The latest test method changes

In spite of the fact that our new test method dated 2007 generally satisfied our readers, we nevertheless decided to make several corrections to further improve it. Here's a brief changelog (the test method article will also be updated):

• Excluded I/O scores from SPEC tests for Maya, Pro/ENGINEER, and SolidWorks, as they only reflect disk subsystem performance. We are naturally not interested in these from the angle of CPUs benchmarking. While these I/O scores will be added to the detailed report (Microsoft Excel format), they won't be considered in total score.
• Added weighting coefficients for the Web Server subtest. The average score is now calculated, assuming that 60% of requests are "light" (small-sized), 30% are "average", and 10% are "large". We believe that weighting coefficients reflect real web server activity better than when we consider the aforementioned loads to be equal.
• Added Supreme Commander strategy to the Games section. This game is seriously optimized for multi-way (multi-core) configurations. We believe it to be a good expansion of our game list. After all, PC gamers play not only first-person shooters.
• Changed the method of game average score calculation. Now we calculate an average score for each game, based on its results at low quality (20% weighting coefficient), average quality (60% weighting coefficient), and high quality (20% weighting coefficient). Values obtained are used to calculate general average score.

This benchmarking doesn't pretend to be complete, it's just our first attempt to use the new test method. Judging by this, we considered the first test should produce some "references". Therefore it gathers only 3 CPUs: Intel Core 2 Duo E4300 approved as a reference for the 2007 test method (its performance is considered 100 points and is used as a reference for other processors); quad-core Intel Core 2 eXtreme QX6700, which is still the top; and AMD's top-level Athlon 64 X2 6000+, which we hadn't tested before.

Of course, we'll expand our test results database in the near future, since an "honed" test method enables to get results much faster than a "raw" one.

Hardware and software

Testbeds

• 2 x 1024 MB of RAM
• Samsung SP1614C (SATA) HDD
• Boxed coolers
• Chieftec GPS-550AB A PSU

* - "2 x ..." means per core
** - for AMD processors this is memory controller bus clock rate
*** - measured differently for Intel and AMD processors; impossible to compare directly.

Software

1. Windows XP Professional x64 edition SP1
2. 3ds max 9 x64 edition
3. Maya 8.5 x64 edition
4. Lightwave 3D 9 x64 edition
5. MATLAB R2006a (7.2.0.32) x64 edition
6. Pro/ENGINEER Wildfire 2.0
7. SolidWorks 2005
8. Photoshop CS2 (9.0)
9. Visual Studio 2005 Professional
10. Apache HTTP Server 2.2.4
11. CPU RightMark 2005 Lite (1.3) x64 edition
12. WinRAR 3.62
13. 7-Zip 4.42 x64 edition
14. FineReader 8.0 Professional
15. LAME 3.97
16. Monkey Audio 4.01
17. OGG Encoder 2.83
18. Windows Media Encoder 9 x64 edition
19. Canopus ProCoder 2.01.30
20. DivX 6.4
21. Windows Media Video VCM 9
22. x264 v.604
23. XviD 1.1.2
24. F.E.A.R. 1.08
25. Half-Life 2 1.0
26. Quake 4 1.3
27. Call of Duty 2 1.2
28. Serious Sam 2 2.07
29. Supreme Commander 1.0.3220

Benchmarking

Essential foreword to charts

Our test method has two peculiarities of data representation: (1) all data types are reduced to one - integer relative score (performance of a given processor relative to that of Intel Core 2 Duo E4300, given its performance is 100 points), and (2) detailed results are published in this Microsoft Excel table, while the article contains only summary charts by benchmark classes. We will nevertheless focus your attention on detailed results, when needed.

3D Modelling

Without a doubt, QX6700 won in this test thanks to its rendering performance provided by 4 cores vs. 2 cores of the closest rival. But if you note the details, you will see that Intel CPU outperforms A64 X2 6000+ in other aspects as well: in interactive operations (Graphics score) and even in Hardware Shaders performance. The latter, by the way, proves that "hardware" shaders of modern graphics cards are not that hardware sometimes, otherwise how could the same graphics card produce noticeably different Hardware Shaders performance results with different processors?

