My program run correctly with one GPU, however the same code give wrong result with one thread with dual GPUs (the other is still correct) , when i try to debug and it turn out some problem related to cudaArray and texture. In that operation I copy device memory to cudaArray that attached to a texture. I use that texture to perform bilinear interpolation.
I don’t know the problem from the cudaMemcpyToArray funtion, or my bilinear interpolation using texture, because i don’t know how to debug the contain of cudaArray and texture.
My question are :
- How can i debug the contain of cudaArray and texture
- How can I know the current device of the array and texture, to make sure that each one is generated per device context.
Can any one help help me. Thank you.