Cuda runtime error (48) : no kernel image is available for execution, how to solve it?

[GarrisonD (Ihor Dotsenko)]

You can try with CUDA 8.0 and I believe it will solve your problem. Otherwise if you really need support of CUDA 9.0 or CUDA 9.1, drop me a line and I will help you to compile PyTorch from source with support for your GPU. Just did it yesterday for my GTX 760.

PyTorch with CUDA 9.1 is super slow compared to 9.0

Ah, thanks @smth ! I must say it’s a very weird case, I never got a warning when installing/importing it. And despite the fact that it detects the CUDA 9.1 lib, it does run on CPU (0% utilization when I check via nvidia-smi). When I install the CUDA 8.0 PyTorch version (also via conda), it works just fine though with 100% utilization.