Original Link: https://www.anandtech.com/show/1529



Intel's plans for higher Front Side Bus frequencies come as no surprise to those that have been carefully reading for years now. Back in October of 2001 we outlined Intel's FSB targets for the major core releases of the coming years:

"Although not nearly as high, FSB frequencies are also ramping up as clock frequencies do. Intel's Northwood (0.13-micron Pentium 4) core will eventually use a 533MHz FSB, Intel's Prescott core will use a 800MHz FSB, and their Tejas core will use an incredible 1.2GHz FSB upon its release."

Things have obviously changed since we made that statement, primarily that Tejas has since been canceled and Intel introduced an Extreme Edition of the Pentium 4 based on the Gallatin Xeon core. Given the revised plans, what do present day roadmaps say about Intel's FSB targets for the future?

Today Intel is making the move from 800MHz up to the 1066MHz FSB, with the release of the Pentium 4 3.46 Extreme Edition. Priced at $999, the 3.46EE will be followed up by the 3.73EE as the only two chips to support the faster FSB for almost a year. Current Intel roadmaps show absolutely no support for the 1066MHz FSB from any processor other than the two aforementioned Extreme Edition chips until the end of next year, when Prescott-2M chips will finally receive 1066MHz FSB support; there are also plans for dual core 1066MHz FSB chips although they have not been publicly outlined on any roadmap.

It seems as if Intel isn't interested in releasing a new revision of their mainstream chipsets and wants to thus save the move for Prescott to 1066 for their upcoming Lakeport chipset. What that means is that today we are seeing a very limited launch of a new FSB as a feature exclusive to the Extreme Edition CPUs, no doubt a way of adding more value to the very expensive Extreme Edition offering.

The faster FSB is enabled through a revision to the 925X chipset called the 925XE. The 925XE simply adds 1066MHz FSB support, and nothing more, to the feature set of 925X. By enabling 1066MHz FSB support, the 925XE chipset can finally run the FSB and DDR2-533 memory buses synchronously, offering a balanced 8.5GB/s of bandwidth over each bus while operating at multiples of the same base 266MHz frequency. The end result is that the 1066MHz FSB should guarantee lower latency memory accesses for any chip that uses it with DDR2-533. However it seems that Intel paired the faster FSB with the worst possible choice out of their processor line; armed with an on-die 2MB L3 cache, the Pentium 4 Extreme Edition already has the lowest latency memory accesses out of the entire Pentium 4 line, thus softening the impact of the synchronous 1066MHz FSB.

The other thing to keep in mind is that faster FSB frequencies truly become beneficial as higher CPU speeds cause the bus to saturate with data requests and transfers - strike 2 against Intel's choice of introducing the FSB with the Extreme Edition. Prescott continues to be the higher clocked CPU (not to mention the deeper pipelined CPU of the two), thus making it the better candidate of the two for the 1066MHz FSB.

While some have speculated that Prescott isn't yielding well at the 1066MHz FSB, it would seem that the reasons for not moving Prescott to 1066MHz are much more marketing related. The current 915 and 925X chipsets are not selling well at all and the thought of replacing both of those chipsets with new, 1066MHz FSB versions (especially the high quantity mainstream 915 chipset), so that Intel could migrate the entire Prescott line to the 1066MHz FSB wouldn't go over very well with the motherboard manufacturers. From talking with numerous motherboard manufacturers it would seem that Intel is trying to save any major performance improvements for their upcoming Glenwood/Lakeport chipsets, in order to avoid what happened with 925X/915P where there was no performance improvement over the previous generation of chipsets. This may also be why there is no talk of 1066MHz FSB Pentium 4s on Intel's current roadmaps, all conversation about the future 1066MHz FSB parts appears to be happening over telephone and face-to-face meetings only, the former of which we have been privy to upon numerous occasions recently. It could very well be that even Intel is unsure about what their future 1066MHz FSB plans will be at this point.

We shouldn't eliminate the possibility that Prescott may not be yielding well at the faster FSB. Remember that the major limitation to FSB frequency is CPU packaging, the connection and routing of signals from the solder bumps on the silicon core to the pins (or in this case pads) on the chip itself. It could very well be that Prescott's 7-layer design has a correspondingly more complex packaging layout, resulting in some present day hurdles to increasing FSB frequency. That being said, we have heard more reports of people being able to run their Prescott CPUs at the 1066MHz FSB than those that haven't. Take that for what it's worth.

The Pentium 4 3.46EE replaces the 3.40EE as the fastest Extreme Edition CPU from Intel. However with only a 66MHz clock speed advantage, we would expect almost all of any performance increase we see to be caused by the 1066MHz FSB. But how much performance can we expect from a FSB increase on a processor that isn't the best choice for a FSB increase? Given the simplicity of today's launch (just one slightly faster chip with a slightly modified chipset), let's just find out.



Does the 1066MHz FSB Improve Memory Performance?

Quite possibly the biggest feature of the 1066MHz FSB today is the fact that it runs at a clock multiple of DDR2-533's frequency. Why is that such a big feature? It's analogy time:

If two people are having a conversation and they can both talk and listen at the same rate, then the conversation will flow as smoothly as possible. If person A talks and listens slower than person B, then person B will always be waiting for person A instead of communicating as fast as possible - a frustrating situation for those that have been here before. Running a FSB and memory bus asynchronously is just as frustrating to the CPU; if the two frequencies aren't synchronous then there is an additional latency penalty incurred while transferring data between the two buses and because of that additional latency penalty, there is a reduction in usable bandwidth.

With the original 925X chipset we were a bit unhappy to see that the Pentium 4's 800MHz FSB was paired with DDR2-533, creating one of those frustrating asynchronous situations. But with a 1066MHz FSB (266MHz x 4), the 925XE can communicate synchronously with DDR2-533 (266MHz x 2), thus reducing memory latency and increasing memory bandwidth in theory. What do we see in practice? To answer this question we look to two trusted measures of memory bandwidth and latency: CacheMem and ScienceMark 2.0.

First looking at latency we see that with the 1066MHz FSB, memory latency with DDR2-533 looks like it's hardly improved. Running the FSB at 1066MHz manages to shave off a just a few clock cycles.

