday, March 6, 2021 - link
You have misses that Intel has not released the product yet and the samples you can find on the market come from a seller that broke the NDA and start selling them before the official release time.They are actually are in beta support with BIOS, microcode, drivers and such on not final motherboards as well.
Once you understand that you'll understand why this is a preview or a beta test, not a full review of the product.
Qasar - Saturday, March 6, 2021 - link
um THE HAVE RELEASED THEM. the store just started selling them early, that is the stores fault. you are grasping at straws.at_clucks - Monday, March 15, 2021 - link
Oh CiccioB, give it a break. That exact CPU was packaged by Intel for retail, it was meant to be sold as is, just a few weeks from now. Yet, that *exact* piece of hardware. You keep implying "it was not ready for retail" as if Intel was gonna start etching this CPU some more to turn it from a hot grill into a cool as a cucumber lightning fast CPU. Intel may tweak it a bit and have a new revision but it's not like they're just getting the line moving now, they've been building stock for a while so retailers are ready to sell this silicon.But let's be honest, this very CPU that AT put to the test would have been something that an end user would have bought days or weeks from now. A real customer would have used it as is.
The only straw you could grab is that the BIOS might be tweaked until launch. And while it's true, it also probably doesn't work in your favor. I doubt retail MoBos will have CPUs running at over 100C so likely they will limit power more aggressively, and there's very little microcode optimization that one can do to squeeze that much performance.
This is a retail CPU that was sold some days or weeks early. But still retail, and still exactly what consumers will get. Performance won't get better. Power will but only at the price of performance. Stop shilling, unlike many other articles, this one actually made it clear that the BIOS was not final. Any person who doesn't understand that the BIOS can't magically fix this CPU probably don't bother with reading AT anyway.
Makste - Saturday, March 6, 2021 - link
Thank you very much for this comment. My thoughts exactly.To help change some people's perspectives regarding this review. Look at it as a review for those who bought those RL cpus who are about to buy them from that retailer. Otherwise, for the rest who can't handle such an early review, just look away and wait for the launch date and the reviews which will commence.
Billy Tallis - Friday, March 5, 2021 - link
" you could also try PCI4 connected memory storage to see how good Intel implementation of the technology is."That's odd phrasing. Do you mean something other than off the shelf PCIe 4.0 NVMe SSDs? We do plan to test storage performance.
CiccioB - Saturday, March 6, 2021 - link
I mean testing a whatever PCI4 SSD to see if they are working correctly.My suspects is that the BIOS used on that motherboard was so early that there was not PCI4 support at all and that's the reason there are not those test, which would have been a normal thing to add seen PCI4 is one of the new feature brought by these new CPUs, newer than the AVX-512 instruction set.
terroradagio - Friday, March 5, 2021 - link
Why would they show PCI4 when its quite clear this review was not aimed to show any benefits - just to feed the AMD fanboys.Billy Tallis - Saturday, March 6, 2021 - link
PCIe Gen4 support is certainly a welcome improvement over Intel's previous desktop processors, but it can't be considered much of a benefit over AMD's alternatives. We will be investigating whether there are any measurable differences in PCIe 4 storage performance between Intel and AMD hosts, but given how limited the benefits of PCIe Gen4 over Gen3 are for NVMe storage, it's pretty clear that differences between PCIe Gen4 hosts will be insignificant.Spunjji - Saturday, March 6, 2021 - link
@Billy - they're demanding you validate Ryan Shrout's claimed benefits of Rocket Lake over Zen 3 in a synthetic storage benchmark. Who could possibly imagine why 🧐schujj07 - Friday, March 5, 2021 - link
I don't see how a BIOS update will do much for performance. The motherboard is already running with unlimited turbo so the CPU is pegged at 4.6GHz for MT tests. That is the rated all core turbo frequency.haukionkannel - Saturday, March 6, 2021 - link
Ofcourse bios updates will improve. Just like They have improve amd performance... but not buy much. 1% I prove is a big in these changes... bios upgrades Are more to clean up bugs.chrcoluk - Saturday, March 6, 2021 - link
He reviewed on available now retail parts, even if a microcode update iss issued, there will still be people using this cpu on same microcode as in this review, as its down to the user to manually update it.I have always hated specially coordinated review programs where everyone agrees to publish shame time (wtf?), and the reviewers are working with vendor to make sure review doesnt upset them, wild west reviews like this need to be more frequent.
Timoo - Monday, March 8, 2021 - link
"You are reviewing a device that is not ready to be sold yet."Sorry?
It came from a European retailer. Therefore it is ready to be sold.
"Not officially launched yet" would be more accurate.
ballsystemlord - Friday, March 5, 2021 - link
Nice one @Ryan . Keep up the good work.Spunjji - Saturday, March 6, 2021 - link
Well said, Ryan.pentiuman - Sunday, March 7, 2021 - link
I understand AnandTech has honored their applicable NDA. And that you informed Intel of yourintentions or whatever. And that you didn't break any laws. And you also considered the early release OK because the chip was (what would be sold here in the US) retail. But I think I agree w/ User terroradagio and some others in that, Anandtech shouldn't have released their review early because they happened upon a favorable, early deal - (which itself may have been contrary to an Intel company policy w/ the retailer), not available to any other reviewer or consumer. It's taking advantage of a slip in how the system was supposed to work. You don't want to see it as wrong because it's almost like time doesn't really matter. In the end, you're still buying the product, doing the work, publishing and maintaining the website and revisiting the numbers and updating the motherboard and more and more work. You do all this hard work, and you're highly respected, (and for good reason), so for these good reasons, and more, I think this clouds your decision on this matter. I just feel that all tech sites should respect the same release review date! To not do so reminds me of the less ethical journalism methods used by some photographers, who then sell them legally to the newspapers. But integrity goes deep - more than 1 level.
The benefit Anandtech COULD take is the one that they have apparently become so used to, that it is assumed. The ability to buy the chip before consumers, test it, write their review, and click
the mouse to post it 1 second after the NDA says they could. (My point here is, some reviewers are either not able to buy them early, or not given the chips, don't have the connections to buy them, and have to wait to buy them like any other consumer, to test them, review them and then publish.)
In other words, you are already at an advantage over some reviewers by your early access to the
chip - and have weeks more than them to test it. Publishing it 3 weeks earlier than your standard
NDA (that may not apply), before nearly anyone else, is (in my opinion) an unfair advantage. You
are a well established website and reviewer - so I'm not saying you did it for the views. I just
feel it's not right. I get it - you must have a different ethical view. Thank you for the review otherwise.
Qasar - Sunday, March 7, 2021 - link
pentiuman, and would you be saying the same thing if another site did this, or were also able to get one of these cpus to test ? or maybe, like others have suggested, some just dont like to see intel in such a disappointing light ?Timoo - Monday, March 8, 2021 - link
Why has Anandtech here an advantage?If you wish to write an early review, you can too.
The CPU is simply for sale, apparently.
https://www.techradar.com/news/intel-core-i7-11700...
Timoo - Monday, March 8, 2021 - link
*update: not anymore, apparently.Sold out or rebuked by Intel?
Spunjji - Monday, March 8, 2021 - link
Every other reviewer out there had the chance to take advantage of this "slip", so it's not unethical.Unethical would be taking advantage of insider contacts to produce an officially-sanctioned "preview" prior to release of a product and formal reviews that provides a misleading picture of the product's performance, like DF did with Nvidia and the RTX 3080.
lmcd - Friday, March 5, 2021 - link
Any coverage of Rocket Lake is good coverage at this point. "People know it exists so hopefully they'll buy it."movax2 - Friday, March 5, 2021 - link
I really don't understand why You attack anadtech. Intel Rocket Lake sucks... not anadtech!!Get it right already!
CiccioB - Friday, March 5, 2021 - link
I would like to know what you will say if Anantech will do the same with AMD Ryzen 4, that is reviewing it on an early motherboard with a beta BIOS and not yet tuned microcode and it will result not being as fast as you would expect (or hoped it to be) one month head of the actual release date.And present it as an official review of the product.
I would bet you (and your "friends") would go and cry out for a payed article by Intel to make AMD product look worse that it really is "like the good old times when it payed everywhere on Earth to not make AMD sell its products".
supdawgwtfd - Friday, March 5, 2021 - link
Except it appears the CPU is operating at advertised performance levels...It provides +19% improved performance for some things but not others.
the_eraser1 - Friday, March 5, 2021 - link
It would actually be fine because Zen 4 is already showing 30% performance uplift clock-for-clock, and that's a full year ahead of launch.Be honest with yourself. Achieving even a 5% uplift with microcode optimizations ahead of launch is a pipe dream. The review successfully shows the kind of performance you would expect. Is there room for improvement over time? Of course, but that applies to any product.
CiccioB - Friday, March 5, 2021 - link
I'm not speaking about the improvements in IPC. IPC is not everything to evaluate a product. If that were true ARM chip would be the winner since early '90s.I'm speaking about the fact that in many tests shwn here this architecture shows worse results than the previous one. That would mean there's something really broken in the architecture or in the SW they execute.
About Zen 4, don't old your breath because 5nm for HP are not that close. Even Intel Ocean Cove is said to be the really new revolutionary architecture that is finally going to show what the new 7nm PP could have really brought if it was available today as we speak.
These are speculations, while this chip will be out in less than a month and waiting for the final tuning would have just made a better service to anyone that really wants to know how really it behaves. Not how it looks like on a unknown motherboard with a not updated BIOS and not the final (or even the first version) of the microcode.
the_eraser1 - Friday, March 5, 2021 - link
This performance from RKL is unsurprising if you had been paying attention. We've known for months that Intel hasn't had great results with RKL and that's why they're pushing for Alder Lake ASAP.Once again, you cannot honestly expect substantial performance uplifts a month before launch. It's possible the ring/uncore frequency was low for this review, however that will only make a significant difference in games, or other latency sensitive scenarios.
As for Zen 4, I know for fact from reputable sources that it's coming around the middle of 2022, with large uplifts in performance.
Qasar - Saturday, March 6, 2021 - link
" I'm speaking about the fact that in many tests shwn here this architecture shows worse results than the previous one. That would mean there's something really broken in the architecture or in the SW they execute. " no your crying cause intel still lost. and that well. rocket lake, isnt the performer it was made out to be. AND it looks like some of the performance regressions, were explained/accounted for, in the review, which you obviously did not readOtritus - Saturday, March 6, 2021 - link
Originally ocean cove was leaked to be a revolutionary architecture with a massive ipc uplift that can serve as the backbone for future architectures like Conroe did in the original Core2 line. Leaks have later said that Intel cancelled the revolutionary nature of the product and ocean cove is simply going to be another microarchitecture like sunny or golden cove. By that point AMD should have zen 5 and we the consumer can enjoy healthy competition.As to the point on how it really behaves. We can see from the frequency graphs that rocket lake is turbo boosting as it should. A new BIOS could change power limits which would change behavior, but when rocket lake is already given infinite turbo time, such changes are likely to lead to performance regressions. The only other possibility is that the slight maturation of the bios leads to a small performance uplift. This WOULD BE IPC, but such improvement would be small at best given that z590 (and the similar sunny cove) has been out for a while. At best gaming performance may not suck as hard (or be fixed), but overall the performance improvements should be less than 5%. Not enough to change any conclusion.
My only gripe is the usage of "review" over something that indicates it's a pre-release product. However, given that Intel themselves didn't have any comments on the article, the final performance is likely to be so similar that this is basically the review of the final-release product.
Spunjji - Saturday, March 6, 2021 - link
They've already done that with a bunch of AMD products - like the OEM-only 4000 series APU and the weird Xbox One S APU desktop board.Speaking as a tech enthusiast, if they get hold of Zen 4 before release and can do a preview that doesn't break NDA, I would be over the moon. I love to get an idea of how a release will shape up, as long as there are caveats that it may not be final performance - which is exactly what we got here.
Timoo - Monday, March 8, 2021 - link
I'd Lóve it!Beta BIOS might be a slight disadvantage, but we've seen it with the release of ZEN1. At the time everyone blamed the BIOS for memory compatibility, etc. etc. etc.
In the end, not much improvement was found, once stable BIOSes were out. Bugs were fixed, but ZEN1 was still not beating Intel.
TheinsanegamerN - Monday, March 8, 2021 - link
It would still match the expected out of box performance. Stop being so salty over Intel sucking the big one, RKL is a total dud performance wise, a microcode tweak is not going to increase IPC. It already boosts to where it should be and draws LMFAO power.zakelwe - Saturday, March 6, 2021 - link
terroradagio, your argument would have had more weight and be less hypocritical if you had not read the article that was posted. You lost the moral high ground when you did.brunosalezze - Friday, March 5, 2021 - link
If you dont like it dont read it. Just wait for the NDA lift and be happy. I'm still waiting for the STH Milan early reviewterroradagio - Friday, March 5, 2021 - link
I will wait, because I will read the reviews that waited like everyone else and for the microcode update that very well may fix some of the issues. Anandtech has just given the middle finger to a bunch of other sites and channels who are doing the right thing and waiting to see what happens.TheinsanegamerN - Friday, March 5, 2021 - link
Yeah a microcode update is going to magically fix 290W power draw. LMFAOterroradagio - Friday, March 5, 2021 - link
I didn't say that. I don't know what it will do. And neither do you. And that is why you wait to see before the official launch. Using a CPU that wasn't supposed to be sold is a total backdoor and deceptive way of handling this. Totally unprofessional.lmcd - Friday, March 5, 2021 - link
Dude this isn't a new CPU core or a new GPU core. They've both already been released. There's no magic microcode fix. There is only an upcoming die shrink for desktop. That is all.terroradagio - Friday, March 5, 2021 - link
It is entirely possible a new update could fix issues. For example, in this review, the latency issues they were seeing. Look at AMD. AMD has put things out many times that have later been fixed with updates.schujj07 - Saturday, March 6, 2021 - link
Those updates allowed for better turbo or memory frequency. Microcode updates won't fix cache latency issues if the physical SRAM is already slower. Same as it won't fix intercore latency as that deals with the mesh fabric.chrcoluk - Saturday, March 6, 2021 - link
microcode updates are extremely rarely anything to do with performance, they are usually to fix erattas. Since there was no instability in this review there is nothing for a microcode update to fix.If you think a microcode update is going to give any kind of performance boost or power usage drop you going to be disappointed.
Qasar - Saturday, March 6, 2021 - link
terroradagio please explain HOW this cpu wasnt supposed to be sold ? it was bought AT RETAIL.Fulljack - Saturday, March 6, 2021 - link
I guess to his understanding either this CPU are engineering sample or/and that using beta BIOS, microcode, and firmware so this is not "full retail".romrunning - Saturday, March 6, 2021 - link
@terroradagio - "Using a CPU that wasn't supposed to be sold is a total backdoor and deceptive way of handling this. "You're really not making sense. They bought a retail boxed CPU from a retailer; they didn't test an engineering sample. The same box would have ended up in someone's hands when purchased by them, and the retailer's stock of RKL CPUs will also be sold to customers. To say Anandtech used "a CPU that wasn't supposed to be sold" is a complete lie. A retailer sells products; they don't stock products just for themselves.
If you're going to argue something, at least argue something with more validity.
CiccioB - Friday, March 5, 2021 - link
Not that I want to defend Intel, but to you and all others that have difficulties at understanding numbers (you just look at them and report the value without thinking about their meaning):The 290W power consumption was achieved by using AVX-512 instructions. In the test that uses them Intel is <b>6 times</b> faster at double the power consumption than AMD CPU.
So under the point of perf/W with AVX-512 compatible workload (responsible for those high power consumption) there no doubts that Intel is the winner with a very large margin.
So better to concentrate on other power consumption terms to make this chip appear the fail it actually is with respect to the new architecture that does not really improve on almost anything.
lmcd - Friday, March 5, 2021 - link
Frankly I think some of us (at least, I) got confused by the revision of CPU architecture going into this design. Tiger Lake's Cove version might've changed the outcome, even if the timeline didn't work. Maybe would've been way too big though.CiccioB - Friday, March 5, 2021 - link
I'm not a CPU architectural engineer so I just can guess that this compromise of architecture and process is the best Intel could come up also taking into account production limitations.Probably it is not the best thing Intel could have done in absolute terms, but it for sure will be available at big quantities (unlike "super fast advance mega efficient hyper many core" AMD CPUs, and worse APUs) and that could be enough to fill the market until Alder lake is presented.
