GPU is slower than CPU on my wsl

Accelerated Computing CUDA CUDA on Windows Subsystem for Linux
1

Hi.
system details:
windows 10: 2004 build 20175.1000
wsl 4.19.121
ubuntu: happens on both 16.04 and 20.04
cuda-toolkit on winsows 11.1
nvidia driver: 455.41
GPU: nvidia geforce gtx 1070
on the linux I installed: cuda-toolkit-11-0

I did everything as it was in the guide.
and I ran this Tensorflow notebook as a docker container:
docker run -it --gpus all -p 8888:8888 tensorflow/tensorflow:latest-gpu-py3-jupyter

when I use the GPU, it’s about 3 times slower than CPU. What could be the problem?

1 Like
2

First of all you need to update your nvidia Cuda Drivers to the new 460.15 and also your WSL2 kernel to 4.19.128.
(with wsl.exe --update from elevated cmd)

3

Thank you for your answer.
However, I did those steps but still, and the GPU performs poorly.

1 Like