My workstation has two identical CUDA enabled graphics cards. How to run in this same time the same program (kernel) on both cards with a different set of input data ?
Have you taken a look at the sample app simpleMultiGPU that ships with CUDA?
Just a few seconds ago I looked there - I have to “digest” this example - looks fine; Thenk you :)
For working with CUDA with .NET I created a DLL wrapper that works fine.
By the way, running the same kernel on a multi GPU: is it enough to run two tasks (threads) on .NET and in each of them to run with a different card run cudaSetDevice ?
I know nothing about .NET.