stonedwaldo420
Limp Gawd
- Joined
- Mar 30, 2005
- Messages
- 505
I think I understand the concept behind why most distributed computing applications cannot fully utilize the computational power of a graphics card. I believe it is because they cannot be separated into 16 to 24 separate threads, all with outcomes independent of one another.. or at least that is what it would take to fully utilize a GPU for folding.
However, why not use just one pipeline of the graphics card???
Anyone know why they dont come up with an ad-on for the folding client that downloads a second smaller work unit and uses just one pipeline of the graphics card to crunch it, along with the CPU crunching its regular work unit? I understand that it would only be using a small portion of the GPU, but that is still one processor running at 350-450MHZ, with 256MB of 1 1.3 GHZ RAM. I am sure that would do something. Obviously, the second GPU folding client would shut down completely during any 3D gaming application, so as not to take up any GPU memory.
It just bugs the hell out of me that I have this MASSIVE, incredibly fast processor with its own memory subsystem, just sitting idle while people die of horrible diseases.
HARD FOLDING FOREVER!!!!
However, why not use just one pipeline of the graphics card???
Anyone know why they dont come up with an ad-on for the folding client that downloads a second smaller work unit and uses just one pipeline of the graphics card to crunch it, along with the CPU crunching its regular work unit? I understand that it would only be using a small portion of the GPU, but that is still one processor running at 350-450MHZ, with 256MB of 1 1.3 GHZ RAM. I am sure that would do something. Obviously, the second GPU folding client would shut down completely during any 3D gaming application, so as not to take up any GPU memory.
It just bugs the hell out of me that I have this MASSIVE, incredibly fast processor with its own memory subsystem, just sitting idle while people die of horrible diseases.
HARD FOLDING FOREVER!!!!