You know, this generation is just a fill gap as it is not a 10nm generation and can't get all those Cove's improvements on the now obsolete (but still high performing and delivering) 14nm process.
Cooe - Saturday, March 6, 2021 - link
Zen 3 in the form of the directly competing R7 5800X is widely available at MSRP. How many times must you be told this before it gets into your thick ass skull??? O_oSpunjji - Saturday, March 6, 2021 - link
The complaint here is that 290W exceeds the capabilities of the vast majority of air coolers. The Perf/W you get is irrelevant if you melt your CPU.In directly comparable tests where the same work.is being done, Intel's Perf/W is still roughly 50% worse than the 5800X.
TheinsanegamerN - Friday, March 5, 2021 - link
Cant handle Intel dumping arse all over the bed eh?terroradagio - Friday, March 5, 2021 - link
Why are you guys making this an Intel vs AMD thing, when I am not? Get a life. This is about professionalism. Go take your AMD worshipping and Intel hating elsewhere.Makaveli - Friday, March 5, 2021 - link
And who are you to question Anandtech professionalism you sound like a butt hurt fan boy. Quit the crying and just move on.arashi - Friday, March 5, 2021 - link
He's probably just a salty competing review website owner upset that he didn't think to do this first.jonathan1683 - Friday, March 5, 2021 - link
I am an intel fanboy, but I dont agree with you at all either, intel scewed up and they have to deal with that, they wont fix anything the motherboards are made the cpus are retail and being sold. Anand did me a solid. I literally was about to pre buy all my stuff for rocket lake now I am not. If intel does pull some kind of magic they can change the reviews. No harm if something changes, but I don't think it will. I don't want AMD either so I am without a home right now also not really interested in big little cores either unless it pulls off some magic. This is also just a hardware review site not some major corporation not sure why you hold them to such a high standard.SaturnusDK - Saturday, March 6, 2021 - link
big-little cores will help improve efficiency so the mobile parts may become competitive again but for desktop parts it will almost certainly decrease performance rather than increasing it. There is literally zero chance that the scheduler will always get it right which processes are assigned to which cores, and moving processes between different core types will always incur a performance hit.Cooe - Saturday, March 6, 2021 - link
"I don't want AMD either"Uhhh... why not? Deliberately not buying the best products available on the market because of the company that made them doesn't make ANY SENSE unless the company being referred to tends to be on the wrong side of basic business ethics/morality. You are cutting off your nose to spite your face here for no other reason than some misplaced stupid fanboy loyalty... -_-
TheinsanegamerN - Monday, March 8, 2021 - link
Criticism of intel for releasing a hot dung heap isnt "AMD worshipping".Spunjji - Saturday, March 6, 2021 - link
Trying to make a moralistic critique out of your own personal objections is a desperate move.If Intel want to do pre-orders before they've even announced the product, yet they're going to release stock to a retailer, then personally I'm only too happy for one of the best CPU testing websites out there to buy one and put the screws to it.
scineram - Saturday, March 6, 2021 - link
Why didn’t they buy one then?dsplover - Friday, March 5, 2021 - link
What a waste of time.5xxx is the better Design.
3xxx series is when Intel should’ve noticed they were in trouble.
Lucky for them I don’t need the best, just the most mature hassle free CPU.
But I’m not buying anything until the server parts 480/1200 Xeons drop prices.
If AMD can deliver on the 5xxx APU before that, I’m going to jump ship.
blppt - Saturday, March 6, 2021 - link
I'm not sure anybody saw the massive single thread or core performance jump from the 3 series to the 5 series. I remember being astonished at the geekbench and CB synthetics, thinking there was something bugged somewhere.3xxx Zen was good, but 5xxx series is outstanding.
Otritus - Saturday, March 6, 2021 - link
Zen 2 was always market as a a zen++ that for some reason had a massive enough performance uplift to be comparable to a new microarchitecture. Zen 3 on the other hand was marketed as a newly designed microarchitecture with the largest ipc improvements (from AMD) since zen. AMD delivered exactly what was promised, a significantly faster and more efficient product on 7nm.The reason why you would have been astonished by the synthetics is because we haven't had a typical (15-30%) microarchitecture performance uplift since sandy bridge. Haswell was only ~11% faster than ivy bridge. The thing that we didnt see coming was zen 3 being called 5000 instead of 4000 and the price hikes.
Makste - Saturday, March 6, 2021 - link
Yes. And we also got spoiled in the process. Now these are the kind of jumps we want with every release because AMD showed us that they were possible. The thrill of these performance jumps AMD has brought to the table are nothing short of exciting to the entire semiconductor realm. Things Intel should have been doing.Qasar - Saturday, March 6, 2021 - link
i read else where that the " reason " intel didnt go above quad cores, or increase performance more then 10% gen on gen before Zen, was cause if they did, then amd would of been put out of business.i would like to know is, when did/has intel ever done anything that was good for anyone other then intel ?
blppt - Sunday, March 7, 2021 - link
Well, if AMD were to go under, then Intel would be subject to increased scrutiny of being a near-monopoly. Thats probably why they would have done such a thing (if this was true).Its similar to what Microsoft did way back in the day, propping up troubled Apple with cash infusions.
Qasar - Sunday, March 7, 2021 - link
but the part is, NO one really knows what would of, or could of happened then. all it is, is one big IF.Spunjji - Monday, March 8, 2021 - link
That sounds like a rationalisation more than a reason - they had the clear motive of protecting the revenue from their "HEDT" and server processor lines. It was telling how quickly they dropped their trousers on those products after Ryzen launched.Zizy - Monday, March 8, 2021 - link
That would imply some incredibly hefty performance reserves to be unlocked when Intel felt threatened.Nah, all Intel seems to have done is to keep shrinking chip sizes at same 4C. This maybe was picked to help AMD, or just to help their bottom line. All clocks, IPC and whatever other improvements they had available were put in end products - otherwise these would be unlocked in Ice and Tiger lake, or this Rocket Lake abomination.
TheinsanegamerN - Monday, March 8, 2021 - link
The biggest reason, and thisis one thing people dont want to admit, is there was simply no reason to. Their 6 core HDET parts were consistently slower then their 4 core peasant parts in consumer applications. So spending the money on making a new die just to have 6 cores on peasant class would have been a waste of money.Cooe - Saturday, March 6, 2021 - link
Zen 2 was NOT "Zen++". Not even close. It had a ton of MASSIVE changes vs Zen 1.Sure, Zen 2 is closer to Zen 1 than Zen 3 is to Zen 2, but that's not saying much at all....
Mighty Molecule - Friday, March 5, 2021 - link
Remarkably underwhelming and disappointing. If the only distinction between 11900K and 10700K is left to binned core frequencies, you’re then left with the upgrades to PCIe 4 (which really is a designed delay from Comet Lake) and backported IPC’s at a tragic TDP. Yikes.shabby - Friday, March 5, 2021 - link
Ian can you guys start posting the actual wattage usage during each test beside the 125w rating? Would be nice seeing how much each cpu used during each specific test, if possible.DV8_MKD - Friday, March 5, 2021 - link
Yeah, agree. That would be interesting to see.IanCutress - Friday, March 5, 2021 - link
Adding in power detection during a test will often slow it down a bit, or require running the tests twice (they already take 24hr) and then parsing the data (probably more time than it's worth). Which would all lend to a slower review cycle. No easy way to automate it, unfortunately.lmcd - Friday, March 5, 2021 - link
Maybe true but hand-picking a few benchmarks would be nice. The power draw section is opaque from our vantage because we don't know what performance looks like at max power draw.idimitro - Friday, March 5, 2021 - link
A wall meter can be automated, easily. Even system level power consumption will be more useful that the TDP figures published by the manufacturers these days.shabby - Friday, March 5, 2021 - link
Ah ok, can you maybe hand pick a few tests? It's an interesting metric.Spunjji - Saturday, March 6, 2021 - link
Agreed here. I get that it's a big ask, but it would be very useful to see in terms of sizing cooling up for system builds.SL2 - Friday, March 5, 2021 - link
"...the only parts readily available on retail shelves right now are from Intel."Maybe in North America, but not in Europe. The only CPU missing here is the 5950X, the other models are in stock.
Milk Fondu - Friday, March 5, 2021 - link
This.. is rough. As someone who has been team blue for a long time, and who knew this launch was going to be bad: this is bad. I'm sure things will get better with a bios update or two, but 14nm has just been stretched too thin. It's begging to be put down.lmcd - Friday, March 5, 2021 - link
I mean, part of the problem is simple though: AVX is a waste of die space. An onboard GPU appears to be a waste of die space -- AMD sure doesn't need to sell one on its desktops. And a modular approach on desktop does wonders for yields.This launch should've been 14nm, 10 cores, more spaced out in the die (if possible) and no iGPU on the die at all. Saves the Xe backport money and produces the same effect, since these real-world power numbers are unsustainable in anything other than a powerful workstation.
nandnandnand - Friday, March 5, 2021 - link
AMD is expected to add an iGPU chiplet starting with Zen 4. It might be added for machine learning acceleration in addition to entry-level graphics.Intel will actually have RKL chips without iGPU, for what that's worth. i7-11700KF and i9-11900KF. I don't think it will improve the thermal performance though, just lower the price a few dollars.
lmcd - Friday, March 5, 2021 - link
Those are binned. I mean a die that is designed without an iGPU at all. It sets up the entire product for failure.iGPU chiplet is exciting. Hopefully Intel follows suit as soon as they can.
SaturnusDK - Friday, March 5, 2021 - link
Wouldn't that technically be an eGPU when on a separate die instead of an iGPU?nandnandnand - Friday, March 5, 2021 - link
Maybe. I just know that I want it. If it has graphics and L4 cache, I'll get it ASAP and use it without a discrete GPU.Alex_Haddock - Saturday, March 6, 2021 - link
Having just been unable to use my PC for 10 weeks whilst waiting for a Zotac 3080 RMA (pray nothing fails in this silicon shortage..) I’m all for every CPU having a basic GPU chiplet in the future...The_Assimilator - Saturday, March 6, 2021 - link
This is why you buy a low-end spare GPU off Ebay.CiccioB - Friday, March 5, 2021 - link
I think you, as many others, over estimate the need for a dedicated GPU.Intel sells more GPUs (and that means they are all iGPUs up to now) than AMD and Nvidia together, meaning that most users are happy with the iGPU and don't want to spend more for another component. Especially in this period where a 4 year old GPU costs twice what it costed at its launch.
Not putting a iGPU for how small it can be would be a failure, not the opposite.
AMD is struggling in trying to put an iGPU into its desktop CPUs and will have to resort to a dedicated chiplet for that (with performances and power consumptions stil to be seen) and Intel would do the wrong step in going backward directions to the "full on package integration (and in die if possible)".
They do not sell discrete GPUs (not yet though, and not decent ones for a while when they will) and not providing a integrated GPU will just mean giving the chance to its competitors to make more money when they could easily avoid that and still use more power.
Because in all those benchmarks none has evidenced that AMD's set up needs a discrete GPUs (if you want to see something on your monitor) that needs much more power than Intel integrated one, so you should take into account the sum of AMD CPU+dGPU vs Intel CPU only to see what are the real differences between the two.
arashi - Friday, March 5, 2021 - link
Wow you're so wrong about so many things.Qasar - Saturday, March 6, 2021 - link
and you over estimate the need for an iGPU on a high end cpu, that most will probably DISABLE, and use a discrete one." Intel sells more GPUs (and that means they are all iGPUs up to now) than AMD and Nvidia together, " of course they do, cause they sell one with EVERY desktop CPU they sell.
"Because in all those benchmarks none has evidenced that AMD's set up needs a discrete GPUs (if you want to see something on your monitor) that needs much more power than Intel integrated one, so you should take into account the sum of AMD CPU+dGPU vs Intel CPU only to see what are the real differences between the two. " um no, cause as i said above, most will probably disable the igpu on the higher end cpus, to use a discrete one. would YOU buy this cpu, and use the iGPU to play ANY game that was released in the last say 2-3 years ? lets see you play CP2077 or and of the games that were tested in this article.
haukionkannel - Saturday, March 6, 2021 - link
True, Intel is the biggest gpu seller in the earth because of integrated gpu in cpus...And Also true that most people use cpus in the Office Computers, so They don`t need more speed that intregrated gpu offer. Gamers Are really small minority in the world.
But for gamers neither Intel nor amd makes GPUs fast enough. Maybe in the few years integrated GPUs will become gamers choise, but not yet. But all in all gamers don`t count in the big picture.
Otritus - Saturday, March 6, 2021 - link
If you are the type of person to need an igpu, you don't need more than 4 cores, or you need a massively parallel processor (many cores for compiling, simulations, etc). AMD not having an igpu is not a fail because if I need that much cpu, I likely need a fast gpu. Intel's 96 EU Xe does not make the cut, let alone these 32 EU parts.If Intel gave an igpu fast enough for at least 1080p medium, your argument may have some merit because I can still game in spite of not being able to find a dgpu. Most non-gaming or compute-heavy work tasks people can already do with their existing computers/phones. Meaning the inclusion of an igpu is meaningless because I can already web browse on another device.
scineram - Saturday, March 6, 2021 - link
I am literally waiting for Zen 3 APU.idimitro - Friday, March 5, 2021 - link
Not exactly true - the 10nm is pushed too thick :).Cooe - Saturday, March 6, 2021 - link
No BIOS update is going to be able to "fix" the physically much larger cores & caches creating higher core-for-core AND core-to-memory latency. That's just an inherent part of the new design.madseven7 - Friday, March 5, 2021 - link
3 generations of processors within 18 months and intel still won't beat AMD. LOL LOL LOL. I LOVE IT!!!Achaios - Friday, March 5, 2021 - link
We are back to 2007.Check the review for Core 2 Extreme QX9650 here on ANANDTECH.
This kind of energy consumption was noted on the QX9650 when it hit around 4 GHz.
Intel has come Full Circle.
blppt - Friday, March 5, 2021 - link
Its actually worse now---Intel has lost the gaming AND power consumption lead here. QX9650 may have been a blast furnace, but it was still speedy.haukionkannel - Saturday, March 6, 2021 - link
Well Gaming is not important to most people... but power usage does matter. So not bad product, but shows that Intel needs the next step Sooner than later.blppt - Sunday, March 7, 2021 - link
I only posted "gaming" because especially back when the 9650 was in its heyday, single thread/core/IPC performance was by far the most important thing for gaming. Other apps benefit from such things as well---especially back then when coders were still learning to properly code for more than 2 cores.favro - Saturday, March 6, 2021 - link
The first thought that came to me was Pentium 4!!! Twice around the circleLekz - Friday, March 5, 2021 - link
> However, due to high demand and prioritizing commercial and enterprise contracts, the only parts readily available on retail shelves right now are from Intel.Wonder if this will continue to be the case. At least in the US, there seems to be plenty of retail Ryzen 5000 availability for about 3 weeks now.
TheinsanegamerN - Friday, March 5, 2021 - link
What planet are you living on? The only chips reliably in stock are the 5600x and sometimes the 5800x, and $100+ over MSRP. The ryzen 9s are nowhere to be found.Slash3 - Friday, March 5, 2021 - link
The 5800X is in stock right now on AMD's webstore, for anyone reading this at date of publication.I agree with them that the situation is improving, but also agree with you in that the 5900X is very tough to come by, with the 5950X effectively ceasing to exist.
blppt - Friday, March 5, 2021 - link
As usual, you have to watch Microcenter stocks daily and hope you get to the store in time to pick one up (they do not allow you to reserve it online).Newegg has been hopeless for me with the new gpus and cpus.
Slash3 - Friday, March 5, 2021 - link
As someone 3,500+ miles from the nearest Microcenter, I press my face up against the glass longingly every time someone mentions their in-store pricing and availability.I briefly attempted to purchase a 5950X from Best Buy, but their Zen 3 CPUs were listed both as "Online only" as well as "Cannot be shipped." I then tried ship to store, and was met with "Unavailable for ship to store within 250 miles."
So.
That's awkward.
