Rotate-and-Render icon indicating copy to clipboard operation
Rotate-and-Render copied to clipboard

no kernel image is available for the execution of device

Open ekantbagri opened this issue 3 years ago • 4 comments

I installed pytorch 1.2 torchvision 0.4 for the neural_renderer but when I ran the v100_test.sh file, it said no kernel image is available for the execution of device. I searched online & it appeared that it is because of the cuda torch compatibility issue but when I upgraded my pytorch version, the neural_renderer doesn't work. I'm stuck, can you help me with this? I'm attaching my error output here as well. Screenshot (46)

ekantbagri avatar Oct 18 '21 06:10 ekantbagri

"https://github.com/daniilidis-group/neural_renderer/tree/projection_fix" git clone this repo and run the setup.py file, worked for me.

ekantbagri avatar Oct 18 '21 09:10 ekantbagri

Hello ! I'm having a similar issue on google colab, but your workaround did not fix the issue in my case, do you have any other lead on this ? I have a Tesla K80 GPU.

MaxLmqr avatar Feb 08 '22 09:02 MaxLmqr

I solved it by upgrading to google colab pro, it looks like the issue was indeed related to a compatibility issue between the GPU used and CUDA. Switching to pro gave me access to another kind of GPU (Tesla T4) and the error went away.

MaxLmqr avatar Feb 08 '22 17:02 MaxLmqr

you don't need to upgrade to colab pro, just use install the neural renderer 'projection_fix' branch. note: you might need to change the v100_test.sh file (gpu_id to 0,1 or 0) as it by default uses multi gpus. hope it helps!

ekantbagri avatar Feb 08 '22 17:02 ekantbagri