CAD/CAE

This is one of two tests, in which Athlon 64 X2 6000+ could outperform Intel's quad-core. At that, not only in the total score, but in each of 3 subtests as well.

Digital Image Processing

An excellent showcase of 4-way vs. 2-way competition. Still, if an application is well optimized for multi-way (multi-core) configurations, amount of cores may become a clincher. Let's wait for a quad-core solution from AMD...

Compiling

This tests continues the previous. Microsoft Visual Studio 2005 seemingly puts all system processors to good use. A good point to move to a newer compiler in tests.

Web Server

The complete collapse of Intel's quad-core despite seemingly perfect conditions - a "very multi-threaded" application. We believe that might be due to a very specific L2 cache structure of QX6700 (two dedicated L2 caches for each couple of cores.) Because of this it can't operate "correctly" with Windows XP task manager. When the latter moves threads from one core to another, sometimes threads get to cores using another set of L2. As a result, it has to "re-cache" all thread data, but this places a heavy load on memory subsystem and, consequently, leads to a performance loss. It's a good example of that you shouldn't hurry. If a concept of shared L2 for all cores is approved, you should use it in all products. However, you can act simpler - clarify to Microsoft how threads should be distributed among cores of the very specific QX6700. :)

Synthetic tests

CPU RightMark rendering system uses any amount of cores up to 32, so the QX6700 is an obvious winner.

Archiving

Intel Core 2 eXtreme QX6700 wins fair and square in all applications (see details). You can't say the victory is critical though.

OCR

As we have mentioned before, it's a pity, but FineReader's batch recognition doesn't support multi-way configurations for reasons completely unclear. Therefore, based on this test results, we can state that Intel Core 2 architecture is more preferable than that of AMD Athlon 64 X2. This doesn't consider amount of cores, just core architecture.

Audio encoding

An "olde" test group which nearly lost its importance today due to high result predictability. No comments.

Video encoding

The critical advantage of Intel's quad-core is essentially based on two tests: Canopus ProCoder and x264. A simple riddle: why is that so, considering both coders use all system cores? :)

Games

A complete fiasco of AMD. Considering this test group includes games not optimized for multi-way (multi-core) configurations, we can obviously state that Intel's top solution is more preferable for both enthusiasts of multi-way, and users of single-threaded classics.

Total score

The total score indicate the trend very clearly. Perhaps, in certain specific situations AMD's top solution performs better, but it generally loses to the quad-core Core 2 eXtreme QX6700. Speaking of the "reference" Core 2 Duo E4300, we can state that today it's not far from processor performance peak - the currently most powerful CPU is just 1.5 times faster.

Supposed energy consumption

Such a mystery it is: at 100% load Intel's quad-core always consumes less power than AMD's processor. Moreover, in the idle state Core 2 eXtreme QX6700 reduces power consumption by more than 3 times. While Athlon 64 X2 6000+ only halves it to terrific 54W! To comfort the AMD followers (and running ahead a bit) we can say that such "phenomenal" values are seemingly appropriate only to A64 X2 6000+. Because A64 X2 4400+ indicated the sane 27W idle state consumption on the same testbed.

Of course, we can assume that the BIOS of M2N32-SLI Deluxe motherboard doesn't support AMD Cool'n'Quiet with this very processor. However, ASUS website officially states there IS support for Athlon 64 X2 6000+ since BIOS 0903. And we not only used this very BIOS in our tests, but also re-ran them with the latest beta 1001 - to find no differences.

Conclusions

It's actually simple: Intel's current core is really better than that of AMD. Whether it's a fair 64-bit application, or an older 32-bit one; whether it's multi-way (multi-core) optimized, or is single-threaded classics. In almost every situation Intel Core 2 outperforms old AMD K8 even in its DDR2 modification. And this is so natural: a new core beats an old one anyway. On the other hand, this doesn't mean the show is over. One company just released new-core processors slightly earlier than the other. We've seen that before many times. Those who need the best performance here and now should select the current leader. And the followers of certain vendors may wait a bit all right, until it acts in return. And judging by our experience, we are sure AMD is bound to act...

Write a comment below. No registration needed!