I'll ride it out.
blppt - Saturday, March 6, 2021 - link
I feel for you, man. I'm blessed in the sense that there are like 5 microcenters within a 100 mile radius of me. Its still basically luck though---I would have no chance of getting there in time to get a 6900XT or 5900X/5950X unless they had double digit stock, because I wouldn't be able to get there until work is over, and the stock dives during the day.As somebody noted before, they still have nice stock of the 5800X.
the_eraser1 - Friday, March 5, 2021 - link
I bought a 5800X (for myself) and 5600X (for my dad) at MSRP from my local Microcenter, with literal piles more remaining on the shelves (and more in the "back"). It's only Ryzen 9 that's hard to find.Lekz - Friday, March 5, 2021 - link
I live in GA, and both Microcenters are well stocked of 5600X/5800X at MSRP. Not only that, but Walmart, Newegg, Amazon, and others have had 5600X/5800X stock come up daily - I kept the notifications on despite getting what I needed. Ryzen 9 has had a lot less stock - that I agree with.kmmatney - Saturday, March 6, 2021 - link
At my local Microcenter they've had both 5600X and 5800X cpus in stock for a while now.Cooe - Saturday, March 6, 2021 - link
The 5800X is available at MSRP almost everywhere... Has been for awhile now.rolfaalto - Friday, March 5, 2021 - link
"When AVX-512 comes to play, every-one else goes home" pretty much nails it. If your code runs AVX-512 and is limited by single core speed (plus max bandwidth to a RTX3090) then no other processor comes close.lobz - Friday, March 5, 2021 - link
The customers needing that truly rejoice! All 6 of them worldwide.Moravid - Friday, March 5, 2021 - link
Exactly how many people buying client PC processors intend to run their AVX 512? Very niche workloadSaturnusDK - Friday, March 5, 2021 - link
AVX512 in client class hardware could perhaps get more sales for these parts at the expense of lower sales of higher margin hardware that would normally be required. An extremely poor financial decision to include AVX512 in client class hardware, especially considering it very well may be at the expense of two regular CPU dies which would have narrowed the performance gap to it's competition.Spunjji - Saturday, March 6, 2021 - link
Only if it's 100% AVX-512 though. This CPU still loses to AMD in tests where there is a sprinkling of AVX-512, which is how things will always look in client workloads. Most people running simulations will be running them on workstations that support ECC RAM.Otritus - Saturday, March 6, 2021 - link
From my understanding, low end xeons take the consumer dies and enable ECC. So the dies could have meaning if intel is trying to increase the volume of avx-512 computers in the market. If the goal was for the die to be used to retake the gaming crown or remain competitive with AMD, it was an absolute fail. The only reason it was included was probably because sunny cove was designed around it's inclusion, and removing it from cypress cove would have delayed the launch to the point that Intel may as well just wait for Alder Lake.supdawgwtfd - Friday, March 5, 2021 - link
"This made the advantage of AVX-512 suitably only for strong high-performance server code. But now Intel has enabled AVX-512 across its product line, from notebook to enterprise, with the running AI code faster, and enabling a new use cases."As usual multiple spelling mistakes...
Is Anandtech ever going to proof read their articles before releasing them?
Billy Tallis - Friday, March 5, 2021 - link
Those aren't exactly spelling mistakes. Those typos amount to grammar mistakes, but using correctly-spelled words. That makes them a lot harder for an automated program to accurately catch. With humans proofreading, it's very hard to catch your own typos, and each new person who proofreads might catch 80-90% of the remaining mistakes. The only ways around that are to have several more people do proofreading passes, or to set the article aside for a few days before going over it again with less familiarity.MartinP - Friday, March 5, 2021 - link
It is not just the grammar. It is inept and shoddy to be using a cooler from 2008, which has not been in production for years and refer to it as 'the best air cooler on the market'.Actually, it's a down right untruth. The Thermalright True Cooper was not considered the best cooler in 2008 (see reviews on Overclockers.com) so it most definitely not the best one in 2021.
Why not use a Noctua D15? Or even at least use exactly the same Noctua on both platforms.
This is methodology that Anandtech should be ashamed to publish.
Billy Tallis - Friday, March 5, 2021 - link
Did you mean to reply to a different comment? Because it looks like you're not even trying to say anything that relates to my comment or the one above it.Cooe - Saturday, March 6, 2021 - link
Because it IS the best air cooler on the market. The 100% cooper cooler they used absolutely CRUSHES an NH-D15. If they can't do it with the pure copper, a Noctua had no chance in HELL!!!Qasar - Sunday, March 7, 2021 - link
but, can you even buy that cooler anymore ?Zizy - Monday, March 8, 2021 - link
Importance of this depends what exactly you mean by "on the market". Cooler WAS on the market - ie it isn't something custom made. If even that cooler was insufficient, anything else currently on the market would be even worse (assuming it is actually the best air cooler for this, something I have no idea about)CiccioB - Monday, March 8, 2021 - link
Right, should test the CPU with the same components.So as Intel does not need a dGPU, it should not be used on AMD as well. And use the AMD platform without any video out, as it is sold.
Sound fair, doesn't it?
Qasar - Monday, March 8, 2021 - link
a little sour on the fact that rocket lake is garbage, and you cant handle it ?Ryan Smith - Saturday, March 6, 2021 - link
Thank you for pointing out that grammatical error. I've gone ahead and corrected it."Is Anandtech ever going to proof read their articles before releasing them?"
This article actually did get a full proof-reading pass by me. But even then, I missed that error.
But to answer the question at hand, for the typical article the answer is "no, it will not get proof read". We let our copy editor go almost 15 years ago during the Great Recession. And while I'd still like to have one, it simply doesn't make economic sense. The modern, Google-driven news cycle does not reward copy editing for online news. It's better to post more articles, and to post them sooner.
GeoffreyA - Saturday, March 6, 2021 - link
Thanks, Ryan. You know, these people who keep on pointing out tiiiiinnnyyy errors are getting tedious. I sigh every time I see another one of these comments castigating grammar or spelling. It's better to have excellent material, like Anandtech has, with the occasional typographical slip, than lacklustre material + "perfection" (like some other sites).BubiSlav - Friday, March 5, 2021 - link
Been planning a rocket lake build for over a year. Already bought everything else including motherboard, was only waiting on CPU. Thanks for ruining my day. What the hell am I supposed to do now?Slash3 - Friday, March 5, 2021 - link
Devil's advocate: It's not a home run, certainly not anything revolutionary, but it's still a fine chip.If you were already interested in the i9-10900K, this is essentially the same thing with a few feature improvements. It's not a great value proposition (assuming it lands at ~$450), but you still have:
· Improvements to single core performance due to IPC uplift
· AVX 512 (if you should need it for anything)
· Gen4 support for the primary PCI Express x16 slot (or x8/x8 for some boards)
· Gen4 support for the dedicated NVMe M.2 slot
· 8GB/s (vs 4GB/s) DMI link bandwidth, for storage or peripherals connected through the chipset
· 32CU Xe iGPU, for ~33% improvement vs 10900K's Gen11 IGP. Full features, including AV1 decode
· 20Gbit USB 3
· TB3/4 support (depending on board model)
If you're wanting an Intel system, it's still arguably a solid option. It offers useful features that the 10th gen CPUs lack, and Gen4 alone may be worth it if you intend to use fast storage ("RTX I/O" or DirectStorage down the line) or want to be able to upgrade to a faster GPU in the future.
Is it a good value? Not particularly, but the competing 5800X isn't a bargain at $450, either.
You won't find yourself shouting from the rooftops, but it'll do the job and let you skip the first iteration of Intel and AMD's DDR5 platforms. By then, something more interesting will definitely have come along, and the early adopter tax should have eased up.
BubiSlav - Friday, March 5, 2021 - link
All I really care about is gaming performance so none of those things interest me too much."it'll do the job and let you skip the first iteration of Intel and AMD's DDR5 platforms"
This was basically my plan. Use this PC for a few years then upgrade again when these new technologies matured a little.
DigitalFreak - Friday, March 5, 2021 - link
"What the hell am I supposed to do now?"Learn a lesson that you shouldn't buy hardware before you read the reviews?
Alistair - Friday, March 5, 2021 - link
1 month return policy on that motherboard.jonathan1683 - Friday, March 5, 2021 - link
Literally was on newegg earlier at work doing the same, only reason I didnt was i got busy! ugh very disappointingJimmyTheFish - Friday, March 5, 2021 - link
Well that was a bit silly wasn't it?What you're supposed to do now is base future purchasing decisions off benchmarks and not hype, empty promises and marketing, especially when it comes to a company who has given themselves a reputation for under delivering over the last half decade or so.
jonathan1683 - Saturday, March 6, 2021 - link
YES it was!MadAd - Saturday, March 6, 2021 - link
A technical annoyance maybe so eBay the motherboard quickly to some sweaty palmed Intel fanboy who doesnt read Anand and reuse everything else in a more modern Ryzen build?Otherwise maybe email Intel to see if theres any discount coupon available for 2 kilo coolers? Or maybe go head with that 4 fan twin 240mm water build you always wanted?
scineram - Saturday, March 6, 2021 - link
Get 10700K or 10850K. Good pricing.The_Assimilator - Saturday, March 6, 2021 - link
Accept that this is the price you pay for not being a smart man.AntonErtl - Sunday, March 7, 2021 - link
Be happy that you can get a CPU early (If Mindfactory has some in stock again) and enjoy! Including the various problems of being one of the first on the platform. Take it as an adventure!airdrifting - Friday, March 5, 2021 - link
This little 8-core has the power consumption of a 32 core Threadripper (3970X / 2990WX), this is pure madness.ywyak - Friday, March 5, 2021 - link
Greeting,Would it be possible going forward to list in the review on whether the current CPU supports both AV1 encoding & decoding? (To my understanding, RKL only support decoding).
Slash3 - Saturday, March 6, 2021 - link
Correct, AV1 decoding only. There was an early Intel slide that made the rounds which listed support for AV1 encode, but it was quickly acknowledged as a typo by Intel. Decode only.GeoffreyA - Sunday, March 7, 2021 - link
My guess is that you're looking for faster AV1 encoding. It might take a while before it reaches hardware, and even then, quality and size will likely suffer. The best option at present is to use libaom, through FFmpeg for example, and set the "cpu-used" parameter to 8 and see how that goes. If fast, keep on dropping by one. Intel's SVT-AV1 is another option: fast but lacking in quality. Also, keep an eye on VVC and x266: it's in the same tier as AV1 but, from my testing with the Fraunhofer encoder, slightly better, sharper, and faster.zamroni - Friday, March 5, 2021 - link
Seeing the power consumption, this rocket lake should be called bulldozer lakeGeoffreyA - Saturday, March 6, 2021 - link
Yep, nice one.RanFodar - Friday, March 5, 2021 - link
At this point the spotlight will shine at the i5, not this one. This is, to my view, an embarassing feat since Pentium 4.Cooe - Saturday, March 6, 2021 - link
The spotlight won't shine on any part of this lineup...Chaser - Friday, March 5, 2021 - link
INTEL fell asleep at the wheel.DigitalFreak - Friday, March 5, 2021 - link
They've been asleep at the wheel for so long they must be driving a Tesla.croc - Friday, March 5, 2021 - link
OK, so once more into the gap with 14nm... looking about as good as could be expected for a node this old. Then there are all the AMD faobois.. To whom I will ask, where is my 5970x? Hmmm??? And to either camp, why can't I have 32 or more (40 seems about right...) PCIe 4.0 lanes on a desktop CPU? From my perspective, neither camp is executing particularly well. Intel at least has Gelsinger back, we'll see what he can do in about a year (or two). AMD, Dunno... If they can't execute at 7nm, what makes anyone think 5nm will be easier? Then there is poor TSMC, having to worry about China's takeover. And mark my words, China WILL take over.Alistair - Friday, March 5, 2021 - link
China's been threatening Taiwan for 70 years. Nothing new. TSMC is the top because they work hard for it.Oxford Guy - Sunday, March 28, 2021 - link
I'm certain everyone in Hong Kong agrees.Thanny - Friday, March 5, 2021 - link
Both Intel and AMD failed utterly to avoid turning "mainstream" computers into toys when they moved PCIe from the chipset to the processor.Intel moved their non-toy processors into the HEDT space, before it even had a name (I had to buy Sandy Bridge-E rather than Ivy Bridge to get a usable computer back in 2012), at ever increasing prices.
AMD at least brought that buy-in price down a little bit with Threadripper, then turned it way back up with the Zen 2 incarnation. I'm hoping the Zen 3 TR chips start at 12 or 16 cores, so an upgrade is reasonable. I don't need 24 cores, but I do need more lanes.
I put 40 lanes as the bare minimum for a non-toy PC.
lmcd - Saturday, March 6, 2021 - link
40 seems high but 20 is definitely too low. 28 wouldn't be bad.Spunjji - Saturday, March 6, 2021 - link
24 is fine for the vast majority of people, 28 would certainly be nice. I don't tend to take comments seriously when they refer to computers with 16 cores and the ability to run as much IO as 90% of users need as "toys", though.PixyMisa - Saturday, March 6, 2021 - link
Well, there's ThreadRipper Pro now. $1150 for 16 cores, but gives you 128 lanes of PCIe 4.0.Slash3 - Saturday, March 6, 2021 - link
It's a good option, but the $1,000 motherboard does put a bit of a damper on things. I was also hoping for a 16C/32T Threadripper 3000 part, but when it became obvious that they didn't intend to offer one I picked up a 3950X instead. While it has in no way disappointed me, the extra lanes and memory bandwidth of the Threadripper platform would have been appreciated. A $1,000 16C version would have been an absolute no-brainer vs the 10980XE and still leave a market space for the $750 3950X.The lack of 8, 12 or 16 core Threadrippers at the end of 2019 is essentially the only reason X299 still exists; it baffles me why they didn't pursue it at the time.
SaturnusDK - Saturday, March 6, 2021 - link
Are you sure you need 40 PCIe 4.0 lanes though. I'm willing to be you probably just need more PCIe lanes or PCIe 2.0 lanes at most. For that you can get bifurcation cards that splits or doubles PCIe lanes. AmeriRack is an example of a company that makes these.Zan Lynx - Sunday, March 14, 2021 - link
The answer to "why can't I have 32 or more PCIe 4.0 lanes" is pretty obvious. It's the socket size and pin count. Duh.So for AMD on the AM4 socket there is absolutely no way to increase the PCIe lanes without a new socket. Which is what the sTRX and sWRX sockets are for. 4,094 pins for the latest Threadripper Pro which lets them have 8 memory channels and 128 PCIe lanes. As compared to AM4's 1,331 pins.
AM4 on the x570 chipset is still an improvement on PCIe lanes because the PCIe4 speed allows doubling the chipset bandwidth. So it can drive twice the number of PCIe 3.0 devices as before.
Oxford Guy - Friday, March 5, 2021 - link
'A marginal win for Intel in Blender is a good result'Is it a win, though?
The power consumption and hearing loss... Getting tinnitus and running up the power bill are both not advantages.
Also, were any of the AMD 5800X results thermally limited because of the use of a quieter less performant cooler?
Apples to apples testing would be better. When you do your full review it would be best to use the same cooler for AMD and Intel.
