Some may think this material provides a too narrow view as it only deals with products of one manufacturer. But we believe it is the optimal form to express our attitude. Indeed, ASUS is a worldwide respected trademark. And when we received a number of new and really significant ASUSTeK products for tests recently we decided to share our impressions in the form of a separate review.

According to the company's officials, not only the modified versions of popular models are better equipped, but they are also supposed to work faster. Such speed-focused approach is something quite rare for these days when most mainboard manufacturers have put up with the idea that similar chipsets mean similar board performance. And since the new products have provoked curiosity, why not bate it?

Where does board performance come from?

This digressing chapter is needed to finish once and for all with widespread delusions about the source of board performance. It is hardly a secret that two ready (assembled) systems that only differ in their mainboards may show different performance, though such situations are getting increasingly rare. Why does this happen, even when the boards are based on one and the same chipset?

In general, it results from the fact that the chipset has some controlable parameters. Memory timings, the best-known of which is CAS (Column Access Strobe), are the most influential of them. The solution seems simple: we take the slower board, set the same timings as on the faster one, and it will work at the same speed. Yes, but only if it works at all. To make a system function correctly with fast timings one needs to have a good layout, quality elemental base (including resistors and capacitators), well-designed functioning units, to say nothing of just careful manufacturing. And these are actually factors that make some mainboards differ from others.

That is why accelerated and otherwise optimised versions (revisions) of motherboards differ from their previous variants either in a better layout, or in a slightly modified elemental base, or in both at once. Due to such modifications, some chipset settings can be made more aggressive, which will result in a performance gain. The changes mostly concern those "hidden" parameters which can't be altered in BIOS Setup. All in all, there are several dozens of them.

But it also happens sometimes that a new revision is just based on an R&D engineer's idea to make the board less expensive. For example, he can find out that when the system is not overclocked, this or that capacitator may be removed or replaced with a cheaper one. And sometimes, when there's too much economy, the manufacturer decides to make the board slower with the help of the same hidden BIOS settings. Clearly, such revision will be worse than the previous variant: less stable at overclocking, more sensitive to the quality of other components, etc. That is why we are not advised to believe the manufacturers on their bare word when they announce a new board revision.

But don't take it all too close to your heart. Not that what we have said above is not true, but just in most cases, mainboard manufacturers pay no special attention to performance and are oriented to certain average parameters which will create no problems even for a mediocre PCB design. And actually, we can't say they're wrong. An accurate manual layout followed by test runs and estimation of minimal stable timings will have about the same effect as if we increased FSB frequency by 10 MHz. So, is it worth racking one's brains about accuracy if people who have high requirements for board performance are usually overclockers?

Back to the subject

Thus, we can say that our tests of the new boards were quite to the point, especially considering that the manufacturer focused so much on their increased speed. But we thought we could do with a rather small number of tests since we were not interested in global preformance aspects but just wanted to know about memory access speed in the new revisions. Therefore, we limited ourselves to four test subtypes: synthetic (RightMark Memory Analyzer), work with raster graphics in Adobe Photoshop, 7-Zip archiving, and games represented by Return to Castle Wolfenstein, Serious Sam: The Second Encounter, and Unreal Tournament 2003.

Performance tests

Testbed:

• CPU: Intel Pentium 4 3.2 GHz (16x200 MHz), Socket 478
• ASUS mainboards:
  • ASUS P4P800-E Wireless Edition
  • ASUS P4P800 SE
  • ASUS P4P800
  • ASUS P4P800S SE
  • ASUS P4P800S

• Memory: 2x256 MB PC4000(DDR500) DDR SDRAM DIMM OCZ, CL 2 in DDR400 mode
• Graphic card: Manli ATI Radeon 9800 Pro 256 MB
• HDD: Western Digital WD360 (SATA), 10000 rpm

Software:

OSs and drivers:

• Windows XP Professional SP1
• DirectX 9.0b
• Intel Chipset Software Installation Utility 5.0.2.1003
• ATI Catalyst 3.7

Test applications:

• RightMark Memory Analyzer 2.5
• Adobe Photoshop 7.0
• 7-Zip 3.13
• Gray Matter Studios & Nerve Software Return to Castle Wolfenstein v1.1
• Croteam/GodGames Serious Sam: The Second Encounter v1.07
• Digital Extremes/Epic Games/Atari Unreal Tournament 2003 v2225

The comparative table below contains brief characteristics of the tested boards:

Test results

RightMark Memory Analyzer: memory BW

It's easy to see that the greatest difference is not that between the first and the second revisions, but between boards based on a single-channel i848P and those on a dual-channel i865PE. It confirms the old truth that whatever the layout, the board on a dual-channel chipset will be faster than that on a single-channel one. But to be fair, we should say that second revisions are really a bit faster than the first ones, that is, the manufacturer lives up to his promise. The fact that the gain is not very significant can probably be accounted for by the impossibility to visibly improve a board if it's good enough as it is.

RightMark Memory Analyzer: latency

Here, the picture is more complicated. Both results taken together, P4P800S SE is really a little better than its predecessor, P4P800 / P4P800 SE show identical results, while the top-equipped P4P800-E Wireless Edition has maximal latency even 1ns higher than the simpler variant on the same chipset. However, the loss (if it is one) can be explained by the board's complexity, and it is even good that the price of increased functionality proved to be so low :).

Adobe Photoshop 7.0

The results match the general SE-is-a-little-better-but-nothing-extraordinary scheme. But then again, what can we expect of real applications if even low-level tests couldn't revealed anything extraordinary?

7-Zip archiving

The picture is similar to that in the case of Adobe Photoshop, despite a great difference between the task types and the absence of any connection between the programs. And that is, in fact, a yet another confirmation of our preliminary conclusions.

Game applications

The same thing is here: SE boards are sometimes better and never worse.

Conclusions

So, what can we say about ASUS's SE series within the P4P800 line? On one hand, some may be disappointed that it hasn't shown a huge performance growth compared to the first revisions. On the other hand, we have already mentioned that the better the product, the harder it is to modify it. ASUS boards have always been very fast, and fans of the trademark can treat the test results in the following way: in spite of all difficulties, the company still managed to get some more from the boards.

But let us be more pragmatic: it's not all about performance. Functionality has been enriched, the chips have been modified, and some of the boards (P4P800S SE) will become cheaper, if the manufacturer is to be believed. All this has been achieved at least with no decrease in performance. Is that not enough? Certainly, the idea behind the release of the SE series was to draw attention to ASUS and its products. But the way it was implemented was very subtle. Not just a lot of money was spent on the advertising campaign, but also some really significant modifications were introduced into the product line. And this approach can only be acclaimed as it is useful for both the manufacturer and the end customer.

Write a comment below. No registration needed!