AI
Tuesday, February 01, 2005 12:48:16 PM
Re: I think there's a mis-conception about Dual Core power in general. It appears that there is a general opinion among the population, that if 1 core = 80watts, then 2 cores must = 160 watts. This is not so. AFAIK, a slight reduction in linked core clock speed, can result in huge reduction of power consumption, while still improving performance, so linking 2 cores together, and slightly reducing the clock speed, can be made to result in NO Additional Power consumption, over the original single chip/higher speed. I think the naysayers are going to be quite surprised at the power consumption numbers of dual core. Wait, you'll see 
Semi, I've discussed this before on this forum. If you pick the lower leakage parts and cut the voltage and frequency, you can get very good power dissipations from dual core chips without sacrificing more than 3-4 bins of scalar performance.
By the way, I think the proof is in the pudding. Prescott maxed out around 3.8GHz, but Smithfield is on the roadmap at 3.2GHz, which is minus 3 bins. Similarly, if AMD gets out dual core chips at 2.4GHz, I would expect single core speeds of up to 3.0GHz.
Semi, I've discussed this before on this forum. If you pick the lower leakage parts and cut the voltage and frequency, you can get very good power dissipations from dual core chips without sacrificing more than 3-4 bins of scalar performance.
By the way, I think the proof is in the pudding. Prescott maxed out around 3.8GHz, but Smithfield is on the roadmap at 3.2GHz, which is minus 3 bins. Similarly, if AMD gets out dual core chips at 2.4GHz, I would expect single core speeds of up to 3.0GHz.