Chaitanya - Friday, March 5, 2021 - link
Yikes, wont be too long before new CEO and other staff at Intel will be on firing line.nandnandnand - Friday, March 5, 2021 - link
As long as they are continuing to make tons of profit and enterprise customers aren't leaving, it doesn't matter.Chaitanya - Saturday, March 6, 2021 - link
How long with enterprise endure these literal useless products? Apple already left for ARM and customer size of Apple leaving isn't good optics .Machinus - Friday, March 5, 2021 - link
14nm++++++++nandnandnand - Friday, March 5, 2021 - link
14nm+++++++-NeatOman - Friday, March 5, 2021 - link
*Intel Core FX-1170KSpunjji - Saturday, March 6, 2021 - link
Oof, burn!blppt - Saturday, March 13, 2021 - link
Its hard to believe that the same company that gave us one of the worst cpus of all time (FX-9590) has now birthed Zen 3.I had a 9590. What a piece of overheating junk that thing was.
hansip87 - Saturday, March 6, 2021 - link
I feel like the 11700k is essentially nerfed in order to make room for i9 to shine hotter (pun intended)lmcd - Saturday, March 6, 2021 - link
it's not a pun if one form doesn't work at all. Shine hotter? Pun better, padawan.Hifihedgehog - Saturday, March 6, 2021 - link
https://i.redd.it/cf7wqkvhicl61.pngI predicted ~300W peak power about a year or so ago when I had first heard they were making the big mistake of bringing AVX-512 to 14nm mainstream consumer processors. Why on earth? There was a valid, wise, very, very, very good reason Intel had reserved AVX-512 to just their 14-nm HEDT processors and that had always been the furnace-like heat and nuclear-like power consumption of it. Even with the best logical design improvements from a new microarchitecture, it is still an extremely intensive logical pill of a task to swallow. Now, we see that reason in full, unadulterated display. You bring a massively complex instruction set extension to a higher process node where you have far lengthier physical networks (meaning essentially longer wires, increased resistance, higher power, and maximum heat) and, of course, you are going to have a steaming pile. Remember this review is only looking at the number two product, the Core i7-11700K which has lower clocks and lower power draw. The Core i9-11900K will likely need a 360- or 420-mm AIO just to not thermal throttle like mad. My 5950X with its meager 240mm AIO (Corsair H100i RGB Platinum) that runs at the quiet mode setting is laughing its butt off right about now. When Ian Cutress had to use an obnoxiously loud 170 CFM fan (I have used 100 CFM Deltas and those already annoy most PC enthusiasts) on a massive 4-pound, full copper heatsink to tame the 11700K's 290W, I shudder to think. Will the 11900K be outdoing the FX-9590's record-making peak power draw of 350W? Ian easily could have gotten the 11900K also at retail, but I think he is holding back on that because he already knows the 11900K is going to be a throttling disaster and only the 11700K is an actually usable processor. *mike drop*
Slash3 - Saturday, March 6, 2021 - link
There's a reason MSI includes a 360mm AIO with their Z590 MEG Ace reviewer kits.nandnandnand - Saturday, March 6, 2021 - link
Only the magic of binning can save the 11900K.I'll note that the leaks have 11900K at a 100 MHz lower base clock.
Everett F Sargent - Saturday, March 6, 2021 - link
This graph suggests that the i9-11900K will pull 280-285W at an all core clock of 4.8Ghz for AVX2 type operations ...https://i.imgur.com/QJOx2k7.png
Everett F Sargent - Saturday, March 6, 2021 - link
5960X should be 5950X in above graph, like this ...https://i.imgur.com/8BEsGVo.png
Material3600 - Saturday, March 6, 2021 - link
哟哟哟,这不是Intel吗?Hifihedgehog - Saturday, March 6, 2021 - link
No, this is Bulldozer, back from the dead.https://i.redd.it/cf7wqkvhicl61.png
GeoffreyA - Sunday, March 7, 2021 - link
It's ironic that this descends straight from the illustrious CPUs that put Bulldozer to shame in the last decade.ozzuneoj86 - Saturday, March 6, 2021 - link
Wow... 5800X looks fantastic here compared to the competition.I'm happy with my 3600, but it's really nice to know that the 5000 series is available as a future upgrade.
Intel will come back eventually... because $$$$$$$... but I'm honestly surprised that AMD's chips have so thoroughly beaten Intel's on in basically all areas for two generations. This hasn't happened since the Athlon 64 vs Pentium 4\D days. Whether Intel's next big performance jump will be a complete game changer like Core2 though... time will tell.
Atom2 - Saturday, March 6, 2021 - link
Typical applications these days are not optimized for AVX512 or even AVX. That is the main problem for Intel. As soon as you do benchmarks using software compiled with Intel C++ and their libraries, AMD is not even on the map. Despite huge possible gains (10x), most big companies never invested.Targon - Saturday, March 6, 2021 - link
So, what you are suggesting is that because companies are not being 100 percent focused on Intel performance, Intel loses in the majority of situations now?Atom2 - Sunday, March 7, 2021 - link
Most companies are obviously not keen on investing in performance. When first compilers came out in 1960s, they brushed away with assembler, which produced 3x faster code, because it was cheaper to write software with them. We have a similar situation now, but in reverse. Intel adds new instructions, which require "more development cost" in order to achieve higher speed. AMD instead focused on speeding up existing code, which does not require more development cost. On Anandtech you have one (!) benchmark which is hand optimized (3DPM) for AVX512, but there is ZERO reports, on out of the box performance of Intel C++ (never mind optimizing C++ for SIMD). The whole situation is further skewed, by the so called power inefficiency of AVX512, but completely forgetting, that when used correctly, these instructions result in 4x faster code per core than AMD AVX2. Even Intel AVX2 is 2x faster than AMD AVX2, when done properly. Yes, Intel can draw 300W, which is horrible, but it can run 4x faster than best AMD with that power draw. The final power/performance ratio for what you could call "quality software" is in fact still two to one in favor of Intel. But when running GCC compiled code, written by just "anybody", this is not the case and Intel struggles. If you buy Intel, you also need a qualified development team and no prejudice for using Intel compilers and libraries to make TCO stand.AntonErtl - Sunday, March 7, 2021 - link
Here are the number of cycles for 2000x2000 matrix multiplication with 1 thread on libopenblas-r0.3.5, which is done properly.1158M Skylake (6600K)
2198M Zen (1600X)
1451M Zen2 (3700X)
So, Skylake is indeed almost twice as fast compared to Zen, but not compared to Zen2 (assuming similar clock rates). It will be interesting to see Zen3 and Sunny/Cypress Cove.
Atom2 - Sunday, March 7, 2021 - link
Matrix multiply is already maxed out with AVX2 and Intel MKL shows no gains with AVX512 capable CPU. Using one thread still requires blocked matrix multiply to have data in cache. On Skylake X-series 4GHz, with 8 cores, I get 65ms for multiplying of two matrices, double precision 2000x2000 in size. With a single thread, it needs 290ms.AntonErtl - Sunday, March 7, 2021 - link
4GHz*290ms=1160M cycles, same as the consumer Skylake above. I would have expected AVX-512 to gain a lot for matrix multiplication: wider units, and more registers for register blocking.Atom2 - Sunday, March 7, 2021 - link
Matrix Multiply depends on cache L1 to CPU core bandwidth, which is about 1.9TBytes/s and this is relatively fixed in the last years/generations. To get more speed with AVX2 and especially AVX512 you need an algorithm, which has more than one or two CPU instructions (multiply + add for dot product) per one memory read/write. This is a similar situation as with GPUs, which require algorithms with 50 cycles of instructions or so per one memory read to really show their muscles. During those 50 cycles, the GPU thread can work with registers or wait for memory to respond.AntonErtl - Sunday, March 7, 2021 - link
For 2000x2000 DGEMM, libopenblas on AVX performs 1G loads, 0.285G stores, and 2G AVX FMAs. E.g., a Skylake can perform 2 AVX256 loads and 2 AVX FMAs per cycle, so this is bounded by the FMAs. The Sunny Cove (and, I expect, it's 14nm offspring Cypress Cove) can do 2 twice-as-wide loads and 2 stores, as well as 3 twice-as-wide FMAs; and AVX512 has more registers, allowing to reduce the number of loads per FMA. So I expect a speedup by a factor of two from the wider units, and some additional speedup from the additional resources available.Atom2 - Monday, March 8, 2021 - link
It does not happen. Intel started introducing AVX512 some five years ago in to their toolset and for large matrix multiply the presence of AVX512 with 2FMAs has zero effect. There is difference however for other algorithms, which in fact are 2x faster. For example vectorized math functions: Sin, Cos, etc.. where the memory bandwidth requirement is much less.arashi - Saturday, March 6, 2021 - link
Intel wins bigly... mainly because ICC nerfs the shit out of performance on non Intel PCs? Although both are large corporations, at least AMD is much less scummy.scineram - Saturday, March 6, 2021 - link
No, AVX2 is getting more heavily used over time.Bik - Saturday, March 6, 2021 - link
For apple to apple comparision, Avx2 workload is "only" 225w vs 142w on AMD. That's 50% more power for marginal less performance, not looking good.maroon1 - Saturday, March 6, 2021 - link
So this is even worse than 10nm sunny cove because it has worse L3 latency which effect gaming performanceIntel should have just released 8 cores tiger lake for desktop. Even the mobile version is beating rocket lake in single-thread. 125w tiger lake would easily destroy that thing
Spunjji - Saturday, March 6, 2021 - link
I don't know if they could put 125W through Tiger Lake and have it survive - the heat at the most dense parts would be intense. I think that's a big part of the problem. The other part is that they are *still* struggling with either yield or fab capacity on 10SF (possibly both), which is why TL-H45 still isn't here. If they had to split that across desktop and notebook lines then they'd have a similar supply problem to AMD, and for Intel that would be a much larger issue.AnnoyedGrunt - Saturday, March 6, 2021 - link
I suspect AMD CPUs are becoming available because no one can build a computer due to the lack of video cards.The one benefit this CPU has over the AMD counterparts is the iGPU, which at least allows a functioning system. I would love to see how this new iGPU performs.
-AG
Otritus - Saturday, March 6, 2021 - link
If I want a functioning system I can use my existing pc, laptop, or phone. The only reason I would upgrade is for a better gaming experience. My i5 2400 just isnt cutting it in games any more, and a 5600x is really enticing. Just waiting for a 5600 or for the x to drop below 300. If I were to be building a PC from scratch, I still can't game with decent settings on intel, so it just isn't that big of an upgrade.To me your argument only makes sense if a person doesnt have any computer or phone, is knowledgeable about PC hardware, physically walks into a store to buy computer parts, and a dgpu is unavailable. It is possible if say someone's house burned down, but it's very unlikely. Especially because buying a phone is probably more important than a computer, meaning the person can already have a basic functioning system.
romrunning - Saturday, March 6, 2021 - link
Well, if you're waiting for the 5600X to come in at $299, then it's there for you. $299 is the list price, and there are retailers who stick to the list price when it comes back in stock. I got a 5800X from Microcenter for list, and they had the 5600X for list as well.Spunjji - Saturday, March 6, 2021 - link
Not very well. It's only a third of the Tiger Lake iGPU, which itself performs serviceably well in optimised games when it has enough TDP headroom and LPDDR4X. This will be unspectacular, albeit still better than the junk bolted onto Sky/Kaby/Coffee/Comet Lake.yeeeeman - Saturday, March 6, 2021 - link
Waste or employees work. They should have pushed alder lake earlier.maroon1 - Saturday, March 6, 2021 - link
Or they should have just released 8 core tiger lake for desktop28w TDP tiger lake already beat rocket lake in singe-thread cinebench.
8 cores tiger lake with enough 125w should easily smoke rocket lake in both single-thread and multi-thread
intel kept 10nm for laptops only because they don't have yield for both laptop and desktop
Otritus - Saturday, March 6, 2021 - link
There are 2 problems for 8 core tiger lake to be released to desktop. One is a lack of available volume because Intel's one advantage over AMD is availability (I assume that Alder Lake is getting much of the volume of 10nm especially with its new iteration ESF). Two, tiger lake probably struggles to scale performance with power beyond that point. If Intel's 4 core tiger lake can have a tdp of 35w and scale up to 65w, then intels 8 core tiger lake can probably have a tdp of 65w and scale to 125w. TDP no longer implies the maximum power draw of a processor, and rocket lake pulling 250-300 watts can probably clock higher. Since IPC is the same, rocket lake wins on performance. Intel obviously no longer cares about efficiency because they know that they can't win on that metric.Hifihedgehog - Saturday, March 6, 2021 - link
Ian, are these two statements for the same thing or intentionally worded different: "Intel is promoting that the new Cypress Cove core offers ‘up to a +19%’ instruction per clock (IPC) generational improvement," which you mention on page 1; and "Intel is promoting a +19% average performance gain," which you mention on the last page?If so, those statements don't mean the same thing. One statement refers to a best case scenario and the other refers to an average. This puts them at odds with each unless Intel made two separate claims both involving 19% with one "up to" and and the other "average." If so, that is pretty slimy, misleading business practice on their part because that would mean the latter has failed to live up to their advertised promises.
maroon1 - Saturday, March 6, 2021 - link
They said up to 19% in one of the intel slides. Google it.Hifihedgehog - Saturday, March 6, 2021 - link
I know that. My point is it looks like it can’t be one or the other.Spunjji - Saturday, March 6, 2021 - link
Fantastic work. An absolute coup for your page, and especially important in an era where manufacturers are trying to secure presales on products that haven't even been tested yet; especially when Intel seem to be so keen on being misleading (19% IPC) or even straight-up lying (better performance in games!) about their upcoming products.On a personal note, I *finally* the data I was looking for to put a damper on the fanboy agitation that had been hailing the Return of the Gaming King. Perhaps the 11900K will do better at whatever silly price it launches at.
yetanotherhuman - Saturday, March 6, 2021 - link
Well done AnandTech and Dr. Cutress, an excellent scoop. A shite processor, but a great review! It feels like the AnandTech of old. Glad to see it.hansip87 - Saturday, March 6, 2021 - link
CPU aside, the integrated GPU should be reviewed too.. it's hard to buy reasonably priced GPU these days.. might be the new Xe- GPU can serve gamer at low setting better than the predecessor.Spunjji - Saturday, March 6, 2021 - link
Better than the predecessor? Sure! Well enough to be useful for a gamer? No, not at all. You could buy a second or third-hand GTX 750Ti on eBay and get better performance.hansip87 - Saturday, March 6, 2021 - link
How can you be sure if there's no test? That's why those numbers are needed for a conclusion to be made. If Tiger Lake can do reasonable gaming why can't this do the same?Slash3 - Saturday, March 6, 2021 - link
Tiger Lake has a 96CU Xe-LP core. Rocket Lake for the i5-11500 and up use a 36CU core - it is significantly slower than the mobile part.Spunjji - Monday, March 8, 2021 - link
Precisely this. You'd also need to budget for a board and RAM that will support running DDR4 above 4266Mhz to get those 32EUs running at similar performance levels to the ones in the mobile SKUs.TL;DR: even if clocks are higher on desktop, Xe on Rocket Lake can only have (at best) ~50% the performance of Xe on Tiger Lake - so, slower than the Vega 8 on the 4750G.
heickelrrx - Saturday, March 6, 2021 - link
It's seems 11700k is the best Gaming CPU 2021Because it has Xe iGPU, and since this is 2021 Dedicated GPU do not exist
Thank you Intel
Spunjji - Saturday, March 6, 2021 - link
That would still be the AMD 4750G, by that metric.MDD1963 - Saturday, March 6, 2021 - link
Intel had best hope that some last minute chipset /driver/Win10 updates perhaps change these results; gaming FPS being less compared to 9900K (from 2+ years ago!) and 10900K ago is downright pathetic....Cooe - Saturday, March 6, 2021 - link
I'll be totally honest here, I didn't think Intel would reach "Bulldozer status" QUITE this fast... O_oFakThisShttyGame - Saturday, March 6, 2021 - link
An interestingly boring product. But still boring.Leeea - Saturday, March 6, 2021 - link
So much sacrificed just to keep AVX 512.Intel just needs to toss avx512 on the consumer line up. Use that die space for something useful.
Hifihedgehog - Saturday, March 6, 2021 - link
Linus Torvalds was right.San Pedro - Saturday, March 6, 2021 - link
I feel like this review might have just killed interest in 11th gen Intel desktop parts. Maybe there was a reason the retailers were trying to sell these before the official date.Definitely nice to have a review out long before an actual release to give people time to think things over before buying.... Well except for those who have already bought it in the EU.
Lukeforce - Saturday, March 6, 2021 - link
So it's basically a 10700k, which is basically a 9900k.Old habits die hard at Intel I guess...