DDR2-533 3-3-3-12 Latency Comparison - Cachemem
 
1066MHz FSB
800MHz FSB
Performance Improvement
512-byte stride - 32MB block
226 cycles
227 cycles
0.4%
1k stride - 32MB block
239 cycles
241 cycles
0.8%
2k stride - 32MB block
266 cycles
266 cycles
0%
4k stride - 32MB block
311 cycles
311 cycles
0%

We look at ScienceMark and see the same basic situation but with slightly improved performance; looking at the absolute latency values in nanoseconds we see that the 1066MHz FSB manages to reduce memory latency by around 2 - 6%.

DDR2-533 3-3-3-12 Latency Comparison - ScienceMark 2.0
 
1066MHz FSB
800MHz FSB
Performance Improvement
16-byte stride
4.69 ns
5 ns
6.2%
64-byte stride
17.5 ns
18.12 ns
3.4%
256-byte stride
73.43 ns
75.93 ns
3.3%
512-byte stride
75.93 ns
77.81 ns
2.4%

The reduction in latency isn't insignificant under ScienceMark, but what about its affects on memory bandwidth?

Looking at Cachemem once again we see an increase in memory bandwidth of just under 4%.

DDR2-533 3-3-3-12 Bandwidth Comparison
 
1066MHz FSB
800MHz FSB
Performance Improvement
ScienceMark 2.0
4742.02 MB/s
4347.63 MB/s
9.1%
CacheMem
3455.3 MB/s
3324.7 MB/s
3.9%

ScienceMark appears more considerate of Intel's hard work and shows a 9% increase in memory bandwidth. The thing to keep in mind here is that the improvement in memory bandwidth will depend on the types of applications run, but the ScienceMark and Cachemem results should give you an indication of the range of improvements to be expected in applications that are memory bandwidth limited. In those applications that aren't currently bound by memory bandwidth, the impact will be much less.

What is important to keep in mind here is that DDR2 is still running at relatively high latencies. Even while running synchronously with the FSB, the 925XE and DDR2 combination still posts higher memory access latencies than 875/DDR400 platforms thanks to DDR2's high memory timings. We were able to run 3-3-3-12 timings on our DDR2 test platform by keeping memory voltage at a safe, but overclocked, 2.0V, but anything faster than that was too much for today's DDR2-533. It will take even lower latency DDR2 in order for even the 925XE platform to show some further performance advantages.

Given the relatively small increases in memory bandwidth and decreases in latency, the extra 66MHz of the 3.46EE will have to go a long way in order to gain any more ground for Intel. Let's see how things shape up in some real world tests.



Does it Improve Real World Performance?

There is a convenient convergence point between the 1066MHz FSB and the 800MHz FSB - 3.2GHz. By underclocking our 3.4EE and our 3.46EE to 3.2GHz we managed to put together a nice comparison of the impact of FSB on real world performance, independent of CPU and memory clock speed. Granted, the impact of the 1066MHz FSB will be greater at higher CPU clock speeds, but the impact at 3.2GHz should be able to tell us how much of the 3.46EE's performance advantage is due to its faster FSB.

The table below gives a good indication of the lack of performance improvement due to the 1066MHz FSB today in most applications. With an average performance increase of less than 1%, you shouldn't expect the 1066MHz FSB to do much for Intel at all.

Business/General Use
 
1066MHz FSB
800MHz FSB
Performance Improvement
Business Winstone 2004
21.2
21.2
0.00%
SYSMark 2004 - Communication
136
136
0.00%
SYSMark 2004 - Document Creation
201
198
1.49%
SYSMark 2004 - Data Analysis
162
161
0.62%
Microsoft Office XP with SP-2
511
511
0.00%
Mozilla 1.4
401
405
1.00%
ACD Systems ACDSee PowerPack 5.0
593
593
0.00%
Ahead Software Nero Express 6.0.0.3
543
553
1.84%
WinZip Computing WinZip 8.1
419
431
2.86%
WinRAR
419
413
1.43%
Average Performance Increase
 
 
0.92%

Under Multitasking Content Creation applications we see that despite the nature of these applications to be more memory bandwidth intensive, the 800MHz FSB simply wasn't a limitation for the Pentium 4 Extreme Edition. Couple that with the fact that with a very large on-die L3 cache, the Extreme Edition needs to fetch data across the FSB much less frequently, it's no surprise that the biggest performance improvement in our Multitasking Content Creation tests was only 1.52%.

Multitasking Content Creation
 
1066MHz FSB
800MHz FSB
Performance Improvement
Content Creation Winstone 2004
30.9
30.9
0.00%
SYSMark 2004 - 3D Creation
207
204
1.45%
SYSMark 2004 - 2D Creation
264
260
1.52%
SYSMark 2004 - Web Publication
187
185
1.07%
Multitasking: Mozilla and Windows Media Encoder
596
600
0.67%
Average Performance Increase
0.94%

There's not much to see in the Video Creation/Photo Editing tests, the 1066MHz FSB does absolutely nothing for performance here.

Video Creation/Photo Editing
 
1066MHz FSB
800MHz FSB
Performance Improvement
Adobe Photoshop 7.0.1
347
347
0.00%
Adobe Premiere 6.5
533
533
0.00%
Roxio VideoWave Movie Creator 1.5
289
289
0.00%
Average Performance Increase
 
 
0.00%

In the past, DivX encoding has seen reasonable performance increases due to a faster FSB and increased memory bandwidth. With the move to the 1066MHz FSB we seem to have hit a limit, as there's absolutely no performance improvement here either. It looks like it will take much higher clock speeds for the 1066MHz FSB to make a difference.

Audio/Video Encoding
 
1066MHz FSB
800MHz FSB
Performance Improvement
MusicMatch Jukebox 7.10
434
434
0.00%
DivX Encoding
49.9
49.9
0.00%
XV iD Encoding
28.7
28.5
0.70%
Microsoft Windows Media Encoder 9.0
2.32
2.32
0.00%
Average Performance Increase
 
 
0.00%

Games have also been areas where faster FSB frequencies have benefited Intel, but once again we see that the average performance increase is less than a percent. Starwars Battlefront shows the greatest increase in performance at 2.8% due to the 1066MHz FSB.

