Conclusion: Asus Gets The Glory, Intel Gets The Headaches
After the last few days, in which there has been much discussion about factory overclocking and tuning, and MSI made headlines with its 865 Neo 2 , we now have Asus in the picture with its own issues to add. The manufacturer sent us a P4P800 to test, which turns out to be quite different from the version that's available in stores. After contacting Asus in Taiwan, we confirmed some of our suspicions: the manufacturer has succeeded in activating the PAT optimization with the Intel 865PE chipset. However, after Intel intervened, Asus is no longer allowed to call it "PAT" and now calls it "Hyper Path" instead. And so the excellent benchmark results, which are at the level of the expensive Intel 875P, can now be verified for both press samples and retail products.
There are differences between the boards, but we now realize that performance remains the same. We'd also like to revise our statement in our last test comparison that PAT technology doesn't have much of an effect. It is certain that Intel partially puts the brakes on the 865PE chipset (Springdale) in order to justify the price difference with the Intel 875P (Canterwood). Intel thought that it was correct in saying that PAT could not be activated with the 865PE because of details in the hardware. But this proved to be incorrect - as we have shown. Inadvertently, Intel had left a back door open, which remains unfathomable to THG. In the past, Intel has gone to great lengths to put a lock on the multiplier, and the chipsets.
What Asus has done may be good for the consumer, but will cause headaches for Intel. Obviously, with the price difference of the two chipsets being $16 (when bought in 1000 units) translating into a differential of $55 for the mobo, Intel loses out. As long as the buyer goes for the Asus board, since other manufacturers have not mastered this technology yet, the price/ performance advantage is clear. Are the profits for the 875P/ Canterwood chipset in any danger? From a strategic point of view, it's more important for Intel to fulfill the demands of the market for chipsets across all product types. The lower price may not be as relevant, because in the end, Intel's wider range of products is what keeps the competition at bay, as well.
For Asus the problem remains 865PE chips that don't pass the internal Intel PAT qualification tests through speed binning, and others that are artificially degraded in order to fulfill market demand. Only the latter type is suitable for stable operation after "reactivating" PAT. Asus wants to ensure its success by using strict quality control. That should be well within the realms of consumer trust because of the masterful job that Asus did in investing its own resources in a product in order to offer the customer better value compared to its competitors. However, the competition is not resting idle, either. It is highly possible that copies of this technology can be expected soon from other mobo makers.
Update, June 4, 2003: a day after this article was posted, the BIOS version 1.3 Beta for the Abit IS7-G, a rival product, arrived. The initial tests show that Abit has made the same leap in performance and is now at the same level as Asus. Thus, the frame rates in Quake 3 have increased from 366 to 418 fps - this may be a sign that Abit has also activated PAT.