Well, there's MSI's GX60 and GX70 updates -- both are 1080p, but they also have 8970M I think. Still waiting for details on other Richland offerings, but I expect the usual suspects (and sadly, the usual LCDs).
What's annoyong for tbe gx60 and gx70 is the total lockdown of cpu clock and tdp even with the thermal headroom. It could do badly with a 45w version of the cpu at least.
Yeah, a bulky gaming laptop with a dGPU isn't an interesting use case for Richland, as far as I'm concerned. I do hope we get some thin + lights with Richland with good build quality and good displays.
Its normal for both AMD and Intel chips to not run anywhere near max turbo in multi-threaded applications. Turbo is for when you are not using all your cores.
No, intel chips (disregarding ULV) can run at turbo constantly assuming they are cooled properly. My 3630qm runs prime at the max multicore turbo of 3.2 ghz.
The way I read his comment was that he means the max turbo given is usually for 1 or 2 cores active. The max turbo for 3/4 cores active is often a good deal lower. So it will look like the APU/CPU doesn't reach the max turbo stage while in practice it is just running a well threaded workload and reaches the max turbo for that workload.
Note that last I looked, there was no utility that would report the real-time AMD APU Turbo Core clocks. On Trinity A10-4600M for example, I just see a constant 2.3GHz, but presumably the CPU is hitting higher clocks as well. You would need a way to disable Turbo Core on a laptop to determine how much it's helping, which I haven't really looked into. For BIOS implementations, I suspect no one provides this, but maybe some software from the past year has changed things.
As for Intel, on most laptops they consistently hit max or near-max CPU Turbo Boost, at least in CPU workloads. If you hit both the CPU and iGPU, you'll usually see one or the other give way.
AMD overdrive and HWinfo64 will report turbo use. So does PScheck, but the leaked PScheck is extremely buggy.
An A10-4600m will always run at 2.7ghz when 3 or 4 cores are loaded as long as some silly BIOS throttling is not interfering, and it will run 3.2ghz if 2 or more cores are in C6 (long idle).
You can use PCI config space editing programs to disable BIOS throttling algorithms, while still leaving the built in 100C throttle in place (you can also disable the 100C throttle if you want...). PScheck tries to, but it does so incorrectly and breaks 3.2ghz use. You can also do things like set CAS latency (greyed out in overdrive) and force the northbridge to not downclock on IGP use. If you're interested in how to do any of this, let me know.
As for testing if IGP overclocks actually work, just max out the slider in afterburner of overdrive. If it crashes then it actually worked, if not...
Looks great. I'm in a need for a laptop in the upcoming months so the a10 could be the one powering it. The performance is well enough for my taste. Gotta wait for these to become available in my country though
Jarred you managed to overclock a trinity GPU? Are you sure? From my experience all it does it lock the GPU to turbo (685MHz), which is why you get a performance increase without actually overclocking. I've had afterburner and GPU-z say I was running over 2GHz...
So AFAICT, with MSI Afterburner I am able to OC Trinity. I'm not sure which is right, but Afterburner reports the overclocked speeds in 1MHz increments whereas GPU-Z shows larger jumps -- e.g. I saw 720MHz when clocked at 750, and at 760MHz GPU-Z shows 758MHz. I'll do some tests now to see if the OC is real or not....
I can overclock my A10-5800K GPU on my AsRock board. The linux utility (AMDOverdriveCtrl) appears to be reporting correct clock speeds. My board is such that I can lock the GPU speed at a certain value, or have it throttle up to a fixed value when needed. I have mine now ramping up to 1013mhz from 800mhz stock.
I feel like AMD needs to throw out their current naming scheme and introduce some coherency and clarity into their naming convention. What is the interplay between FX & Phenom processors? What is the difference between the high end FX processors and the high end A class APUs other than the lack of integrated graphics?
I know the answers to these questions. But does the general public? Maybe I'm just not grasping their intent and making a fool of myself?