5080 - Saturday, March 6, 2021 - link
Why does Intel even bother releasing this "new" architecture. Intel should keep the previous one until they have a worthy competitor to show off.ThereSheGoes - Saturday, March 6, 2021 - link
It seems that the power graph has been changed. Is AT altering the article without making a note of that?ltkAlpha - Saturday, March 6, 2021 - link
Well, there goes Ian's relationship with Intel...Qasar - Saturday, March 6, 2021 - link
how so ?Hifihedgehog - Saturday, March 6, 2021 - link
RIP Intelschizoide - Saturday, March 6, 2021 - link
5800X and 5600X CPUs are reasonable to find these days, it's only 5900X/5950X that are extremely difficult. There's no reason at all to buy this stinker Intel CPU.Smell This - Saturday, March 6, 2021 - link
I think 'they' (AMD) is running through their AMD Ryzen 9 3900X/3950X Matisse inventory --- currently $449/$679 at my local MC (versus $549/$799). There is still incredible value in the Zen2 chips ...
qwertymac93 - Saturday, March 6, 2021 - link
The improvements seem less impressive when you realize this is the top dog of the product line, being compared to the second-best of Intel's last gen and third-best of AMD's. Granted, we don't know the price so the value may still be a good improvement over what we had. There is just some perspective missing here as 8 cores is only mid-range for the competition.Still, AVX-512 can be a game-chnager for some(if they have the cooling to match).
Hifihedgehog - Saturday, March 6, 2021 - link
> “ if they have the cooling to match”Ian used a 4 pound full copper beast level heatsink and blaring 170 CFM fan. I have used 100 CFM Deltas, the kind you find in servers, and those sound like jet engines taking off at the runway (hence why they sometimes jokingly called Delta airlines) and they annoy the heck out of PC enthusiasts. Good luck with that.
Gurg - Saturday, March 6, 2021 - link
If there isn't room on the CPU itself at 14nm, the answer would appear to be shifting whatever processes/functions that could be moved to the motherboard side of the CPU interface where there is more room. For example if the motherboard has the PCIe 4 connectors for discrete graphics card why can't it also include a chip for iGPU rather than having it on CPU.qwertymac93 - Saturday, March 6, 2021 - link
Because then it'd need its own memory controller, or hog a huge amount of bandwidth/power transferring all that data to and from the CPU's.The IGPU block does more than just 3d graphics so they can't just drop it entirely.
abufrejoval - Saturday, March 6, 2021 - link
I wonder how they sell F type CPUs then (iGPU disabled)...They could easily fit two more cores into 32EU iGPU space, but they they'd sell fewer Xeons.
The fact that AMD sold cores on die area Intel is giving away for free (they basically charge $5 for an iGPU) has been the main appeal of Zen (except APUs) from the start.
Gigaplex - Sunday, March 7, 2021 - link
They only disable the parts that aren't required.Qasar - Sunday, March 7, 2021 - link
abufrejoval the times i have seen a intel cpu with, and with out an igp, for sale there was no price differenceSpunjji - Monday, March 8, 2021 - link
"die area Intel is giving away for free (they basically charge $5 for an iGPU)"Trying to fathom the cost of Intel's iGPU by seeing what they charge for a SKU with it disabled (but still physically there in silicon) is a fool's errand.
Intel don't give anything away for free - their margins didn't go down after they put their iGPU on-die.
IUU - Saturday, March 6, 2021 - link
Whatever the advantages of the new microarchitecture, Intel managed to negate them all, by trying to sell overclocked products. Once upon a time when you overclocked a chip Intel would void your warranty. Sometimes it would directly ban overclocking or distinguish its products between these that can be overclocked and these that can't be.I would buy a true 125W TDP chip, with a new microarchitecture, even knowing it is slower, with a reasonable price of course. I would not buy overclocked chips at any price. If Intel wants to command prices comparable with AMD, they need to offer chips with comparable performance per watt power.
It would be less embarrassing for them to sell a 70% slower product with the right power than an equal regarding speed chip with twice the power draw. In my eyes of course. There are people who only care about speed , no matter the power or the money.
You can have a fast porsche, or a fast ferrari, or a fast koenigsegg.
And you can have a fast dragster.
Not the same thing.
Gigaplex - Sunday, March 7, 2021 - link
Define overclocked. It really just means clocking higher than the manufacturers specifications. If Intel is defining the clock speed it's not overclocking.Techie2 - Saturday, March 6, 2021 - link
While AMD CPUs may be difficult to find at suggested retail prices, it's FAR better to buy the AMD CPU than a POS back ported overheating Intel CPU if you can find one.For the record I have documented that several sources including Newegg have been selling the Ryzen 5 5600x @ $299. as soon as they receive stock which appears to be weekly or more frequently. Newegg advertises the $384 CPU price and (5) offers. When you click on the item you see the $299. Newegg shipped item when they are actually in stock for a short period of time. THAT is when people should buy and not support scalpers with inflated prices. It's worth checking Newegg frequently as they are not advertising the $299. price much with so many scalpers on their network.
In addition for anyone near a Micro Center they are only selling in-store but they have had a continuous supply of AMD 5000 CPUs for some time and at the correct price. This should be the first choice for anyone located near a Micro Center store because there are no games or price gouging involved.
zodiacfml - Saturday, March 6, 2021 - link
shocking. Intel allowed some retailers to sell the CPU as the shortage in AMD is severe.jayjr1105 - Saturday, March 6, 2021 - link
5600X 5800X are becoming readily available recentlyHifihedgehog - Saturday, March 6, 2021 - link
This.Antares8001 - Saturday, March 6, 2021 - link
So what it the point of Rocket Lake?The fact that it is only available with a maximum of 8 cores makes it pretty undesirable for most workstation tasks, compared to competition from AMD and even themselves. Also the fact that they seemingly sacrificed their architectural advantage for games to a point where it gets outperformed comet lake makes it seem like a chip that is completely dead on arrival with no viable market whatsoever.
As it is it does everything worse than both zen 3 and comet lake while running a lot hotter and consuming more power.
I mean great that intel managed to backport it to 14nm, but what the hell was the point.
Hifihedgehog - Saturday, March 6, 2021 - link
They have to do an annual refresh or release to appeal to their investors. That's literally it. Honestly, Skylake+++++++++++++++ would have been a better choice because at least it would have better gaming performance than this hot mess. Now, you have idiots like Usman Pirzada at fake news WCCTech playing arm-chair apologist, saying that the peak power here is a lie since it is AVX-512 and other garbage. Well, newsflash: AVX-512 is one of a few reasons why Intel got improvements in IPC, so if you want an AVX2 to AVX2 load test, also test the 11700K with AVX-512 disabled. Intel loses some of their minor gains in synthetic benchmarks, where they are still losing to the Ryzen 5000.GeoffreyA - Sunday, March 7, 2021 - link
Leaving aside the 14 nm handicap, I reckon that Intel's engineers took Sunny Cove a bit too far, widening the out-of-order window excessively, adding more decoders, etc. One will notice that Zen 3, while raising performance considerably, was pretty conservative in its design.Hifihedgehog - Monday, March 8, 2021 - link
The reason here is the 14nm handicap. You cannot bring long lengths of transistor networks without having to counteract latency, noise, and a host of other problems that come from stretching a microarchitecture intended for shorter physical paths. As an extreme example, that is why Zen 3 will never be backported to a 1 micron process. ;)GeoffreyA - Monday, March 8, 2021 - link
Would be interesting to see Sunny Cove fabricated on TSMC's 7 nm. I always wanted to see such a comparison, with Zen 3, taking the process out of the picture.jayjr1105 - Saturday, March 6, 2021 - link
Intel apologists out in full force. What a joke. I can't believe I invested in the stock of Intel recently thinking they would turn it around. Can't wait for the GPU dept to underwhelm next.Makste - Saturday, March 6, 2021 - link
Tech journalism at its peak. It is in such a way that even a tech illiterate can easily understand what is being explained. Thanks for the effort Ian and Anandtech.I previously had categorized RL as mostly optimised for gaming. But now even with the peak power draw when using AVX 512 instructions leaves me without an immediate use recommendation.
outsideloop - Saturday, March 6, 2021 - link
Intel has been forced to release a Bulldozer. All AMD and TSMC need to do now is steadily increase supply in 2021. If Alder Lake is anything less than stupendous, Intel is (insert negative term).outsideloop - Saturday, March 6, 2021 - link
I don't know if you realize this Ian, but your article may have contributed to stabilizing AMD stock yesterday. There had been so much "Intel is returning to dominance!" gobletygook in the world investment community, and this article will do a lot to bring everyone back to earth and the reality of AMD technological and product superiority (for at least 2021).Ryan Smith - Saturday, March 6, 2021 - link
"your article may have contributed to stabilizing AMD stock yesterday. "Although we'd like to take credit for it, this seems unlikely. We didn't publish the article until after the market had closed for the day.
eva02langley - Monday, March 8, 2021 - link
Absolutely not, there is absolutely no connection with the stock price of AMD and this review. The stock stabilized because AMD stock value is undervalued at 75$. The real stock price should be closer to 105$, but the selloff of semiconductors companies last week was general. It happens because some people think the economy is going to restart the good old ways with vaccination... they are all wrong.Scyndek - Saturday, March 6, 2021 - link
lol I said it was bullshit that userbenchmark had this listed at #1 for processors and said it should almost be considered illegal and what they are doing is manipulating public opinion and posted this review stating check for yourself.Of course I was promptly blacklisted and they banned my IP Range for speaking the truth.
Scyndek - Saturday, March 6, 2021 - link
* should note I said all this in the comment section for the processor itself, the comment of course is deleted and barely was there for 5 minutes before they blacklisted my IP range and removed it *Makste - Saturday, March 6, 2021 - link
I stopped reading user benchmark ever since last year (the only year I had started reading on the site) because it became very apparent that they were biased, it is easy to see in their product descriptions.Spunjji - Monday, March 8, 2021 - link
Most of their content would be worthless even if their tests weren't biased as hell - there's no way to tell how people's systems are configured, what background software they're running, etc.JayNor - Saturday, March 6, 2021 - link
"Additional improvements over Comet Lake include AVX512 units, support for 20 PCIe 4.0 lanes,"Along with the avx512, Rocket Lake is also including the dlboost int8 operations ... so quantized ai inference performance can be pushed to around 4x vs Comet Lake, minus any avx512 related clock limiting.
Qasar - Saturday, March 6, 2021 - link
and it looks like it doesnt do any good, this cpu is pretty much a jokeExotica - Saturday, March 6, 2021 - link
Is it the CPU or is it the motherboard and/or early BIOS that’s causing the performance regression? Perhaps consider performing tests on different z490 and z590 boards to see if it is the chip that has regressed or if some motherboards have better performance.Also, many people use AIOs so what is the temperature differential when using a 280 or 360 rad?
Exotica - Saturday, March 6, 2021 - link
Not fair to call this a “review” especially since it’s not been announced yet, and the bioses are still in beta. For example, on my gigabyte vision d z490, the beta f20 bios with rocket lake support is a flaming mess. With boot loop issues and memory training issues. Completely unusable with Comet Lake. Sleep issues as well (computer wakes up immediately after sleep). The earlier non-rocket lake BIOS versions perform much better.So I will wait to reserve judgement on performance until after these early beta BIOS bugs are worked out.
emperoralku - Saturday, March 6, 2021 - link
You are like the amd fanbois who were hoping that magic drivers would fix vega. This is it, face it.hfm - Saturday, March 6, 2021 - link
I'm actually more interested in the H45 notebook parts with embedded TB4 on 10nm. I use a notebook with an eGPU. The higher core counts and direct-to-cpu TB will be a big deal for me. As right now I'm relegated to 4c 28w or marginally H35 right now to get an efficient TB implementation for eGPU gaming. (No I don't want the Asus Flow X13 and it's 52dbA XG mobile)emperoralku - Saturday, March 6, 2021 - link
Bulldozer, is that you? Or are you the cpu version of vega?GeoffreyA - Saturday, March 6, 2021 - link
I think the latter. Just imagine when this thing collapses into a singularity ;)DannyH246 - Saturday, March 6, 2021 - link
What a load of crap. Hot, late, slow. The biggest joke of all is this site actually recommends you buy one “because you can get one”. I didn’t think this place could sink any lower. I was wrong.Spunjji - Monday, March 8, 2021 - link
That's an odd interpretation of what was actually saidRauli Trump - Saturday, March 6, 2021 - link
what ram subtimings where used on 11700klyssword - Saturday, March 6, 2021 - link
Wow, finally upgraded gpu from gtx 980 for gaming tests, good job anandtech!WaltC - Saturday, March 6, 2021 - link
One thing I think that is of interest to people is what security holes have been filled with these newer Intel CPUs, if any, and were these benchmarks run with OS and firmware security mitigations in place? My guess would be: no. It's not really an issue for AMD as that story is well known--1 or possibly 2 mitigations in place--both were handled in firmware for Zen2, probably neither is a problem for Zen 3. Intel, otoh, has been cursed with dozens of security holes patched with Windows microcode patching & firmware patching for quite a while. An update on the situation would have been thoughtful. I don't think Intel believes it is ready to compete just yet.Spunjji - Monday, March 8, 2021 - link
AFAIK this core design is hardened at the hardware level against Spectre and Meltdown.GeoffreyA - Saturday, March 6, 2021 - link
Thanks for the excellent work, Ian. While it hasn't displaced Ryzen, I commend Intel for the effort that Rocket Lake is. Zen 3's architecture is a lot lighter, which suggests that Intel may have to curtail theirs in the next round. Starting from scratch might be of use but will take a few iterations before catching up with Zen.iGigaflop - Saturday, March 6, 2021 - link
Well I finally got my 5950x a couple days ago but I did order it December 15th on special order so it took awhile I think I payed $850ish for it with tax. I was thinking 11th gen intel was gonna be better than this hell seems like 10th gen the smarter bet and going with a 10900k or one of the close versions if you wanted intel. But overall amd is killing intel in every section of the market. I couldn’t imagine this about 5 years ago especially in the laptop market where 7nm efficiency really helps. Now if they can get it together with their gpu’s, the 6000 series is great and all but if you gave most people the choice i still think they would pick a 3080 over a 6800xt. Ray tracing performance is nice but it’s the dlss really helping nvidia. I have a strix 2080ti with a 9900ks right now as my main gaming rig and a 3900x radeon vii for my work/gaming rig. I’m gonna put the ti with the 5950x but I wanted to find a 3080/3090 for the build but I have gave up for now. Hell I need to sell a bunch of my gpu’s while the market is crazy. I have a strix 2080, nitro 5700xt se, msi gaming x 1080, and a couple more thats just collecting dust.watzupken - Saturday, March 6, 2021 - link
Interesting results, but I wasn't surprised by the heat and power consumption to be honest. I was expecting Intel to reclaim single core performance, but at least I don't see it in this independent review and also a credible one. I think I kind of understand why Intel had to rush Alder Lake out on the same year now. While Intel's 14nm has bought them a long time, it is really on its last leg, particularly when AMD and ARM are utilizing more advanced nodes.Santoval - Saturday, March 6, 2021 - link
Wow, if this thing is *that* hot (as was expected) then the 11900K will probably double as an egg fryer, no matter how tightly it was binned. Those 292 W and 104 °C AVX-512 peaks were scary, really. I wonder how Zen 4 will fare with AVX-512 code. Will TSMC's 5nm process node help it keep power and thermals at reasonable levels?Hifihedgehog - Monday, March 8, 2021 - link
Under the absolute worst scenario, you might see AVX-512 on Zen 4 to hit the power level of AVX-2 on Intel. And that’s worst case, mind. Best case, the same power level as Zen 3 with AVX-2. Realistically, somewhere between those two extremes on that spectrum, leaning more towards Zen 3’s maximum power load.dirtyvu - Saturday, March 6, 2021 - link
the sad state of Intel... I'm still waiting for the 5900X to be available.Lord 666 - Sunday, March 7, 2021 - link
Would have preferred to see the Apple M1 cpu included in the comparisons.Klimax - Sunday, March 7, 2021 - link
Two things:Latency increase is most likely down to size of CPU.
Second, why are you using average FPS in games???? It is the worst part of reviews here, because it completely obscures any useful info.
Remember "Inside a second" by Techreport. And with modern CPUs and high variability of clocks, that methodology is bare minimum for even basic comparison between CPUs.
ANOVA would be a good start.
MDD1963 - Sunday, March 7, 2021 - link
"Second, why are you using average FPS in games?"Earth shattering concept reveal: faster processors often produce more min/average/max fps in games...; by seeing the 11700K producing lower results that the last two generations, folks are...disappointed, to say the least.
