Thursday, May 29, 2003
We got the following response from Intel regarding the PAT feature with the 865 chipset; however, its content is open to debate:
|George Alfs (Intel Corporation):No, PAT is an integral part of the 875P chipset, which is set as part of the manufacturing process (maybe a terminology difference, but this is done by manufacturing, not by "QA".) You are right that the silicon is speed binned to shave off two clocks in memory latency/timing. 875P also enables the possibility to use ECC, hence more "balls" or contacts necessitating a different package....You are going to need to ask Asus why they say this. PAT is hardwired in the chipset, and cannot be enabled on 865 based platforms.|
So, Asus gave the hardware enthusiast the joyous news that PAT can be activated with all P4P800 motherboards, making us all wonder what the sense is in having the 875P chipset at all. Not surprisingly, Intel requested that Asus promptly remove their press release about the 865 chipset. Because of this, the URL mentioned previously is no longer linked on the Asus website.
Friday, May 30, 2003
After discussing the issue with Asus, at the highest levels, Asus made it clear to us that the press release was removed for political reasons, but, technically speaking, the PAT optimization was still available. Therefore, Asus was basically strong-armed by Intel to take back the "PAT-announcement," which is why Asus renamed it something to the tune of "HyperPath," even though there's no difference technically. In the BIOS of the P4P800, a menu item called "Innovation Performance Mode" should appear. How this feature will be marketed is still not clear.
Monday, June 2, 2003
Asus introduces a new BIOS version for the P4P800, namely the 1007.001, replacing all of the older BIOS versions, which are no longer available. THG tests the retail boards once more with the 1007 BIOS. And the result is that all P4P800s run equally fast. Compared to the earlier tests, the retail boards now have performance that's identical to the press sample.
This was a very positive turn of events for us, because we were concerned that the retail variants are equipped with fewer components than our press sample. But Asus provided a clear explanation for this: samples for the press and engineering are frequently identical and have a complete set of components. It's only when a board goes on the production line that selected elements are left out which do not affect performance. This helped to clear up any confusion that had existed, and we were able to verify the results in our follow up tests.
As you can see, having discovered anomalies in boards from two vendors, we were very concerned about the implications. We've addressed the issue with MSI, and now we can confidently address what is happening with Asus, having worked closely with Asus to clear up any misunderstandings. We were very fortunate in having Asus open up to us on this subject because, in sharing its findings with us, Asus leaves their uniquely discovered feature available to their competitors. We have to give kudos to the folks at Asus, who face both the wrath of Intel, and reducing their unique position in this situation.