Originally Posted by TyrantII
To be fair it's actually only a straight 17% edge, because 4 cores are said to be stashed away for OS and other OS uses. Still, Sony tends to push their Devs to code to the hardware more and fully utilize their the hardware around their API's, so there seems to be some wiggle room to use those cores in addition. but who knows, maybe they lock it out as they did with a dedicated OS SPU in the PS3.
Another interesting thing to think about is Sony will still be supporting PSGL (OpenGL ES) and LibGCM, meanwhile MS obviously is pushing DX11 X. So there still will be slight rendering differences in the end, but the parity will be much closer.
Where did you read that? The eurogamer article indicated that 4 of the GPU's CUs were dedicated to compute/GPGPU, not the OS. Normally modern GPUs can just assign it to rendering/compute at will, but this one just has a cordoned off compute block, probably for some sort of efficiency reason. In either case games are going to heavily use it - whatever code is written to utilize those compute blocks on the PS4 is going to have to eat away from the 360's shaders/rendering. It's entirely possible that the majority of games will use 20-30% of the GPU for this kind of code, especially physics. So that full 50% CU advantage should be realized. Plus there's twice the ROPs, and that memory bandwidth advantage.
They can spin the numbers all they want, but if the specs are what we've been seeing, the PS4 is significantly more powerful, and everyone will know it. I really don't think the differences will be subtle this time.