Originally Posted by DaCrusader
The reason why nVidia cards fold better, is because stanford has used CUDA to use all the raw power of the card... which ATi doesn't have.
This makes no sense at all.
the GPU2 Client for AMD/ATi SM4 was released months before the one for NV SM4 Chips.
Do programmers of nVidia and AMD support Stanford.
The Code for the Cores of GPU Clients are mainly written by the companies people.
Is AMD/ATi using CAL as higher level API to access the Shader Units (besides CTM).
It might not be as comfortable as CUDA.
But a skilled Programmer focused on one target (and not bound by timeframe and cost issues - what a dream^^) like Stanford and the people of companies supporting them has, can easily use it like CUDA.
The difference between ATi and NV in folding is a result of different Hardware.
My last knowledge from reading (that doesn't mean it is TEH truth) was that the current Radeons have a bottleneck on the Hardware Shader Code sheduler.
Not a problem for Gaming, Video Encoding or most other Shader Code but making it hard to optimize for folding.
Besides the fact that the current ATi Core is not fully using the 800 SU's of HD48xx Cards.Edited by MESeidel - 12/14/08 at 7:16pm