As far as the general public knowing the difference between the various AMD or Intel products, it's of little concern. Mainstream consumers primarily look for hardware that will do what they need and at the lowest price. Only enthusiasts pay much attention to the hardware specifics.
You can expect Richland chips to run at higher frequencies for longer periods of time compared to Trinity for two reasons. One is that AMD has refined the power regulation to the various areas of the APU so areas that don't need to be running at high frequencies when they have little load, are throttled back as is their power consumption and thus thermal load so other areas of the APU can run faster. Second few have noticed but the Richland chips that AMD announced the other day are not just a refined Trinity chip. If you look really closely at the chip you will notice a significant design change that will contribute to running at higher frequencies for longer periods of time without hitting the TDP or sectional thermal limits. You'll need to squint your eyes very closely to see the detail but it's there for all the world to see.
This is just a stop-gap Trinity refresh to hold put into notebooks untill Kaveri is ready. With Kaveri AMD seriously needs to give a big jump in GPU/CPU performance.
For real Richland performance video check this out!! Lots of other videos too. I think the guy is working flat out. Crossfire videos have started appearing too.
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.
22 Comments
Back to Article
kyuu - Thursday, May 23, 2013 - link
"Several laptops with Richland APUs have already been announced[...]"What are these laptops? Are there any that aren't cheapie 15.6" units with 1366x768 screens?
JarredWalton - Thursday, May 23, 2013 - link
Well, there's MSI's GX60 and GX70 updates -- both are 1080p, but they also have 8970M I think. Still waiting for details on other Richland offerings, but I expect the usual suspects (and sadly, the usual LCDs).Meaker10 - Thursday, May 23, 2013 - link
What's annoyong for tbe gx60 and gx70 is the total lockdown of cpu clock and tdp even with the thermal headroom. It could do badly with a 45w version of the cpu at least.kyuu - Friday, May 24, 2013 - link
Yeah, a bulky gaming laptop with a dGPU isn't an interesting use case for Richland, as far as I'm concerned. I do hope we get some thin + lights with Richland with good build quality and good displays.whyso - Thursday, May 23, 2013 - link
No comment on how well the cpu can keep the clocks up. The 10-4600m usually runs at 2.1-2.5 ghz in multithreaded applications, far from its max turbo.Stuka87 - Thursday, May 23, 2013 - link
Its normal for both AMD and Intel chips to not run anywhere near max turbo in multi-threaded applications. Turbo is for when you are not using all your cores.whyso - Thursday, May 23, 2013 - link
No, intel chips (disregarding ULV) can run at turbo constantly assuming they are cooled properly. My 3630qm runs prime at the max multicore turbo of 3.2 ghz.Death666Angel - Thursday, May 23, 2013 - link
The way I read his comment was that he means the max turbo given is usually for 1 or 2 cores active. The max turbo for 3/4 cores active is often a good deal lower. So it will look like the APU/CPU doesn't reach the max turbo stage while in practice it is just running a well threaded workload and reaches the max turbo for that workload.JarredWalton - Thursday, May 23, 2013 - link
Note that last I looked, there was no utility that would report the real-time AMD APU Turbo Core clocks. On Trinity A10-4600M for example, I just see a constant 2.3GHz, but presumably the CPU is hitting higher clocks as well. You would need a way to disable Turbo Core on a laptop to determine how much it's helping, which I haven't really looked into. For BIOS implementations, I suspect no one provides this, but maybe some software from the past year has changed things.As for Intel, on most laptops they consistently hit max or near-max CPU Turbo Boost, at least in CPU workloads. If you hit both the CPU and iGPU, you'll usually see one or the other give way.
Khenglish - Friday, May 24, 2013 - link
AMD overdrive and HWinfo64 will report turbo use. So does PScheck, but the leaked PScheck is extremely buggy.An A10-4600m will always run at 2.7ghz when 3 or 4 cores are loaded as long as some silly BIOS throttling is not interfering, and it will run 3.2ghz if 2 or more cores are in C6 (long idle).