Zizy - Monday, March 8, 2021 - link
You have 95 percentile results here, presented as FPS. This is standard fare these days: You get a number X ms where 95% of frames have lower or equal time vs previous frame than that. Then you convert those X ms frame time numbers to the more typical FPS (=1000/X).Yeah, there are no percentile graphs that illustrate some issues with CPUs or GPUs, but outside of extremely few cases those were mostly boring linear increase in frame time to ~99% then a sizeable spike for the last ~1% and that's it. Dragging a line between average (50%) and 95% and extrapolating to 99% would give you essentially all the same info.
FatalError - Sunday, March 7, 2021 - link
Comparing gaming benchmarks to my i7-5775c (reminder base 3.3 turbo 3.7) with DDR3 at to my surprise the i7-5775c wins a few benchmarks and loses to some. I get it more cores and better IO but why would someone that does not do heavy professional workloads buy this (and to some extent anything that AMD bought out).A CPU released in 2015.
FatalError - Sunday, March 7, 2021 - link
i7-11700k | i7-5775c(m-7) Strange Brigade DX12 - 1080p Ultra - Average FPS
219.6 | 225.9
Civilization VI - 4K Min - Average FPS
94.3 | 113.1
Gears Tactics - 4K Low - Average FPS
49.2 | 53
sure enough, these are handpicked and the 11700k wins some benchmarks with a big margin, but we have here 4 real cores less and 1.3 GHz less turbo and a mere DDR3 memory interface. I wonder if Intel went the wrong way by ending the L4 monster caches. At least they would have retained the gaming crown if they would introduce an "extreme" version with a decent-sized L4 low latency cache.
Netmsm - Sunday, March 7, 2021 - link
Hi Ian;I appreciate your work and also I have a suggestion hoping to be considered.
I think it's better not to mention processors by their formal or nominal power in benches. for us, it can be much more justifiable when we know how much a processor consumes power in each test separately.
I know there's a separate test for power consumption but, as you know, in such an old-fashioned yet common way, the efficacy of each processor for each test is not clear. and I'm afraid, it undermines your work's articulacy.
regards =)
Polacott - Sunday, March 7, 2021 - link
Charlie Demerjian was right long time ago, when he stated this was going to be a fiasco.Seems like a desperate move from intel, more than lift off, this rocket is hitting the ground.
eastcoast_pete - Sunday, March 7, 2021 - link
Thanks Ian! I actually applaud the decision to go ahead and put a legally purchased CPU through the paces before the official embargo date for the official, free-for-testing samples; you were very up front how and where you got your test candidate. And, once an authorized reseller sells them, they're "public" and fair game.These performance tests are also a unique opportunity to see by just how much the official sample Rocket Lake with the then-valid BIOS firmware might differ from this store-bought one.
Actually, if Intel would want to be smart about this, they'd send you a coupon to order a RL and MoBo from any authorized retailer, to avoid accusations of a "review edition" hand-selected to be the best binning possible.
Lastly, I wish you (AT) and other review sites would occasionally cross-check the results of their review samples (sent by the manufacturer) to the same unit bought through retail. I know that's not always feasible (costs $$$) , but doing so every once in while might help keep the manufacturers honest and assure us, their customers.
pman6 - Sunday, March 7, 2021 - link
hey, it comes with a free space heater.nice
pman6 - Sunday, March 7, 2021 - link
why wasn't there a comparison of the XE integrated graphics?Ryan Smith - Sunday, March 7, 2021 - link
Partially for timeliness reasons. Partially because we've already seen Xe in Tiger Lake. Rocket Lake's Xe-LP implementation is much slower than the mobile chips, since it has only one-third the number of EUs. So while we'll collect that data eventually, it's nothing terribly exciting.terroradagio - Sunday, March 7, 2021 - link
Intel Rocket Lake Core i7-11700K vs Core i9-10900K CPU Gaming Benchmarks Leaked – Reportedly Faster Than Comet Lake With New BIOShttps://wccftech.com/intel-rocket-lake-core-i7-117...
eastcoast_pete - Sunday, March 7, 2021 - link
The boost clock frequency in that WCCF article for the 11700K is especially interesting; according to the table in it, the boost clock goes up to 470 GHz (: Now, that's a clock speed worth writing about!Hifihedgehog - Monday, March 8, 2021 - link
WCCFTech has been Intel’s lap dog ever since Zen 2 hit the scene with article after article with a spin in Intel’s favor and I would not be the least bit surprised if they are just like UserBenchmark. Their notorious Usman Pirzada in particular harped on Ian for showing peak power for Rocket Lake, claiming it was disingenuous and misleading since AVX-512 is the reason and that puts the processors on unequal footing. Well, Mr. Pirzada, you forget. AVX-512 also gives a performance uplift so you better also ask to note all the benchmarks where it is engaged so we are playing fair here. If you truly want to level the playing field and throw out AVX-512, you have to throw out a key item that has contributed to an uplift in performance and turn it off in both benchmarks and power tests alike. Either both ways or no ways, sonny Jim.Spunjji - Monday, March 8, 2021 - link
"So coming to the benchmarks, only three titles were tested which include Crysis Remastered, Star Wars: Jedi Fallen Order, and Cyberpunk 2077. The test settings and resolution are not mentioned."It's bad enough that they only showed tests from games that Anandtech don't have in their test suite - so we can't compare directly - but it's even worse that we have no data about the settings used, so nobody else could possibly compare directly, either.
Honestly, at this point I consider myself to have a good indicator of final performance, and will draw final conclusions when the embargo lifts.
shabby - Sunday, March 7, 2021 - link
I hear intel will be bringing back btx to cope with the 300w+ 11900k...Doug1820 - Sunday, March 7, 2021 - link
Rocket Lake is looking more and more like Netburst 2.0.Oxford Guy - Sunday, March 7, 2021 - link
Someone literally wrote ‘I commend Intel for this release’. Commend? Perhaps the word originally sought was condemn?This product exists due to inadequate competition. It is a gift from monopolization. While the mantra of the corporation is ‘sell less for more’, it’s adequate competition that’s supposed to be its saving grace. That hasn’t been the case in many areas of tech for a long time.
Not only is AMD not enough competition, it took this long for the company to finally beat Intel badly. If Intel had managed its business better it would still be competitive.
And, to top things off, just like in GPUs (dire lack of competition) it’s extremely profitable to fail. Nvidia is failing to meet gamer demand, for various reasons that come down to inadequate competition. Despite that, it is taking record money. Intel is profitable despite failing in the consumer desktop CPU market.
Allison Kilkenny joked that America is special because one can ‘fail upward’. But, really, tech has a huge problem with inadequate competition — a global one.
People see brief glimmers of competition and mistake it for adequate competition. This is not the first time AMD had the better CPU design and we all know how that turned out.
Being held captive by one and 1/2 companies (the typical tech competition ratio) isn’t at all close to approaching ideal.
If huge profitability while failing very badly to meet demand is the recipe for a good state of business...
GeoffreyA - Monday, March 8, 2021 - link
I believe that was me. "While it hasn't displaced Ryzen, I commend Intel for the effort that Rocket Lake is."I should have worded it better, I agree, but what I meant was, instead of just releasing another Skylake refresh, they did the unthinkable, in an attempt to release something: porting Sunny Cove to their legacy node. "We will fail, but we'll try anyway, futile though our efforts may be." Their being the underdog at present makes me feel a bit sorry for them, but then again, when we consider Intel's bank account, we realise they don't need our sympathy.
For my part, I am not an Intel shill or proponent. In fact, from my teenage years I've had a distaste towards them, and as I see it, Rocket Lake is a disappointment: still behind Zen 3, using far more power, slower graphics than Tiger Lake. There's practically no good in it. For the elusive Sunny Cove, I had expected a lot more. Despite that, I do commend them for making an attempt, something in life that's more important than winning.
GeoffreyA - Monday, March 8, 2021 - link
As for the business/economic side of the matter, I don't understand these things too much to comment, but from that side, Intel has always been wretched; and all these companies are only out to make profit. We've even learned that those who offer something for nothing---it comes at a cost (Google, Facebook, etc.).Spunjji - Monday, March 8, 2021 - link
100%Oxford Guy - Tuesday, March 9, 2021 - link
'all these companies are only out to make profit'Which means consumers need to get their heads out of the sand and fight for value for their dollars.
Spunjji - Monday, March 8, 2021 - link
Intel are very much not the underdog - they're still the 800 lb gorilla, they've just got themselves trapped inside a cage of their own making. Tiger Lake is an example of the fact that they're still capable of first-tier core designs; they're just not (currently) capable of bringing them to market on a competitive process.Hifihedgehog - Tuesday, March 9, 2021 - link
They are 800-lbs on a fixed rations in self-imposed solitary confinement. Watch them slim down in a jiffy to as skinny as a rail if Alder Lake fails to impress. It was a slippery slope with Bulldozer. The only issue with Intel is the sheer amount of corporate waste. If they go down, others will be ready to swoop in and pick up the layoff broken pieces—pieces, once surrendered, that never again be recovered without a fight. Intel has never been fully challenged but they just might fall into Bulldozer-like obscurity this time around.GeoffreyA - Tuesday, March 9, 2021 - link
"self-imposed solitary confinement"Hard to believe how the tables have turned in a matter of four years, and worse for Intel, this isn't the slack, "we've done a good job, let's rest" AMD of the K8 days. This time round, they're relentlessly executing, with a vigour I don't remember them possessing. Whether it's Lisa or just the pain they went through in the Bulldozer days, I don't know, but it's magnificent to watch.
GeoffreyA - Tuesday, March 9, 2021 - link
Cage of their own making: I like that phrase. Well, I think for Intel, the best path is to design a microarchitecture from scratch, which I'm sure they're doing. Possibly Intel Israel. Sunny Cove, descending all the way from P6, is not holding up any more.Zim - Sunday, March 7, 2021 - link
Time to be me some more AMD stock.alhopper - Sunday, March 7, 2021 - link
What an awesome deep-dive technical expose/review of this processor. No stone left un-turned - incredibly detailed investigation and easily understood explanations. With this level of detail every reader can make up their own mind about the pros/cons of this product - there's no need for me to offer an opinion.Many thanks for this well written treasure trove of technical information and the graphics are wonderful. Kudos to the entire team and keep up the Good Work (TM) !!
poohbear - Sunday, March 7, 2021 - link
Wow how the tables have turned. I remember for AMD's Bulldozer review, which was supposed to be AMD's answer to Core, Bulldozer was similar to Rocket Lake...it consumed MORE power and was SLOWER than the competition. Well, here's to Intel making a comeback with Alder Lake, but 2021 clearly belongs to AMD.Hifihedgehog - Monday, March 8, 2021 - link
Alder Lake, just confirmed in a lineup leak to use a heavy amount of little or low power cores, first has to ensure that they get heterogenous core architecture working in the Windows scheduler. The biggest roadblock Microsoft claimed to have already fixed it with Lakefield but even after they did and additionally Cinebench, per Intel’s special emergency request, to recompile a special update to R20 after that, Lakefield was an absolute failure in performance compared to a Y-class (or M notated) Core product. Personally, I have little faith in Microsoft and I also do not expect software companies recompiling every last piece of software made in the last decade either just to accommodate a radical new design all for Intel’s sake. Developers would much rather assume a simple, vanilla core architecture than an unbalanced one with little and big cores and therein lies a huge risk factor for Intel. I foresee a huge misfire here because Microsoft isn’t exactly a good company to bet on refixing what they claimed they have already fixed with Lakefield.GeoffreyA - Monday, March 8, 2021 - link
I'm sceptical of this whole fast/slow core approach myself and believe it's a mistake. They will learn the hard way. Perhaps it works for Apple, I don't know, but Intel and AMD are better of spending their efforts in cutting down power in their ordinary cores dynamically.Hifihedgehog - Monday, March 8, 2021 - link
It works for Apple because they have a iron-tight grasp on the full stack which requires developers to use their latest compilation and development tools to get their apps to the App Store and ultimately in users’ hands. You can run pretty much anything under the sun for 2 decades plus on Windows and without a recompile, you will never utilize their cores correctly. The Windows scheduler as it stands is garbage tier compared to Linux with a plain vanilla homogenous core design so I expect very little from Microsoft unless they actually redesign their scheduler from the ground up. Don’t hold your breath on that one.GeoffreyA - Monday, March 8, 2021 - link
The whole idea smacks of Bulldozer, and that means disaster.blppt - Monday, March 8, 2021 - link
Only if their single thread performance sucks (which from all I've seen, it won't)---Bulldozer's problem was being behind in manufacturing process and requiring a highly-threaded application to shine.Hifihedgehog - Tuesday, March 9, 2021 - link
Well, the problem then is how smartly it assigns threads to the right cores and the Windows scheduler still has serious problems with that to this very day. If it doesn’t do that right, it could appear like an overclocked Core i3 in comparison to Ryzen and that is where the Bulldozer analogy is fitting. AMD banked on their unique topology getting OS and development optimization down the road—FineWine to coin a phrase—but that never materialized.blppt - Saturday, March 13, 2021 - link
Not really---the Bulldozer design had far more problems than just the inadequate Windows scheduler---while "optimized" linux kernels gave better performance, the true issue with Bulldozer was that you had 8 relatively weak cores versus 4 strong cores from Intel, and at the time, coders were still struggling to optimize for anything over 2 cores.In a sense, the designers of Bulldozer REALLY misread the timeline of highly multithreaded coding taking over the market. Heck, even if it was released now alongside the equivalent Core 2 quads, it would still stink for the majority of users, because no game gets any significant advantage of 4+ physical cores even today---and most games still value high single thread/core performance.
Oxford Guy - Sunday, March 28, 2021 - link
'the true issue with Bulldozer was that you had 8 relatively weak cores versus 4 strong cores from Intel, and at the time, coders were still struggling to optimize for anything over 2 cores. In a sense, the designers of Bulldozer REALLY misread the timeline of highly multithreaded coding taking over the market.'My guess is that AMD designed Bulldozer for the enterprise market and didn't want to invest in an additional design more suited to the consumer desktop space. Instead, its additional design priority was the console scam (Jaguar). While that was a good move for AMD it wasn't beneficial for consumers, as consoles are a parasitic redundancy.
One thing many ignore is that Piledriver supercomputers occupied quite high spots in the world performance lists. Mostly that was due to the majority of their work being done by the GPUs, though. Even the original Bulldozer, in Opteron branding, was used in some.
The cheapness of Piledriver chips was also probably a factor in the adoption of the design for supercomputers. Turn down the voltage/wattage so that you're in the efficient part of the improved 32nm SOI node and rely almost completely on heavily threaded code when not running GPU-specific code... and voila — you have an alternative to the monopoly-priced Intel stuff.
But, on the desktop, Piledriver was a bad joke. That's because of its very poor single-thread performance mainly. Not everything can be multi-threaded and even if it is that can mean a speed regression sometimes. The slowness of the L3, the lack of enough operations caching... the design wasn't even all that optimized for multi-thread performance — especially FPU stuff. The cores were very deeply pipelined, designed to use very high clocks. They were not efficient with avoiding bubbles and such. I read that AMD relied too heavily on automated tools due to cost sensitivity.
