- Joined
- Jul 19, 2002
- Location
- WPAFB, OH
This topic is more "Specialized Computing" however there's much more visibility here, sorry Mod's.
Has anyone been keeping up with the Nvidia TESLA launch? Link
I'm planning to get several TESLA C870's for dev work when they arrive in August. However even after blasting through whatever documentation is available and posing some Q/A's with Nvidia, theres still alot of gray area questions about these. So I figured I might start up a discussion with some of the brains on the forum so I can share what I learn and pose some questions. Maybe some of you are interested in these as well.
- One thing I have yet to have answered is whether or not the CUDA environment would utilize two C870 cards on a board with two PCIExpress x16/x8 slots, or is there only native support in the guts after Nvidia crams two of them into the D870 or four into the S870 (Link and bumps up the price by $4100 (2x C870's = $3000, 1x D870 = $7000)?
- I've also seen it noted that the next iteration will feature double precicion floating point (Arstechnica - Link) but haven't been able to confirm it with Nvidia.
Has anyone been keeping up with the Nvidia TESLA launch? Link
I'm planning to get several TESLA C870's for dev work when they arrive in August. However even after blasting through whatever documentation is available and posing some Q/A's with Nvidia, theres still alot of gray area questions about these. So I figured I might start up a discussion with some of the brains on the forum so I can share what I learn and pose some questions. Maybe some of you are interested in these as well.
- One thing I have yet to have answered is whether or not the CUDA environment would utilize two C870 cards on a board with two PCIExpress x16/x8 slots, or is there only native support in the guts after Nvidia crams two of them into the D870 or four into the S870 (Link and bumps up the price by $4100 (2x C870's = $3000, 1x D870 = $7000)?
- I've also seen it noted that the next iteration will feature double precicion floating point (Arstechnica - Link) but haven't been able to confirm it with Nvidia.