Gaming
 
1066MHz FSB
800MHz FSB
Performance Improvement
Doom 3
86.1
85.2
1.05%
Sims 2
46
46
0.00%
CS: Source
156.8
156.4
0.26%
Halo
88.4
88
0.45%
Far Cry
133.5
132
1.12%
Star Wars Battlefront
143
139
2.80%
Battlefield Vietnam
239
239
0.00%
UT2004
59
58.6
0.68%
Wolf: ET
98
96.9
1.12%
Warcraft III
60
59
1.67%
Average Performance Increase
 
 
0.91%

We weren't expecting to see much in the 3D rendering tests and the 1066MHz FSB did not disappoint with only a 0.74% average performance increase here.

3D Rendering
 
1066MHz FSB
800MHz FSB
Performance Improvement
Discreet 3ds max 5.1 (DirectX)
280
282
0.71%
Discreet 3ds max 5.1 (OpenGL)
339
342
0.88%
SPECapc 3dsmax 6
1.63
1.62
0.61%
Average Performance Increase
 
 
0.74%

Our final suite of tests are the professional applications tested by SPECviewperf 8. Here we see the largest overall gains provided by the 1066MHz FSB, with performance improvements approaching 5%, and average performance improvements approaching 3%. There's very little gain in compiling performance but in the realm of 3D professional application performance the 1066MHz FSB begins to show its worth. The gains here will only get better as clock speeds increase, so maybe the 1066MHz FSB will pay off for those running demanding enough applications to require a $1000+ 3.46EE CPU.

Professional Apps
 
1066MHz FSB
800MHz FSB
Performance Improvement
SPECviewperf 8 - 3dsmax-03
15.99
15.99
0.00%
SPECviewperf 8 - catia-01
12.62
12.08
4.28%
SPECviewperf 8 - light-07
12.89
12.41
3.72%
SPECviewperf 8 - maya-01
12.66
12.32
2.69%
SPECviewperf 8 - proe-03
15.9
15.31
3.71%
SPECviewperf 8 - sw-01
12.87
12.53
2.64%
SPECviewperf 8 - ugs-04
13.71
13.1
4.45%
Visual Studio 6
16.8
16.7
0.60%
Average Performance Increase
 
 
2.76%



Intel D925XECV2: Intel's Enthusiast motherboard

For this review Intel sent us their latest motherboard based on the 925XE chipset. Our own Wesley Fink gives his brief look at the motherboard:

Computer users have always admired the speed and stability of Intel motherboards. When testing other boards the standard for stock performance has been Intel brand motherboards. This has been true for both boards based on Intel chipsets and boards based on new competing chipsets. They are measured against Intel performance, as that performance has been as close to a standard in stock motherboard performance as exists in the computer industry.

However, when a Computer Enthusiast shops for a motherboard Intel is rarely on the list for consideration. The reasons are simple - Intel has not offered the kinds of control most performance enthusiasts want. This list of wants includes controls like adjustable memory timings, a full range of adjustable memory speeds, adjustable memory voltage, CPU FSB adjustments, and CPU voltage adjustments. Intel's philosophy has been to offer the safe and accepted options that would not compromise the specifications of the processor and system. This assured the Intel motherboard was not run out-of-spec, keeping failures and RMA's at an acceptable level.

Performance Enthusiasts often seem to care less about motherboard features, but motherboard makers also know a large segment of computer component buyers do care about features like high-definition audio, flexible IO options, RIAD, and flexible storage options. Intel has done a better job in the recent past in addressing these "wants", but the Performance Enthusiast often felt left out of the Intel marketing plan.

This philosophy has gradually been changing at Intel, as we have seen more talk about catering to the Enthusiast in the last year. However, directions for Intel seem to shift frequently, which leads to talk on the one-side about Enthusiast control options on motherboards at the same time the 10% overclock lock is being implemented. This performance schizophrenia also led to Intel calling some very limited efforts in the past an enthusiast board.

All of this leads to the new Intel 925XE chipset motherboard, the D925XECV2. Those who have not been following the evolving interest in the enthusiast at Intel will be surprised to find many of the things Performance Enthusiasts want are a part of the new Intel 925XE motherboard. They didn't come out of nowhere; they are just a continuation of Intel's renewed efforts to attract Enthusiasts to Intel brand motherboards.



Intel D925XECV2: Enthusiast Options

The new Intel 925XE includes adjustments for memory timings, memory speeds, memory voltage, and CPU FSB adjustments. These options are well-hidden in the Intel BIOS, but they are much more complete than you might expect on an Intel board.

The Main menu provides system information, but there is little indication of what is lurking inside the Intel BIOS. The real action for the Enthusiast is in the Advanced menu.

In the Advanced menu are the options to adjust PCI, Boot, Peripheral, Drive, Floppy, Video, USB, etc. The main controls for the Enthusiast are found in the Chipset Configuration menu

Chipset Configuration includes two important sub-menus - Burn-In Mode and memory

Continue and Extended Burn-In Mode should be enabled. You can then select Host Burn-In Mode.

Host Burn-In Mode now reveals an adjustment range of -2% to 10% in 1% increments. With a base frequency of 266 or 1066FSB, this is an adjustment range of 1045 FSB to 1173 FSB in 10.66MHz increments. While this is a very limited range compared to the average Enthusiast board, it is certainly a beginning for Enthusiast controls on an Intel board.

PCI speed can be adjusted form the default 33.33 to 36.36 and 40.0.

PCI Express also has an adjustment range from the default 100 to around 110 in 1.32Mhz increments.

Returning to Chipset Configuration, the Memory sub-menu has a wide range of memory adjustments.

After selecting Manual-User Defined you can select memory timings. This includes CAS Latency (tCL), RAS-to-CAS Delay, tRP., and tRAS.

DDR2 Memory Speed can be adjusted to base frequencies of 266, 333, 400, 533, and 667.

It was particularly surprising to find a reasonable range of memory voltage adjustments in the Intel BIOS. The available range is from default 1.8V to 2.1V. this is wide enough to allow many DDR533 rated DDR2 dimms to run at DDR667 or faster.

There are just two major controls missing for the Enthusiast. There is no option in the BIOS that we could find to adjust the base CPU frequency. If this could be forced to lower standard speeds such as 200 (800), there would be much more flexibility available for controlling the CPU. There is also no provision for CPU voltage, which is becoming less important as processor operating voltages have decreased. Our test CPU did have limited multiplier adjustments of 12X-13X but we do not know if shipping 1066 processors will also have this feature.

We could wish for wider ranges on several options, but the good news is this Intel board does contain the kind of tweaking controls Enthusiasts have been asking for from Intel.



