An x86 PlayStation 4 could signal a sea-change in the console industry

[Evilsizer]

yes with gpus getting faster and faster, you still need a cpu with a good push to feed them. it takes a pretty fast quad core i7 to get the most out of a 3 way sli setup. you would be kidding your self if you said you could get the same performance from the 3 way sli setup with a dual core. you still have to have the balance, gpu might be come more and more involed in our computing. it will never replace a cpu, intel killed what would have been there new entry [ala larrabee] into the video card market. what reasons unknown to me, they killed going to market with it as a video card and it is now a add-in compute device. would have been interesting to see what kind of performance it would have had. not all was lost, some of the tech/ideas did make their way into the onboard cpu video. intel is just a bit slow to update those to be competive in the cpu-video area.

 

[mjw21a]

yes with gpus getting faster and faster, you still need a cpu with a good push to feed them. it takes a pretty fast quad core i7 to get the most out of a 3 way sli setup. you would be kidding your self if you said you could get the same performance from the 3 way sli setup with a dual core. you still have to have the balance, gpu might be come more and more involed in our computing. it will never replace a cpu, intel killed what would have been there new entry [ala larrabee] into the video card market. what reasons unknown to me, they killed going to market with it as a video card and it is now a add-in compute device. would have been interesting to see what kind of performance it would have had. not all was lost, some of the tech/ideas did make their way into the onboard cpu video. intel is just a bit slow to update those to be competive in the cpu-video area.

No, though I think AMD has the right idea with integrating the GPU component, and an even better idea in changing from VLIW5 to GCN which will make the GPU much more versatile for compute type tasks. Whats even more interesting is that nVidia still hasn't done anything similar. Their CUDA still only utilizes VLIW. If they're not careful they'll be left behind too.