You can use PCI config space editing programs to disable BIOS throttling algorithms, while still leaving the built in 100C throttle in place (you can also disable the 100C throttle if you want...). PScheck tries to, but it does so incorrectly and breaks 3.2ghz use. You can also do things like set CAS latency (greyed out in overdrive) and force the northbridge to not downclock on IGP use. If you're interested in how to do any of this, let me know.
As for testing if IGP overclocks actually work, just max out the slider in afterburner of overdrive. If it crashes then it actually worked, if not...
testbug00 - Sunday, May 26, 2013 - link
AMD has a different version of turbo. derp.different things work differently.
aalvisk - Thursday, May 23, 2013 - link
Looks great. I'm in a need for a laptop in the upcoming months so the a10 could be the one powering it. The performance is well enough for my taste. Gotta wait for these to become available in my country thoughKhenglish - Thursday, May 23, 2013 - link
Jarred you managed to overclock a trinity GPU? Are you sure? From my experience all it does it lock the GPU to turbo (685MHz), which is why you get a performance increase without actually overclocking. I've had afterburner and GPU-z say I was running over 2GHz...JarredWalton - Thursday, May 23, 2013 - link
Hmm...perhaps you're right. Afterburner at least is showing 750MHz, but I would need to do some benchmarking to see if it's actually helping.JarredWalton - Friday, May 24, 2013 - link
So AFAICT, with MSI Afterburner I am able to OC Trinity. I'm not sure which is right, but Afterburner reports the overclocked speeds in 1MHz increments whereas GPU-Z shows larger jumps -- e.g. I saw 720MHz when clocked at 750, and at 760MHz GPU-Z shows 758MHz. I'll do some tests now to see if the OC is real or not....johnny_boy - Tuesday, May 28, 2013 - link
I can overclock my A10-5800K GPU on my AsRock board. The linux utility (AMDOverdriveCtrl) appears to be reporting correct clock speeds. My board is such that I can lock the GPU speed at a certain value, or have it throttle up to a fixed value when needed. I have mine now ramping up to 1013mhz from 800mhz stock.MFK - Monday, May 27, 2013 - link
I feel like AMD needs to throw out their current naming scheme and introduce some coherency and clarity into their naming convention.What is the interplay between FX & Phenom processors?
What is the difference between the high end FX processors and the high end A class APUs other than the lack of integrated graphics?
I know the answers to these questions. But does the general public?
Maybe I'm just not grasping their intent and making a fool of myself?
Myrandex - Wednesday, May 29, 2013 - link
Note that Newegg has been selling a Desktop Richland APU for some time now...Jason
Beenthere - Saturday, June 1, 2013 - link
As far as the general public knowing the difference between the various AMD or Intel products, it's of little concern. Mainstream consumers primarily look for hardware that will do what they need and at the lowest price. Only enthusiasts pay much attention to the hardware specifics.You can expect Richland chips to run at higher frequencies for longer periods of time compared to Trinity for two reasons. One is that AMD has refined the power regulation to the various areas of the APU so areas that don't need to be running at high frequencies when they have little load, are throttled back as is their power consumption and thus thermal load so other areas of the APU can run faster. Second few have noticed but the Richland chips that AMD announced the other day are not just a refined Trinity chip. If you look really closely at the chip you will notice a significant design change that will contribute to running at higher frequencies for longer periods of time without hitting the TDP or sectional thermal limits. You'll need to squint your eyes very closely to see the detail but it's there for all the world to see.
http://products.amd.com/en-us/NotebookAPUSideBySid...
Gaugamela - Saturday, June 1, 2013 - link
This is just a stop-gap Trinity refresh to hold put into notebooks untill Kaveri is ready. With Kaveri AMD seriously needs to give a big jump in GPU/CPU performance.majorleague - Thursday, June 13, 2013 - link
For real Richland performance video check this out!!Lots of other videos too.
I think the guy is working flat out.
Crossfire videos have started appearing too.
:-)
https://www.youtube.com/watch?v=bQq50l3bnLs
Nyceis - Thursday, September 5, 2013 - link
good stuff