Intel D925XECV2: Utilities

Intel provides a very wide range of Software and Utilities with the D925XECV2 to support the BIOS adjustments and features of the 925XE chipset. This includes SoundMAX, Sonic Focus, MusicMatch Jukebox, Intervideo Home Theater Silver and Win DVD Creator 2, NTI CD-Maker, RestoreIT! Lite, Norton Internet Security 2004, LANDesk, Embassy Trust Suite, Webex, and Intel Desktop Utilities. Many of the features are integrated into two utilities that tie together 925XE adjustment options, features, and software.

The Desktop Control Center is used to Tune, Stress, and Monitor system performance. You can adjust and test the memory timings, memory voltage, bus speed, and frequency settings in the Control Center.

Temperature monitoring and Fan Speed adjustments are important to many users. You can also adjust fan settings and cooling strategies in DCC. Temperatures and Speeds can be monitored in the Control Center as well.

Intel Audio Studio integrates High-Definition Audio functions and the software that supports those audio features. Sonic Focus, SoundMAX and Dolby Digital Live! can be controlled in the Audio Studio utility.



Intel D925XECV2: Basic Features

The Intel 925XE chipset is essentially the 925X chipset with the addition of a 1066 FSB option. You can find more information on the 925X chipset in the AnandTech 925x/915 launch article.

Intel D925XECV2 Motherboard Specifications
CPU Interface Socket T (Intel LGA-775)
Power Interface 24-pin (775 & Server) ATX and 4-pin 12V
Chipset Intel 925XE/Intel ICH6R
Bus Speeds -2% to +10% in 1% increments
PCI Express Speeds 100 to 109.24 in 1.32MHz increments
PCI Speeds Default (33.33MHz), 36.35, 40.0
Core Voltage No CPU Voltage Adjustments
DRAM Voltage Default, 1.8V to 2.1V in 0.1V increments
PCI Express Voltage No PCIe Voltage Adjustments
DRAM Speeds 266,333,400,533,667
Memory Slots Four 240-pin DDR2 Dual-Channel Slots
Memory to 4GB Total
Expansion Slots 1 x16 PCI Express Slot
2 x1 PCI Express Slots
4 PCI Slots
Onboard SATA 4-Drive SATA by ICH6R
Onboard PATA One Standard PATA IDE 100/66/33 (2-drives)
SATA/IDE RAID 4-Drive Intel Matrix RAID
With Native Command Queuing
Onboard USB 2.0/IEEE-1394 8 USB 2.0 ports supported by ICH6R
3 1394A FireWire ports by ICH6R
Onboard LAN 1 Gigabit PCIe Ethernet
by Marvel 88E8050-NNC
Onboard Audio Realtek ALC8800 High-Definition Audio Codec



The Test

Given that there's pretty much no performance benefit to the 1066MHz FSB as we've illustrated in the previous pages, we will offer the next several pages of benchmarks only as a reference point to show how the 3.46EE stacks up to the CPUs we compared in our most recent CPU review, but without our usual commentary. Based on what you've already seen, it should be no surprise that the 3.46EE is really no faster than the 3.4EE, but if you would like to see all of the individual benchmarks feel free to flip through the coming pages. Those who aren't interested should feel free to skip directly to the conclusion.

In addition to our usual tests we've included PC World's WorldBench 5, an application based test suite much like Winstone and SYSMark that incorporates many popular applications. Unlike the aforementioned benchmarks, WorldBench does not test multitasking power, rather focusing on single application performance, making it very complementary to our existing benchmarks. The one thing to keep in mind about the WorldBench results is that the variation between test runs can be pretty significant; we do everything to make sure that the results are as consistent as possible (multiple runs, throwing out outliers, etc...) but the variation between runs in these tests can be as high as 6% - thus we would suggest looking at performance differences only greater than 10% in these tests for any sort of significance. The rest of the tests have variations between runs of 1 - 3%.

Our hardware configurations are similar what we've used in previous comparisons, with one addition - our Athlon 64 testbed now uses the recently released nForce4 chipset. For a review of that chipset read our own Wesley Fink's review of NVIDIA's latest chipset with SLI support.

AMD Athlon 64 Configuration

Socket-939 Athlon 64 CPUs
2 x 512MB OCZ PC3200 EL Dual Channel DIMMs 2-2-2-10
NVIDIA nForce4 Reference Motherboard
ATI Radeon X800 XT PCI Express

AMD Athlon XP Configuration

Athlon XP 3200+
2 x 512MB OCZ PC3200 EL Dual Channel DIMMs 2-2-2-10
ASUS A7N8X Deluxe nForce2 400 Motherboard
ATI Radeon X800 XT AGP

Intel Pentium 4 Configuration

LGA-775 Intel Pentium 4 and Extreme Edition CPUs
2 x 512MB Crucial DDR-II 533 Dual Channel DIMMs 3-3-3-12
Intel D925XECV2 Motherboard (we used the same board for 1066MHz FSB and 800MHz FSB tests
ATI Radeon X800 XT PCI Express



Business/General Use Performance

Business Winstone 2004

Business Winstone 2004 tests the following applications in various usage scenarios:

. Microsoft Access 2002
. Microsoft Excel 2002
. Microsoft FrontPage 2002
. Microsoft Outlook 2002
. Microsoft PowerPoint 2002
. Microsoft Project 2002
. Microsoft Word 2002
. Norton AntiVirus Professional Edition 2003
. WinZip 8.1

We saw in our initial benchmarks that the 1066MHz FSB does very little for business applications, thus it's no surprise here to see the 3.46EE barely outperform its 3.4GHz counterpart.

Business Winstone 2004

Office Productivity SYSMark 2004

SYSMark's Office Productivity suite consists of three tests, the first of which is the Communication test. The Communication test consists of the following:

"The user receives an email in Outlook 2002 that contains a collection of documents in a zip file. The user reviews his email and updates his calendar while VirusScan 7.0 scans the system. The corporate web site is viewed in Internet Explorer 6.0. Finally, Internet Explorer is used to look at samples of the web pages and documents created during the scenario."

Similar to what we just saw in the Business Winstone tests, the 3.46EE doesn't change things for Intel here. The 3.46EE is about the same speed as the Pentium 4 550 and the older 3.4EE.

