pytorch - docker THCudaCheck FAIL no kernel image is available for execution on the device error (pytorch_1565272279342/work/aten/src/THC/THCGeneral.cpp)

Question

FROM nvidia/cuda:10.0-cudnn7-runtime-ubuntu16.04

... /opt/conda/bin/conda install -y -c pytorch cudatoolkit=10.0 "pytorch=1.2.0=py3.6_cuda10.0.130_cudnn7.6.2_0" torchvision=0.4.0 && 

and error code is

HCudaCheck FAIL file=/opt/conda/conda-bld/pytorch_1565272279342/work/aten/src/THC/THCGeneral.cpp line=371 error=48 : no kernel image is available for execution on the device

RuntimeError: cuDNN error: CUDNN_STATUS_EXECUTION_FAILED

It's docker file so it shouldn't be related to my cuda version. I checked it works well with 2080 ti GPU but failed with 3090 GPU card. It has been installed same cuda (11.2.0) and cudnn.

Has anyone solved this problem before?

question from:https://stackoverflow.com/questions/65929726/docker-thcudacheck-fail-no-kernel-image-is-available-for-execution-on-the-device

Answer 1

Waitting for answers