nVidia chips, what's the DC roadmap?

[CoW]8(0)]

I know ATi has its own F@H client. But nVidia seems to be getting into the general purpose computing frenzy as well with their new Tesla boxes. So what's in nVidia's future in the DC world? And specifically which of nVidia's GPU lines will make it there? (I'm really hoping for the 7xxx series).

 

[vraa]

http://forum.folding-community.org/ftopic16684.html

I expect to see ATi HD 2900 support before nvidia support, but I am hoping that nvidia support is not too far away for obvious reasons.

FWIW, I have switched from nvidia only to ATi only due to being able to run F@H on the cars. Vote with your dollars (says the idealistic libertarian inside of me).

 

[CoW]8(0)]

But in distributed computing in general, what's going on with nVidia? I heard rumors of a Seti client for nVidia GPUs. But I can't find any hard evidence.

 

[ILAMtitan]

At this point AIT provides better support for offloading tasks. Once nVidia integrates that kind of feature into their drivers we will probably start seeing more applications that utilize the power of your GPU when your not playing games. However, until there is a unified standard for the offloading, we probably wont see may applications using this; the case is very similar to the PPU debate at this point.

 

[Xilikon]

MHouston, the guy working at F@H on the GPU folding side, said that currently, the problem is a poor implementation of GPGPU in the drivers so nvidia need to fix this first before we can use this for any kind of general purpose computation.

However, on the bright side, nividia is willing to work on this since he will want to push GPGPU in the mainstream and need some proof of concept to help. But we dunno how much time they will take.