Office Productivity SYSMark 2004

The next test is Document Creation performance, which shows very little difference in drive performance between the contenders:

"The user edits the document using Word 2002. He transcribes an audio file into a document using Dragon NaturallySpeaking 6. Once the document has all the necessary pieces in place, the user changes it into a portable format for easy and secure distribution using Acrobat 5.0.5. The user creates a marketing presentation in PowerPoint 2002 and adds elements to a slide show template."

The extra speed and the faster FSB gives Intel the right to the top bar, but performance remains virtually unchanged from the pre-3.46EE days.

Document Creation SYSMark 2004

The final test in our Office Productivity suite is Data Analysis, which BAPCo describes as:

"The user opens a database using Access 2002 and runs some queries. A collection of documents are archived using WinZip 8.1. The queries' results are imported into a spreadsheet using Excel 2002 and are used to generate graphical charts."

The 3.46EE does a little better here, climbing up one spot to outperform the Pentium 4 530, but still falling behind the Pentium 4 550.

Data Analysis SYSMark 2004

Microsoft Office XP SP-2

Here we see in that the purest of office application tests, performance doesn't vary all too much. The spread of performance here is around 6%, which is nothing terribly significant. While AMD continues to top the charts, Intel is basically on par with them here.

There's no performance difference here between the 3.46EE and the 3.4EE:

Microsoft Office XP with SP-2

Mozilla 1.4

Quite possibly the most frequently used application on any desktop is the one we pay the least amount of attention to when it comes to performance. While a bit older than the core that is now used in Firefox, performance in Mozilla is worth looking at as many users are switching from IE to a much more capable browser on the PC - Firefox.

The faster FSB improves performance a bit here but it is still not enough to dethrone AMD, the 3.46EE performs similar to an Athlon 64 3000+.

Mozilla 1.4

ACD Systems ACDSee PowerPack 5.0

ACDSee is a popular image editing tool that is great for basic image editing options such as batch resizing, rotating, cropping and other such features that are too elementary to justify purchasing something as powerful as Photoshop for. There are no extremely complex filters here, just pure batch image processing.

ACD Systems ACDSee PowerPack 5.0

Ahead Software Nero Express 6.0.0.3

While it was a major issue in the past, these days buffer underrun errors while burning a CD or DVD are few and far between thanks to high performance CPUs as well as vastly improved optical drives. When you take the optical drive out of the equation, how do these CPU's stack up with burning performance?

As you'd guess, they're all pretty much the same, with the slight variations between chips falling within expectations. Any of these chips will do just fine. The 3.46EE is just as fast as its predecessor.

Ahead Software Nero Express 6.0.0.3

Winzip

Archiving performance ends up being fairly CPU bound as well as I/O limited. The faster FSB and slightly higher clock speed give the 3.46EE a slight advantage over the 3.4EE, but nothing noticeable. On the charts it puts Intel within finger's reach of AMD.

WinZip Computing WinZip 8.1

WinRAR 3.40

Pulling the hard disk out of the equation we can get a much better idea of which processors are truly best suited for file compression. While the hard drive hid a lot of the shortcomings of the Athlon XP in WorldBench's WinZip test, they are all revealed in WinRAR's built in benchmark that is largely disk I/O independent.

Much like the WinZip test, we see only a slight performance boost under WinRAR.

WinRAR 3.40 Archiving Performance



Multitasking Content Creation

MCC Winstone 2004

Multimedia Content Creation Winstone 2004 tests the following applications in various usage scenarios:

. Adobe® Photoshop® 7.0.1
. Adobe® Premiere® 6.50
. Macromedia® Director MX 9.0
. Macromedia® Dreamweaver MX 6.1
. Microsoft® Windows MediaTM Encoder 9 Version 9.00.00.2980
. NewTek's LightWave® 3D 7.5b
. SteinbergTM WaveLabTM 4.0f

As you can see above, Lightwave is part of the MCC Winstone 2004 benchmark suite. As an individual application, Lightwave does manage to get a healthy performance benefit with multithreaded rendering enabled, especially when paired with Hyperthreading enabled CPUs like the Pentium 4s here today. All chips were tested with Lightwave set to spawn 4 threads.

There's a slight boost here, but the 3.46EE is still slower than the Athlon 64 3400+.

Multimedia Content Creation Winstone 2004

ICC SYSMark 2004

The first category that we will deal with is 3D Content Creation. The tests that make up this benchmark are described below:

"The user renders a 3D model to a bitmap using 3ds max 5.1, while preparing web pages in Dreamweaver MX. Then the user renders a 3D animation in a vector graphics format."

The 3.46EE comes within striking distance of the Pentium 4 560 but doesn't have what it takes to dethrone it.

3D Content Creation SYSMark 2004

Next, we have 2D Content Creation performance:

"The user uses Premiere 6.5 to create a movie from several raw input movie cuts and sound cuts and starts exporting it. While waiting on this operation, the user imports the rendered image into Photoshop 7.01, modifies it and saves the results. Once the movie is assembled, the user edits it and creates special effects using After Effects 5.5."

2D Content Creation SYSMark 2004

The Internet Content Creation suite is rounded up with a Web Publishing performance test:

"The user extracts content from an archive using WinZip 8.1. Meanwhile, he uses Flash MX to open the exported 3D vector graphics file. He modifies it by including other pictures and optimizes it for faster animation. The final movie with the special effects is then compressed using Windows Media Encoder 9 series in a format that can be broadcast over broadband Internet. The web site is given the final touches in Dreamweaver MX and the system is scanned by VirusScan 7.0."

Web Publication SYSMark 2004

Mozilla + Media Encoder

While AMD dominated in WorldBench 5's Mozilla test, encoding a file using Windows Media Encoder in the background not only makes this test more appreciative of the Pentium 4 but also of Hyper Threading.

Thanks to the multitasking nature of this benchmark there's a 6% performance advantage that the 3.46EE enjoys over the 3.4EE, which is greater than what we saw in the dedicated 1066MHz FSB tests. There are two factors at work here; for one we're looking at a higher clock speed thus taking better advantage over the 1066MHz FSB and at the same time there's a pretty large variance between runs in most of the WorldBench benchmarks so the numbers here aren't too surprising.