My vague understanding of the design is that it was narrow and deep like the Pentium 4. Why AMD tried NetBurst 2.0 is beyond me. Even for the enterprise market it's a bad move because power efficiency is important there, especially with servers (rather than supercomputers which, I think, were more tolerant of high power usage – in terms of acceptable design requirements). Even turning down the clocks/voltage to get the best efficiency from the node doesn't fix the issue of the pipelining inefficiency (although hand-tuned code used for some enterprise/scientific stuff would mask that weakness more than general-purpose consumer-grade apps would).
usiname - Monday, March 8, 2021 - link
Intel know very well that is mistake, but alder lake's big cores are not much better in term of power consumption and 10nm quality wafers so their only choice is to keep the production of max 8 core mainstream. This is bad for them, because even 3 years after AMD show 16 core mainstream they can't and as our very well known Intel they will cheat by introduce their 16 core cpu with 8 fake cores. Even more, when they introduce their "super duper" 12900k with 16 cores they will set price higher than every amd main stream and this is triple win, cheap 8 core cpu for manufacturing, better binned with higher clocks and on price of $800-1000. If you think intel trying to do something new and innovative you are wrong.GeoffreyA - Monday, March 8, 2021 - link
You may well turn out to be right about 16 cores having 8 junk ones, and knowing Intel, that's how they operate, with smoke and mirrors when they can't compete properly.Hifihedgehog - Tuesday, March 9, 2021 - link
The irony here is how Intel used to give ARM smack for having inferior single core performance while they were surpassed in multicore by the likes of Qualcomm. I believe—paraphrasing—what they would say is not all cores are created equal. Well, it looks like Intel is trying to look like they are maintaining parity when they are really just giving us mostly crappy cores that can’t perform well at all.GeoffreyA - Tuesday, March 9, 2021 - link
Also, as others have pointed out before, the nomenclature is just there to obfuscate the whole picture. Not knowing anything much about Alder Lake, I did some searching and saw that it's Golden Cove + Gracemont. Wondering what exactly GC was, I searched a little but couldn't find the answer, so I'll guess it's just Sunny Cove with a new name stuck on.Bagheera - Monday, March 8, 2021 - link
the more I read about Intel's 10nm (check relevant articles on SemiWiki and Semi accurate), the more I feel ADL is designed as a workaround for the power/thermal limitations of their 10nm process, instead of being some sort of revelation for next gen performance.Note how TGL remains Intel's only viable product on 10nm at the moment, with Ice Lake SP now two years late. I think Intel knows their 10nm may never be ready for desktop parts, so ADL is a way to have a desktop product on 10nm except not really (it's more akin to a mobile part).
It will probably do fine for gaming, but highly doubtful it will be a meaningful competition to Zen 4 for the prosumer space.
blppt - Monday, March 8, 2021 - link
Bulldozer/Piledriver were a dumb design because they relied HEAVILY on highly-threaded applications to achieve their performance. Almost none of which existed in the late 2000s when they launched. Single thread/core performance was absolutely pathetic compared to Intel's offerings at the time (Sandy Bridge and on).Zan Lynx - Sunday, March 14, 2021 - link
And yet here we are today with 8 cores in game consoles and AMD's Mantle API being the basis of both DirectX 12 and Vulkan. Multicore was always the future and it was obvious even in 2003.blppt - Monday, March 15, 2021 - link
"And yet here we are today with 8 cores in game consoles and AMD's Mantle API being the basis of both DirectX 12 and Vulkan. Multicore was always the future and it was obvious even in 200"You're missing the point. Even TODAY, few (if any) games gain an advantage of more than 4 physical cores + 4 Virtual. It is still, to this day, far more advantageous for games to have 4 strong physical cores than 8 weak ones. The latter was Bulldozer.
Never mind back in 2011/2012.
zzzxtreme - Sunday, March 7, 2021 - link
I wished you would have tested the XE graphicsFman4 - Monday, March 8, 2021 - link
Am I the only one find that OP plugged 4 RAMs on an X570 ITX motherboard?Fman4 - Monday, March 8, 2021 - link
@Dr. Ian Cutresszodiacfml - Monday, March 8, 2021 - link
bored. just here to say this is unsurprising though this strongly reminds me of the time where AMD is releasing new, well designed CPUs but two process node generations behind intel. I think AMD was 32nm and 28nm while Intel is 22 and 14nm. most comments were really harsh with AMD but I reasoned that it is simply due to the manufacturing superiority of Intelblppt - Monday, March 8, 2021 - link
Bulldozer and Piledriver are not the examples I would put up for "well designed".GeoffreyA - Tuesday, March 9, 2021 - link
Still, within that mess, AMD did a pretty good job raising Bulldozer's IPC and cutting down its power each generation. But the foundation being fatally flawed, it was hopeless. I believe it taught them a lot about cutting power and so on, and when they poured that into Zen, we saw the result. Bulldozer was a fantastic training ground, if one looks at it humorously.Oxford Guy - Tuesday, March 9, 2021 - link
No, AMD did an extremely poor job.Firstly, Bulldozer had worse IPC than Phenom. No engineers with brains release a CPU to replace the entire line while giving it worse IPC. The trap of going for high clocks was a lesson shown to the entire industry via Netburst. AMD's engineers knew all about it, yet someone at the company decided to try Netburst 2.0.
Secondly, AMD was so sloppy and lazy that Piledriver shipped with a performance regression in AVX. It was worse to use AVX than to not use it. How incredibly incompetent can the company have been? It doesn't take a high IQ to understand that one doesn't ship broken AVX.
AMD then refused to replace Piledriver until Zen came out. It tinkered half-heartedly with APU rubbish and focused on pushing junk like Jaguar.
While it's true that the extreme failure of AMD (the construction core line) is due, to a large degree, to Intel abusing its monopoly to starve AMD of customers and cash — cash it needed to do R&D, one does not release a new chip with worse IPC and then very shortly after break AVX and refuse to stop feeding that junk to customers for many years. Just tinkering with Phenom would have been better (Phenom 3).
As for the foundation claim... we have no idea how well the CMT concept could have worked out with competent engineering. Remember, they literally broke AVX in the Piledriver revision that was supposed to fix Bulldozer enough to make it sellable. Operations caching could have been stronger. The L3 cache was almost as slow as main memory. The RAM controller was weak, just like Phenom's. Etc.
We paid for Intel's monopoly and we're still paying today. Only its monopoly and the lack of adequate competition is enabling the company to be so profitable despite failing so badly. Relying on two companies (or one 1/2, when it comes to R&D money ratio and other factors) to deliver adequate competition doesn't work.
Google and Microsoft = Google owns the clearnet. Apparently, they have some sort of cooperation agreement which helps to explain why Bing has such a tiny index and such a poor-quality search.
TSMC and Samsung = Can't meet demand.
AMD and Nvidia = Nvidia keeps breaking profit records while utterly failing to meet demand. Both companies refuse to stop making their cards attractive for mining and have for a long long time. AMD refused to adequately compete beyond the lower midrange (Polaris forever, or you can buy a 'console'!) for a long time, leaving us to pay through the nose for Nvidia's prices. AMD literally competes against the PC market by pushing the console scam. Consoles are gaming PCs in disguise and they're parasitic in multiple ways, including in terms of wafer allocations. AMD's many many years of refusal to compete with Nvidia beyond the Polaris price point caused so much pent-up demand and now the company can enjoy the artificially high price points from that. It let Nvidia keep raising prices to get consumers used to that. Now that it has finally been forced to improve the 'consoles' beyond the garbage-tier Jaguar CPU it has to offer a bit more value to the PC gaming market. And so, after all these years, we have something decent that one can't buy. I can go on about this so-called competition but why bother. People will go to the most extravagant lengths to excuse the problem of lack of adequate competition — like the person who recently said it's easier to create Google's empire from scratch than it is to make a competitive GPU and sell it as a third GPU company.
There are plenty of other areas in tech with inadequate competition, too.
blppt - Tuesday, March 9, 2021 - link
"AMD then refused to replace Piledriver until Zen came out. It tinkered half-heartedly with APU rubbish and focused on pushing junk like Jaguar."To be fair, AMD had put a LOT of time, money and effort into Bulldozer/Piledriver, and were never a company with bottomless wells of cash to toss an architecture out immediately. Plus, Zen took a long time to design and finalize---thankfully, they made literally ALL the right moves in designing it, including hiring the brilliant Jim Keller.
I think if Zen had been another BD like failure, that would have been the almost the end of AMD in the cpu market (leaving them basically as ATI was) The consoles likely would have gone with Intel or ARM for their next iteration. AMD once again spent tons of money that they don't have as disposable income in designing Zen. Two failures in a row would have been disastrous.
Heck, the consoles might go with their own custom ARM design for PS6/Xbox(whatever) anyways.
GeoffreyA - Wednesday, March 10, 2021 - link
blppt. Agreed, that would have been the end of AMD.Oxford Guy - Wednesday, March 10, 2021 - link
AMD did not put a lot of resources into fixing Bulldozer.It shipped Piledriver with broken AVX and never bothered to replace Piledriver on the desktop until Zen.
Inexcusable. It shipped Steamroller and Excavator in cost-cut mode, cutting cores, cutting clocks, cutting the socket standards, and cutting cache. It used a dense library to save money by keeping the die small and used the inferior 28nm bulk process.
Pathetic in basically every respect.
blppt - Saturday, March 13, 2021 - link
They did try to at least 'ride it out' until Zen could get done, and that required smoothing out the rough edges, so they did devote some resources.BD/PD never did any better than a low-end solution for the desktop/laptop market, but they had to offer something until Zen was done.
Oxford Guy - Sunday, March 28, 2021 - link
'They did try to at least 'ride it out' until Zen could get done, and that required smoothing out the rough edges, so they did devote some resources.'Wow... watch the goal posts move.
Riding out = doing nothing. Piledriver was not improved. The entire higher-performance & supercomputer market was unchanged from Piledriver to Zen. All AMD did was ship cheap knock-off APU rubbish and console trash.
The fact that AMD succeeded with Zen is probably mostly a testament to one largely ignored feature of monopoly power: the monopolist can become so slow and inefficient that a nearly dead competitor can come back to best it. That's not symptomatic of a well-run economic system. It's a trainwreck.
AMD should have been wealthy enough to do proper R&D and bulldozer would have never happened in the first place. But, Intel was a huge abusive monopolist and everyone went right along, content to feed the problem. After AMD did Bulldozer and Piledriver the company should have been dead. If there had been adequate competition it would have been. So, ironically, AMD can thank Intel for being its only competition, for resting on its laurels because of its extreme monopolization.
GeoffreyA - Wednesday, March 10, 2021 - link
Oxford Guy. I don't remember the exact details and am running largely from memory here. Yes, I agree, Bulldozer had far lower IPC than Phenom, but, according to their belief, was supposed to restore them to the top and knock Intel down. In practice, it failed miserably and was worse even than Netburst. Credit must be given, however, for their raising Bulldozer's IPC a lot each generation (something like 20-30% if I remember right), and curtailing power. It also addressed weaknesses in K10 and surpassed K10's IPC eventually. Anyway, working against such a hopeless design surely taught them a lot; and pouring that knowledge into a classic x86 design, Zen, took it further than Skylake after just one iteration.AMD would have done better had they just persisted with K10, which wasn't that far behind Nehalem. But, perhaps we wouldn't have had Zen: it took AMD's going through the lowest depths, passing through the fire as it were, to become what they are today, leaving Intel baffled. I agree, they were truly idiotic in the last decade but no more. May it stay that way!
Concerning CMT, I don't know much about it to comment, but think Bulldozer's principal weakness came from sharing execution units---the FP units I believe and others---between modules. Zen kept each core separate and gave it full (and weighty) resources, along with a micro-op cache and other improvements. As for Jaguar, it may be junk from a desktop point of view, yes, but was excellent in its domain and left Atom in the dust.
Oxford Guy - Sunday, March 28, 2021 - link
'Credit must be given, however, for their raising Bulldozer's IPC a lot each generation (something like 20-30% if I remember right), and curtailing power.'Piledriver was a small IPC improvement and regressed in AVX. Piledriver's AVX was so extremely poor that it was faster to not use it. Piledriver was a massive power hog. The 32nm SOI process node, according to 'TheStilt' was improved over time which is probably the main source of power efficiency improvement in Piledriver versus Bulldozer. I do not recall the IPC improvement of Piledriver over Bulldozer but it was nothing close to 20% I think. Instead, it merely made it possible to raise clocks further, along with the aforementioned node improvement. And, 'TheStilt' said the node got better after Piledriver's first generation. The 'E' parts, for instance, were quite a lot improved in leakage — but the whole line (other than the 9000 series which he said should have been sent to the scrapper) improved in leakage. What didn't improve, sadly, is the bad Piledriver design. AMD never bothered to fix it.
While Piledriver, when clocked high (like 4.7 GHz) could be relevant against Sandy in multi-thread (including well-threaded games like Desert of Kharak) it was extremely pitiful in single-thread. And, it sucked down boatloads of power to get to 4.7, even with the best-leakage chips.
And, going back to your 20–30% claim. Steamroller, which was considered a serious disappointment, featured only 4 of the CMT quasi cores, not 8. Excavator cut things in cache land even further. Both were cost-cutting parts, not performance improvements. Piledriver killed both of them simply by turning up the clocks high. The multi-thread performance of Steamroller and Excavator was not competitive because of the lack of cache, lack of cores, and lack of clock. Single-thread was a bit improved but, again, the only thing one could really do was blast current through Piledriver. It was a disgusting situation due to the single-threaded performance, which was unacceptable in 2012 and an abomination for the later years AMD kept peddling Piledriver in.
The only credit AMD deserves for the construction core period is not going out of business, despite trying so hard to do that.
GeoffreyA - Sunday, March 28, 2021 - link
Oxford Guy, while I respect your view, I do not agree with it, and still stand by my statement that AMD deserves credit for improving Bulldozer and executing yearly. Agreed, my 20-30% claim was not sober but I just meant it as a recollection and did qualify my statement.I don't think it's fair to put AMD down for embarking on Bulldozer. When they set out, quite likely they thought it was going to go further than the aging Phenom/K10 design, and the fact is, while falling behind in IPC compared with K10, it improved on a lot of points and laid the foundation. Its chief weakness was the idea of sharing resources, like the fetch, decode, and FP units, as well as going for a deeper pipeline. (The difference from Netburst is that Bulldozer was decently wide.)
Piledriver refined the foundation, raising IPC and adding a perceptron branch predictor, still used in Zen by the way, and I believe finally surpassed K10's IPC (and that of Llano). While being made on the same 32 nm process, it dropped power by switching to hard-edge flip flops, which took some work to put in. They used that lowered power to raise clock speeds, bringing power to the same level as Bulldozer. And Trinity, the Piledriver APU, surpassed Llano. I need to learn more about Steamroller and Excavator before I comment, but note in passing that SR improved the architecture again, giving each integer core its own fetch/decode units, among other things; and Excavator switched to GPU libraries in laying out the circuitry, dropping power and area, the tradeoff being lower frequency.
GeoffreyA - Sunday, March 28, 2021 - link
Also, the reviews show that things were not as bad as we remember, though power was terrible.https://www.anandtech.com/show/6396/the-vishera-re...
https://www.anandtech.com/show/5831/amd-trinity-re...
Oxford Guy - Tuesday, April 6, 2021 - link
I don't need to look at reviews agaih. I know how bad the IPC was in Bulldozer, Piledriver, Steamroller, and Excavator. Single-thread in Cinebench R15, for instance, was really low even at 5.2 GHz in Piledriver. It takes chilled water to get it to bench at that clock.GeoffreyA - Wednesday, March 10, 2021 - link
Lack of competition, high prices, lack of integrity. I agree it's one big mess, but there's so little we can do, except boycotting their products. As it stands, the best advice is likely: find a product at a decent price, buy it, be happy, and let these rotten companies do what they want.Oxford Guy - Sunday, March 28, 2021 - link
'find a product at a decent price, buy it, be happy'Buy a product you can't buy so you can prop up monopolies that cause the problem of shortage + bad pricing + low choice (features to choose from/i.e. innovation, limited).
