considering I only look on this thread, no. Why would I look elsewhere when its right here?

"Cards that offer only compute capabilities at 1.0 level are lacking a hardware features (support for Atomic functions) that Cuda 2.0 relies on, hence Cuda 2.0 compiled apps won't run."

Cuda 2.0 compiled apps will run on cards with compute capabilities at 1.0 level as long as you don't use asynchronous memory copies and atomic global operations.

The FAQ would be right and concise if it said: "Cards that offer only compute capabilities at 1.0 level are lacking a hardware features (support for Atomic functions) that THIS PROJECT relies on".


Atomic functions are used on this project?
If not, the G80 cards are able to run it.
Otherwise, make it clear that this app, instead of any Cuda 2.0 compiled apps won't run.

Good to know, thanks for the information :)
Honestly, I have no clue about Cuda, just collected the cards that would run the app and tried to explain the problem of the G80 without actually knowing for sure.
Unfortunatelly, I can't change my post anymore...


Btw, meanwhile there is also a GTX 260 with 216 Shader units...

I am wondering if refining the list further is possible given that results with 32 shader cards appear to be meeting the 4-day deadline with some room to spare (9500GT and 8600GT cards can be seen in user results with 2-3 day returns...I would assume that the 8600GTS would have similar times). Maybe with a note that these cards can work, but require a 24/7 computation. And clearly given these results, the 16 shader cards would never be able to meet the 4-day deadline.