Multitasking: Mozilla and Windows Media Encoder



Video Creation/Photo Editing

Adobe Photoshop 7.0.1

Adobe Photoshop 7.0.1

Adobe Premier 6.5

Prescott seems to do quite a bit for Intel here, with the 560 and 550 easily offering better performance than the 3.4EE and 3.46EE, despite smaller caches. That being said, even the top of the line Pentium 4 560 can't outperform even the Athlon XP 3200+ in this test. While WorldBench does use an older version of Premier, it is one that is still widely used, making these results quite pertinent.

Adobe Premiere 6.5

Roxio VideoWave Movie Creator 1.5

While Premier is a wonderful professional application, consumers will prefer something a little easier to use. Enter: Roxio's VideoWave Movie Creator, a fairly full featured yet consumer level video editing package.

Intel has worked very closely with Roxio in the past, thus it's no surprise to see Movie Creator take advantage of the Prescott core as best as possible. The Extreme Edition still takes the crown, but the 560 and 550 manage to keep AMD's best at bay.

Roxio VideoWave Movie Creator 1.5



DivX 5.2.1 with AutoGK

Armed with the latest version of DivX (5.2.1) and the AutoGK front end for Gordian Knot, we took all of the processors to task at encoding a chapter out of Pirates of the Caribbean. We set AutoGK to give us 75% quality of the original DVD rip and did not encode audio.

There's a slight performance advantage here to the 3.46EE, but nothing huge.

DiVX 5.2.1 Encoding Performance

XviD 5 with AutoGK

Another very popular codec is the XviD codec, and thus we measured encoding performance using it instead of DivX for this next test. The rest of the variables remained the same as the DivX test.

XviD 5 Encoding Performance

Windows Media Encoder 9

To finish up our look at Video Encoding performance we've got two tests both involving Windows Media Encoder 9. The first test is WorldBench 5's WMV9 encoding test.

Microsoft Windows Media Encoder 9.0

But once we crank up the requirements a bit and start doing some HD quality encoding under WMV9 the situation changes dramatically; the 3.46EE still holds its own at the top of the charts.

Windows Media Video 9 HD Encoding Performance

Closing up our video encoding tests, while AMD does win some, with appropriate optimizations in place Intel seems to be the right candidate here.



Gaming Performance

Doom 3

Our Doom 3 CPU Battlegrounds article already made it quite clear that Intel did not have what it takes to be the highest performer in Doom 3. The 3.46EE doesn't manage to help Intel as much as they need.

Doom 3

Counterstrike: Source

The next big game after Doom 3 is, of course, Half Life 2. And while the game is still not out, Valve's Visual Stress Test that comes with Counterstrike: Source gives us a good glimpse into future performance under Half Life 2.

Half Life 2/Counterstrike: Source VST

Halo

Halo 1.05

Starwars Battlefront

Starwars Battlefront

Battlefield Vietnam

Battlefield Vietnam is similar to Starwars Battlefront not only in its game play but also in its performance; there's no real difference between any of the top performers here. Almost all of these CPUs end up being GPU limited at 1024x768.

Battlefield Vietnam

Unreal Tournament 2004

Unreal Tournament 2004

Wolfenstein: Enemy Territory

An oldie but a goodie, Enemy Territory is still played quite a bit and makes for a great CPU test as today's GPUs can easily handle the rendering load of the Quake 3 based game.

Wolfenstein: Enemy Territory

The Sims 2

While a clear departure from our usual game tests, The Sims 2 is more popular than any of the other games we've featured here in certain crowds - it is effectively the Doom 3 of those who don't play prefer life-simulation to first person shooters. And interestingly enough, it makes for a very impressive CPU benchmark.

The Sims 2

Far Cry

Far Cry 1.2

Warcraft III

Warcraft III The Frozen Throne



3D Rendering

3dsmax 5.1

WorldBench includes two 3dsmax benchmarks using version 5.1 of the popular 3D rendering and animation package: a DirectX and an OpenGL benchmark.

Discreet 3ds Max 5.1 (DirectX)

Discreet 3ds Max 5.1 (OpenGL)

3dsmax 6

For the next 3dsmax test we used version 6 of the program and ran the SPECapc rendering tests to truly stress these CPUs. Since there's not much new to report here we're only going to report the Rendering Composite score. For more details feel free to read our Athlon 64 FX-55 Review where we analyze the performance data in much greater depth.

Discreet 3ds max 6 (OpenGL) - SPECapc Rendering Composite



Workstation Applications

Visual Studio 6

Carried over from our previous CPU reviews, we continue to use Visual Studio 6 for a quick compile test. We are still using the Quake 3 source code as our test and measure compile time in seconds. The results are pretty much in line with what we've seen in the past.

Visual Studio 6 Compiler Performance

SPECviewperf 8

For our next set of professional application benchmarks we turn to SPECviewperf 8. SPECviewperf is a collection of application traces taken from some of the most popular professional applications, and compiled together in a single set of benchmarks used to estimate performance in the various applications the benchmark is used to model. With version 8, SPEC has significantly improved the quality of the benchmark, making it even more of a real world indicator of performance.

We have included SPEC's official description of each one of the 8 tests in the suite.

3dsmax Viewset (3dsmax-03)

"The 3dsmax-03 viewset was created from traces of the graphics workload generated by 3ds max 3.1. To insure a common comparison point, the OpenGL plug-in driver from Discreet was used during tracing.

The models for this viewset came from the SPECapc 3ds max 3.1 benchmark. Each model was measured with two different lighting models to reflect a range of potential 3ds max users. The high-complexity model uses five to seven positional lights as defined by the SPECapc benchmark and reflects how a high-end user would work with 3ds max. The medium-complexity lighting models uses two positional lights, a more common lighting environment.

The viewset is based on a trace of the running application and includes all the state changes found during normal 3ds max operation. Immediate-mode OpenGL calls are used to transfer data to the graphics subsystem."

The biggest surprise here is that there is a huge performance impact (13%) by moving down to a single channel memory subsystem with the Athlon 64. There are also a couple of other surprises, with the Pentium 4 560 doing surprisingly well, coming in at the heels of the Athlon 64 FX-55.

SPECviewperf 8 - 3dsmax 3.1 Performance

CATIA Viewset (catia-01)

"The catia-01 viewset was created from traces of the graphics workload generated by the CATIATM V5R12 application from Dassault Systems.
Three models are measured using various modes in CATIA. Phil Harris of LionHeart Solutions, developer of CATBench2003, supplied SPEC/GPC with the models used to measure the CATIA application. The models are courtesy of CATBench2003 and CATIA Community.