GeoffreyA - Sunday, March 28, 2021 - link
The only solution is a worldwide boycott of their products, till they drop their prices, etc.Franseven - Monday, March 8, 2021 - link
I know is a strange request, but i would like to know the iris integrated graphics benchmarks since i'm using my old 2080 ti for mining and i'm playing Minecraft and simple games with my integrated uhd630 of my 9700k, and unfortunately 5900x does not have integrated graphics, so i would like to know 11700k and 11900k perf with that, i have seen mobile benchmarks but as you know, is not the same thing, would like to see quality gaming benchmark as always, from you. thankskmmatney - Monday, March 8, 2021 - link
Would also be interested in this. I sold my 2070 Super - I owned it for a year, and sold it for what I paid (so free card for a year). The idea was to buy a 30X0 card with that money. That didn't happen, so lately I've just been playing Minecraft and older games on an old GTX 460. I'm curious about how the Xe graphics compares - with current prices on Ebay, the graphics along can add about $60 worth of value to the cpu.terroradagio - Monday, March 8, 2021 - link
ASUS just released another BIOS update with Rocket Lake enhancements. Probably more to come closer to the release too. This is why you don't post your review 3 weeks early.Everett F Sargent - Monday, March 8, 2021 - link
Like maybe an AVX-512 down clocking offset? Either Intel released their Rocket Engine a quarter too early or no amount of BIOS tweaking can do what you think it can do, at this, or any, point in time.From this review "Looking at our data, the all-core turbo under AVX-512 is 4.6 GHz, sometimes dipping to 4.5 GHz. Ouch. ... Our temperature graph looks quite drastic. Within a second of running AVX-512 code, we are in the high 90ºC, or in some cases, 100ºC. Our temperatures peak at 104ºC ... "
So already thermal throttling at Intel's promised 4.6 all core frequency using AVX-512. Makes you wonder what it takes to significantly OC this CPU. Which, you know, has barely been mentioned here in the comments section, OC'ing the damn thing, north or south of 300W or ~300W ...
https://i.imgur.com/8BEsGVo.png
terroradagio - Tuesday, March 9, 2021 - link
I guess you missed also the spot where normal AVX used less power than the 9900k. The vast majority don't care about AVX-512. It is just there so Intel can say it is. People who buy Rocket Lake will be interested because of gaming and there will probably be more stock than 7nm products from AMD.Qasar - Tuesday, March 9, 2021 - link
wow. really ? one test ( of a few) where intel was faster, and used less power ? big deal. over all rocket lake, looks to be a joke." People who buy Rocket Lake will be interested because of gaming " wrong, i know a few peope who are not even looking at intel, and are just waiting for zen 3 to be available, and this is for gaming and non gaming usage.
terroradagio - Tuesday, March 9, 2021 - link
I pointed out facts, and you are cherry picking one very selective AVX 512 test. Go away fanboy.Qasar - Tuesday, March 9, 2021 - link
like you your self have been doing ? and showing how much you love intel?hello pot meet kettle.
Everett F Sargent - Tuesday, March 9, 2021 - link
Yes, a 5.0GHz (all core boost clock) at 231.49W for the i9-9900KS versus a 4.6GHz (all core boost clock) at 224.56W for the i7-11700K. Conclusion? The i7-11700K runs 20-25W higher at the same all core boost frequency (4.6-5.0GHz). The i7-11700K wins at test duration though (by a similar margin as the inverse of the power ratio). The CPU energy used is about the same for both.amanpatel - Monday, March 8, 2021 - link
Few questions:1) Why is apple silicon or ARM equivalents not part of the benchmarks?
2) Why are so many CPU benchmarks needed, especially if they don't tell anything significant about them.
3) I'm not a huge gamer, but I also don't understand the point of so many gaming benchmarks for a CPU review.
Perhaps I'm the wrong audience member here, but it does seem a whole lot of charts that roughly say the same thing!
brucethemoose - Monday, March 8, 2021 - link
1: AT benches on Windows, and right now x86 vs ARM is kinda Apples-to-oranges on that platform, especially when one starts mixing in emulation and AVX.Give it time. More comparisons will come. But you'll probably see comparisons on Linux/Mac, and open source software in general, sooner.
2: People uses CPUs for different things. Some of these benches are relevant to those people.
At the same time, my use cases weren't really covered here, so... I get what you're saying.
3. Yeah, it seems rather silly to me, especially when Anand test GPU limited AAA games.
Where you really need a big CPU is in simulation/sandbox games, especially in servers for such things, and sluggish early access stuff. But no one ever benches those :/
Silver5urfer - Monday, March 8, 2021 - link
So with the 19% IPC claim and losing to 9900KS and 10700K what is the point of releasing this chip, from Intel. I never got much from AT benches a lot and preferred the Hardware unboxed, Gamernexus guys and others. But if this is the final performance figures, then this is really a DOA product from Intel. How can they allow this ? I never saw Intel in such a position..maybe X299 got rekted when Zen 2 dropped but this is mainstream segment.Damn it. AMD processors have the idiotic stock related issues, add that WHEA and USB shitstorm. Intel has bullshit performance over past gen except a Gen4 addition and extra lanes from chipset. GPUs are out of damn stock as well.
2020 and 2021 both are completely fucked up for PC HW purchases.
Gigaplex - Monday, March 8, 2021 - link
"I never saw Intel in such a position.."The Pentium 4, in particular the Prescott architecture was a dud back in the early 2000's. That era spawned the antitrust lawsuit against Intel for illegally blocking AMD sales since the Intel products weren't competitive.
dwbogardus - Monday, March 8, 2021 - link
The fact that Intel can even remain in a close second place, using a 14 nm process is impressive. Imagine what they could do with TSMC's 7 nm process! It would almost certainly outperform AMD by a significant margin.Bagheera - Monday, March 8, 2021 - link
Really tired of Intel fanboys saying this.architectures are designed for specific nodes - RKL's problems are exactly due to porting an arch onto a node it wasn't designed for.
the fact is Intel is not a partner for TSMC and their archs are not designed for TSMC processes. if Intel were to outsource CPU production to TSMC, they will either have to make a new arch or make make tweaks to existing ones - a multi-million $ endeavor with risms of issues like your just read with RKL.
Hifihedgehog - Tuesday, March 9, 2021 - link
> a close second place^Here we see in his natural environment your common everyday dude who fails at reading comprehension. I guess you didn't read the part about the serious gaming performance losses and latency regressions gen-over-gen, the 10% performance gap in single-threaded or 10-20%+ performance gaps in multithread, or the inexcusably high peak power draw? Talk about deluded...
RanFodar - Monday, March 8, 2021 - link
Though their efforts may be futile, I am glad Intel attempted to do something out of the ordinary; not a Skylake refresh, but a backport that is found to have worse performance. And yet, it is an attempt for Intel to learn their lessons for generations to come.Hifihedgehog - Tuesday, March 9, 2021 - link
Backporting is not a lesson; it is a last ditch effort or a fallback when all else fails on the manufacturing side. Half full, half empty cup viewing aside, they wasted even more valuable engineering manhours into a failed backport when it should have been invested into developing new architectures. A best use would have been developing the next release. The problem is Intel had to make Rocket Lake good enough in synthetic benchmarks to appease their investors. That, however, still does not address the elephants in the room of 10-20% single threaded performance gaps or—the one that takes the cake—the latency regressions that makes gaming worse, Intel’s historic crown jewel. Much like movies that fail at release and live on box office bombs that their producers later opine should have been cut early on in development, Intel should have cut this idea early on. If you are looking for a lesson that Intel should have learned here, there it is: avoid another Rocket Lake backporting disaster and just warm over your current microarchitecture with one more middling refresh one last time.Hifihedgehog - Tuesday, March 9, 2021 - link
Ian, I just want to say thank you for the incredible review. Just ignore the haters on social media and in the comments who get their panties in a bunch. If the product is garbage, say it like it is, like you did and quite well I might add. You were incredibly diplomatic about it and even openly and honestly showed when and where Intel did win on the rare occasion in the benchmarks. It is so silly how people make these CPU companies (who don't know them from Sam Hill) their religion, as if erecting a Gordon Moore or Lisa Su shrine would avail them anything. Silly geese.misiu_mp - Wednesday, March 10, 2021 - link
Is that the new bulldozer?Oxford Guy - Wednesday, March 10, 2021 - link
Worse.Bulldozer came from a money-starved little company, a company that couldn’t capitalize on having the superior design because of Intel’s monopolization.
This chip is from the big rich company that stomped on AMD with dirty tricks.
So, no. As stupid as Bulldozer was, this CPU seems to be quite a bit more unjustified.
ThereSheGoes - Wednesday, March 10, 2021 - link
Well, the results in this article are clearly not just wrong, but very wrong. https://www.hardwareluxx.de/index.php/artikel/hard...Bagheera - Wednesday, March 10, 2021 - link
did you even read the article you linked? it's taking 5800X in most benchmarks just like the Anandtech review.if you mean the gaming benches in the other review - 3 games is a terrible sample size and it still loses to 5800x in 2 out of 3.
Beaver M. - Wednesday, March 10, 2021 - link
Look at the results with the new BIOS.This review here is obsolete.
Hifihedgehog - Friday, March 12, 2021 - link
LOL. Fanboy delusion.First off, let's take a quick looksie at the Cinebench R20 results:
https://www.hardwareluxx.de/index.php/artikel/hard...
When switching from BIOS version 0402 to 0603, the 11700K's single-threaded performance actually DROPS from a score of 609 to 600. And its multicore performance is still less than the 10900K and the 5800X.
Switching gears, the games are no less unflattering:
https://www.hardwareluxx.de/index.php/artikel/hard...
The 11700K there, regardless of which of the two BIOS releases it uses, often loses to the 10900K and Ryzen 5000 series. It loses to the Ryzen 5000 series and 10900K in THREE out of the four games: The Division 2, Metro Exodus, The Shadow of the Tomb Raider.
In short: dude, what are you smoking?
Beaver M. - Wednesday, March 10, 2021 - link
As expected new reviews with newer BIOS versions improve performance significantly and puts the numbers to expected levels as well.This was a quick shot of a review and I fear it has tainted Cutress reputation a lot, especially because he defended it that much, even on video.
Oxford Guy - Wednesday, March 10, 2021 - link
With what power consumption?Raising performance by pushing power even higher may improve things like FPS in gaming tests but it obscures the big picture.
Want to talk about reputation? Remember the giant fridge-sized chiller Intel surreptitiously used to give a benchmark demo? Or, remember the ‘GenuineIntel’ fiasco? Or, remember the cute trick of putting a black box CPU inside the one people pay for, so that only special customers get the option of avoiding that particular spyware?
Somehow I think the writers here are going to be very hard pressed to challenge Intel in the cheatiness department, even without mentioning Intel’s history of abusing its monopoly power via OEM deals and the like.
Qasar - Wednesday, March 10, 2021 - link
Oxford Guy you forgot about how intel kept saying that 10 nm is " on track " the last 3-4 years :-)Qasar - Wednesday, March 10, 2021 - link
Beaver M" new reviews with newer BIOS versions improve performance significantly " from what i can tell from the graphs, for the most part, while performance may have improved over previous gen, it looks like it still looses to zen 3, but definitely not significantly, and still using more power then zen 3 overall, not that much of an improvement. a little upset that intel didnt get any performance crowns back, maybe, and that rocket lake still looks to be a dud ?
Bagheera - Thursday, March 11, 2021 - link
I fail to see the performance improvements in this other review. the games chosen were different (only 3 games? really?), and still lost to 5800X in 2/3.is it just "better" relative to last gen performance? the other review only tested a single resolution (and again, 3 whole games!)
Hifihedgehog - Friday, March 12, 2021 - link
LOL. Fanboy delusion.First off, let's take a quick looksie at the Cinebench R20 results:
https://www.hardwareluxx.de/index.php/artikel/hard...
When switching from BIOS version 0402 to 0603, the 11700K's single-threaded performance actually DROPS from a score of 609 to 600. And its multicore performance is still less than the 10900K and the 5800X.
Switching gears, the games are no less unflattering:
https://www.hardwareluxx.de/index.php/artikel/hard...
The 11700K there, regardless of which of the two BIOS releases it uses, often loses to the 10900K and Ryzen 5000 series. It loses to the Ryzen 5000 series and 10900K in THREE out of the four games: The Division 2, Metro Exodus, The Shadow of the Tomb Raider.
In short: dude, what are you smoking?
Technobile - Friday, March 12, 2021 - link
The 10700K costs a pittance at the minute, and after the final bios and microcode 11700K will be around 15 to 20% faster than it. Both a bargain when the only other option is dealing with the 'quirks' (to be kind) of an AMD systemQasar - Friday, March 12, 2021 - link
" The 10700K costs a pittance at the minute, and after the final bios and microcode 11700K will be around 15 to 20% faster than it " i will believe that, when i see it, honestly, that is hopeful thinking." Both a bargain when the only other option is dealing with the 'quirks' (to be kind) of an AMD system " and intel has had its own quirks over the years.
dsplover - Saturday, March 13, 2021 - link
They’re just doing this to give the impression they’re relevant. It’s safe to assume they’ll retake some market share but AMD took servers, laptop and desktop sectors by storm while Intel keeps moving old designs out the door.AMD 5750G, if it exists will render Intel designs useless this summer, while Intel struggles to get Alder Lake up and running.
Motherboard manufacturers will get tired of chasing new sockets after AM5 comes out.
What happened to these guys? It’s embarrassing and I’m an i7 fan boy..
GeoffreyA - Saturday, March 13, 2021 - link
"What happened to these guys?"My guess would be: complacency, underestimating the enemy, putting eggs in too many dead-end baskets, and management that made a mess of excellent engineering talent.
CiccioB - Sunday, March 14, 2021 - link
When reading this I think some of you just ended their school yesterday (with poor results) and just came here to say the first thing they think it is pro AMD. Just to give a (poor) contribution to what they think is an easy (for everyone) task as beating a dead horse (Intel).
I may shock you if I say that that "chasing designs" effort is the secret trick for motherboard producer too... surprise surprise.. MAKE MONEY!
So they do not get tired to do anything if this means selling more motherboards, and this just happens if you have to change your motherboards every couple of generations.
And I may shock you even more if I say you that those that make an upgrade using the same motherboard is just a so small number that no motherboard producer is really interested in supporting.
Usually when you change your CPU you just do not want only it to go a little faster but you want also the new technological improvements that meantime have been created, from faster bus, new and more connectors (M2 vs SATA), faster USBs, Thunderbolts, better memory support and such.
And this doesn't come if you do not also change the motherboard.
And to have and propose a better motherboard to sell, guess what? Yes, motherboard producer have to play the "chasing designs" game.
rfxcasey - Thursday, March 18, 2021 - link
Yeah "useless", please, all of this is splitting hairs, none of these processors are even close to being "useless".rfxcasey - Wednesday, March 17, 2021 - link
To me, looks like the i7 10700k holds it's own against the 5800x in gaming performance and is much cheaper.rfxcasey - Thursday, March 18, 2021 - link
i7 10700K actually beating the 5800x in many game benchmarks. I don't have a preference between AMD and Intel generally, but the i7 10700K is a great gaming processor, Intel did seem to make an embarrassing move with the 11th Gen, but for the cost, the 10700k a top, possibly THE top gaming CPU.quadibloc - Thursday, March 18, 2021 - link
When an official review comes in, with all the details, things may look a little better. But even now, I see one thing that's being overlooked. Since these chips have AVX-512, where that can be used, that will double their performance compared to processors that only have AVX-256. Except, of course, for the necessary slowdown for thermal reasons. So on workloads that involve a lot of AVX-512, they should really shine instead of being as terribly lackluster as they appear when that isn't taken into account.rfxcasey - Friday, March 19, 2021 - link
Yeah, might have to agree with you on this, from what I've seen the new instructions are amazing. They might be a real game changer but one thing is certain, it's the way of the future.Sgtkeebler - Wednesday, March 24, 2021 - link
Should I buy this or the i9-10900k. I have an i7-9700k but I want to blow my stimulus because I have a pto cash out coming in mayOxford Guy - Saturday, March 27, 2021 - link
If wasting money excites you...Priogeth - Tuesday, July 13, 2021 - link
Gesaz what is all this nonesense about?here some values:
(Name; Speed%*; mem; 1core; 2core; 4core; 8core; 64core; Price; )
11400F: 97.6%; 84pts; 159pts; 313pts; 594pts; 930pts; 1127pts; 200€;
7 3700x: 86%; 78pts; 135pts; 267pts; 508pts; 918pts; 1404pts; 264€;
7 5800x: 98%; 87pts; 157pts; 312pts; 589pts; 1062pts; 1637pts; 380€
9 5900x: 101% 89pts; 159pts, 314pts; 602pts; 1167pts; 2347pts; 500€
11700k: 107%; 87pts; 180pts; 358pts; 679pts; 1206pts; 1681pts; 350€
(100% speed performance = i9-9900k)
the values are from userbenchmark dot com, and can be checked your selves.
you might also want to go and check how these values come together b4 making any statements!
in comperison with these values you can clearly see that the i7-11700k is the top runner! not only in gaming but also in number crushing.
tho i would still more than recomend the ryzen9 5900x/5950x or a threadripper if you realy depend in numbercrushing like renderings and such.
also worth mentioning is the I7-11400F if youre on a budget. best gaming cpu for that price.