Joe, Re: The assertion in Infoworld that the dual core K8 will fit within 89W (55W and 30W were also mentioned) creates a bar that Prescott may never reach.
Dual core Prescott is not going to reach the power levels of dual core Opteron on the same process. I don't think Intel is even going to try to compete on the merit of low power with their first dual core designs.
Depending on how far they can scale the voltage and frequency, they may attempt a dual core 90nm Prescott and sacrifice scalar performance for the capability of dual core, but I think it's far more likely that they will use their 65nm process to manufacture their first dual core chips. This would lower power dissipation of a single Prescott to the 65-70W range, making a dual core version 130-140W.
Dual core Opteron's power will depend on how much it dissipates at 90nm. It's possible that AMD has been sandbagging the power of their single core 90nm chips, and that these will be in the range of 50-55W. That would put a dual core version around the range that the first rumors leaked - about 104W. I think this is distinctly possible.
Re: When Intel initially announced the Tejas cancellation and persuit of dual core, all here assumed that it meant Dothan and its derivatives. Now, the information that is emerging is that dual Prescott may be the first dual-core chip, which leaves me puzzled.
I never believed this. While Dothan is a step forward for Intel in lower power and IPC, it's a step backwards in terms of a number of other features, including EM64T and SSE-3. I'm thinking that Dothan's successor, Merom, will get these features. That is what the Inq reported months ago, and other reports came out later to confirm this. The desktop version of Merom, called Conroe, will be the desktop version, and Whitefield will be the server version. Until then, AMD has the low power advantage, and possibly a performance advantage as well. Now is a good time to be invested in AMD.