The car model contains more than two million points. SPECviewperf replicates the geometry represented by the smaller engine block and submarine models to increase complexity and decrease frame rates. After replication, these models contain 1.2 million vertices (engine block) and 1.8 million vertices (submarine).

State changes as made by the application are included throughout the rendering of the model, including matrix, material, light and line-stipple changes. All state changes are derived from a trace of the running application. The state changes put considerably more stress on graphics subsystems than the simple geometry dumps found in older SPECviewperf viewsets.

Mirroring the application, draw arrays are used for some tests and immediate mode used for others."

SPECviewperf 8 - CATIA V5R12 Performance

Lightscape Viewset (light-07)

"The light-07 viewset was created from traces of the graphics workload generated by the Lightscape Visualization System from Discreet Logic. Lightscape combines proprietary radiosity algorithms with a physically based lighting interface.

The most significant feature of Lightscape is its ability to accurately simulate global illumination effects by precalculating the diffuse energy distribution in an environment and storing the lighting distribution as part of the 3D model. The resulting lighting "mesh" can then be rapidly displayed."

SPECviewperf 8 - Lightscape Visualization System Performance

Maya Viewset (maya-01)

"The maya-01 viewset was created from traces of the graphics workload generated by the Maya V5 application from Alias.

The models used in the tests were contributed by artists at NVIDIA. Various modes in the Maya application are measured.

State changes as made by the application are included throughout the rendering of the model, including matrix, material, light and line-stipple changes. All state changes are derived from a trace of the running application. The state changes put considerably more stress on graphics subsystems than the simple geometry dumps found in older viewsets.

As in the Maya V5 application, array element is used to transfer data through the OpenGL API."

SPECviewperf 8 - Maya V5 Performance

Pro/ENGINEER (proe-03)

"The proe-03 viewset was created from traces of the graphics workload generated by the Pro/ENGINEER 2001TM application from PTC.

Two models and three rendering modes are measured during the test. PTC contributed the models to SPEC for use in measurement of the Pro/ENGINEER application. The first of the models, the PTC World Car, represents a large-model workload composed of 3.9 to 5.9 million vertices. This model is measured in shaded, hidden-line removal, and wireframe modes. The wireframe workloads are measured both in normal and antialiased mode. The second model is a copier. It is a medium-sized model made up of 485,000 to 1.6 million vertices. Shaded and hidden-line-removal modes were measured for this model.

This viewset includes state changes as made by the application throughout the rendering of the model, including matrix, material, light and line-stipple changes. The PTC World Car shaded frames include more than 100MB of state and vertex information per frame. All state changes are derived from a trace of the running application. The state changes put considerably more stress on graphics subsystems than the simple geometry dumps found in older viewsets.

Mirroring the application, draw arrays are used for the shaded tests and immediate mode is used for the wireframe. The gradient background used by the Pro/E application is also included to better model the application workload."

SPECviewperf 8 - Pro/ENGINEER Performance

SolidWorks Viewset (sw-01)

"The sw-01 viewset was created from traces of the graphics workload generated by the Solidworks 2004 application from Dassault Systemes.

The model and workloads used were contributed by Solidworks as part of the SPECapc for SolidWorks 2004 benchmark.

State changes as made by the application are included throughout the rendering of the model, including matrix, material, light and line-stipple changes. All state changes are derived from a trace of the running application. The state changes put considerably more stress on graphics subsystems than the simple geometry dumps found in older viewsets.

Mirroring the application, draw arrays are used for some tests and immediate mode used for others."

SPECviewperf 8 - Solidworks 2004 Performance

Unigraphics (ugs-04)

"The ugs-04 viewset was created from traces of the graphics workload generated by Unigraphics V17.

The engine model used was taken from the SPECapc for Unigraphics V17 application benchmark. Three rendering modes are measured -- shaded, shaded with transparency, and wireframe. The wireframe workloads are measured both in normal and anti-alised mode. All tests are repeated twice, rotating once in the center of the screen and then moving about the frame to measure clipping performance.

The viewset is based on a trace of the running application and includes all the state changes found during normal Unigraphics operation. As with the application, OpenGL display lists are used to transfer data to the graphics subsystem. Thousands of display lists of varying sizes go into generating each frame of the model.

To increase model size and complexity, SPECviewperf 8.0 replicates the model two times more than the previous ugs-03 test."

SPECviewperf 8 - Unigraphics V17



Cheap - High End: Athlon 64 3800+ vs. Pentium 4 560

This entire review has been focused on comparing processors that retail either close to or over $1000 since they are flagship chips, but what about the next price class down? There's a very interesting situation here with the most expensive non-EE Pentium 4 being significantly cheaper than the highest end non-FX Athlon 64 processors.

Using our RealTime Pricing Engine we pitted the bargain high end Pentium 4 560, going for $440, against the more expensive Athlon 64 3800+ which is still selling for just over $600. So which CPU is the better choice if you want performance almost as good as the high end chips, but at a much more reasonable price? Let's find out.

First we've got the business/general use application tests; with AMD taking five of the benchmarks and Intel taking three, the score card is pretty close between the two, however AMD does win by higher margins so the nod goes to AMD here. However, once you take into account the price difference between the two CPUs, AMD's margin of victory is cut into. Despite the price argument, we'll still give the crown here to AMD.

Business/General Use
 
Intel Pentium 4 560
AMD Athlon 64 3800+
Winner
Business Winstone 2004
21.4
23.6
AMD (10.2%)
SYSMark 2004 - Communication
144
139
Intel (3.6%)
SYSMark 2004 - Document Creation
201
194
Intel (3.6%)
SYSMark 2004 - Data Analysis
184
145
Intel (27%)
Microsoft Office XP with SP-2
522
520
Tie
Mozilla 1.4
459
316
AMD (31.2%)
ACD Systems ACDSee PowerPack 5.0
547
490
AMD (10.4%)
Ahead Software Nero Express 6.0.0.3
540
491
AMD (9%)
WinZip Computing WinZip 8.1
412
409
Tie
WinRAR
479
617
AMD (28.8%)
Winner
 
 
AMD

Under our Multitasking Content Creation tests Intel comes away with three wins, and AMD with two. One of AMD's victories is in a test with a fairly high margin of error reducing the real world performance advantage; factor in the lower CPU cost and Intel wins this one.

Multitasking Content Creation
 
Intel Pentium 4 560
AMD Athlon 64 3800+
Winner
Content Creation Winstone 2004
32.7
35.3
AMD (8%)
SYSMark 2004 - 3D Creation
231
205
Intel (12.7%)
SYSMark 2004 - 2D Creation
288
265
Intel (8.7%)
SYSMark 2004 - Web Publication
206
188
Intel (9.6%)
Multitasking: Mozilla and Windows Media Encoder
676
590
AMD (12.7%)
Winner
Intel

AMD wins two, Intel wins one, factor in variance between tests and AMD still comes out ahead by a reasonable margin. AMD wins the performance crown here, but its debatable whether or not the performance advantage is worth the price.

Video Creation/Photo Editing
 
Intel Pentium 4 560
AMD Athlon 64 3800+
Winner
Adobe Photoshop 7.0.1
342
315
AMD (7.9%)
Adobe Premiere 6.5
461
371
AMD (19.5%)
Roxio VideoWave Movie Creator 1.5
287
310
Intel (7.4%)
Winner
 
 
AMD

Intel wins the A/V encoding tests hands down, barely losing one of the tests.

Audio/Video Encoding
 
Intel Pentium 4 560
AMD Athlon 64 3800+
Winner
MusicMatch Jukebox 7.10
484
458
AMD (5.4%)
DivX Encoding
55.3
47.9
Intel (15.4%)
XV Encoding
33.9
32.6
Intel (4%)
Microsoft Windows Media Encoder 9.0
2.57
2.21
Intel (16.3%)
Winner
 
 
Intel

There's no argument here, the 3800+ is clearly the faster gaming processor.

Gaming
 
Intel Pentium 4 560
AMD Athlon 64 3800+
Winner
Doom 3
84.6
97.3
AMD (15%)
Sims 2
47.3
55.4
AMD (17.1%)
CS: Source
142.8
171.6
AMD (20.2%)
Halo
87.5
95
AMD (8.6%)
Far Cry
130.3
151.4
AMD (16.2%)
Star Wars Battlefront
140
144
AMD (2.9%)
Battlefield Vietnam
236
240
AMD (1.7%)
UT2004
59.3
67.6
AMD (14%)
Wolf: ET
97.2
107.1
AMD (10%)
Warcraft III
60
62
AMD (3.3%)
Winner
 
 
AMD

AMD wins two benchmarks by less than 5%, while Intel wins one benchmark by less than 7%. We'll call this one a toss up between the two in performance, but factor in price and Intel wins.

3D Rendering
 
Intel Pentium 4 560
AMD Athlon 64 3800+
Winner
Discreet 3ds max 5.1 (DirectX)
267
254
AMD (4.9%)
Discreet 3ds max 5.1 (OpenGL)
327
312
AMD (4.6%)
SPECapc 3dsmax 6
1.64
1.54
Intel (6.5%)
Winner
 
 
Tie

For 3D Professional applications AMD takes the clear lead, not to mention a significant advantage in compiler performance as well. The added cost of the Athlon 64 is well worth it in some of the tests, but is more questionable in others.

Professional Apps
 
Intel Pentium 4 560
AMD Athlon 64 3800+
Winner
SPECviewperf 8 - 3dsmax-03
17.04
16.75
Intel (1.7%)
SPECviewperf 8 - catia-01
13.87
14.03
AMD (1.2%)
SPECviewperf 8 - light-07
14.3
14.3
Tie
SPECviewperf 8 - maya-01
13.12
18.58
AMD (41.6%)
SPECviewperf 8 - proe-03
16.7
17.19
AMD (2.9%)
SPECviewperf 8 - sw-01
13.09
13.72
AMD (4.8%)
SPECviewperf 8 - ugs-04
15.31
16
AMD (4.5%)
Visual Studio 6
16.8
13.1
AMD (22.0%)
Winner
 
 
AMD

So in the end, who takes the crown? AMD or Intel? The 3800+ took four category wins, while the Pentium 4 560 only took two, however with the exception of the gaming and professional apps category, AMD's victories were not overwhelming - especially once you take into account the fact that the 3800+ is priced much higher than the Pentium 4 560. Now that you can purchase at least a couple of 915 based motherboards for less than $130 the total cost of ownership for the Intel platform doesn't eat into the CPU price advantage. For the most part we'd say the 3800+ is faster than the Pentium 4 560 but not always worth the added cost. It's unusual but in many cases, the Pentium 4 560 is actually the bargain high-end chip of the two.



Final Words

So there you have it folks - the 1066MHz FSB does absolutely nothing for performance. The 3.46EE does manage to outperform its 3.4GHz/800MHz FSB predecessor, but the margin of improvement is negligible. Intel desperately needs a win here and other than the more affordable price of the Pentium 4 560, there's very little going for the CPU king these days. It will take higher speed Prescott CPUs or dual core in order for the added bandwidth of the 1066MHz FSB to truly be of any use - and it will take lower latency DDR2 memory to finally give the latest Pentium 4 platforms lower latency memory access than the ones they replaced.

We can only wonder what Intel is thinking, releasing an entirely new chipset just four months after they released the original. Granted with very few 925X designs on the market right now, there shouldn't be too many upset 925X owners, but it's still a very strange situation. Either the 1066MHz FSB is going to make its way to CPUs faster than we have anticipated, or Intel has just introduced the world's first useless FSB improvement for the next 9 months.

The move to the 1066MHz FSB is in sharp contrast to the past two FSB bumps that we've seen from Intel. The introduction of the 533MHz FSB back in 2002 yielded up to a 12% gain in gaming performance, and a 3 - 6% gain in individual applications as it was paired with PC1066 RDRAM. Then came the 800MHz migration that showed a 3 - 9% increase in gaming performance, and just under a 12% increase in professional application performance. But with the move to the 1066MHz FSB we have a platform launch that, in the spirit of the 925X and 915 launches, does virtually nothing for performance.

Is it worth it? Sure, 1066 will be worth it when there are higher clock speed (or dual core) processors to take advantage of it. But given that Intel isn't planning on ramping clock speed up too high anytime soon, we'd say that the 1066MHz FSB is best left for late next year, when more useful implementations of it will appear.

Log in

Don't have an account? Sign up now