I can not create a docker container for cuda 9 which has pycuda

Accelerated Computing CUDA CUDA Setup and Installation
,
1

Environment
Docker nvidia/cuda:9.0-devel-ubuntu16.04
python 3.7
pycuda version 2018.1.1
Cuda version 9 ( Azure Container Instance Limitation )

I get this error

=> [80/80] RUN python3.7 -m pip install pycuda==2017.1.1 --no-cache-dir --user 80.4s
=> => # return PyUnicode_Check(obj) ? _PyUnicode_AsString(obj) : 0;
=> => # ^
=> => # error: command ‘x86_64-linux-gnu-gcc’ failed with exit status 1
=> => # ----------------------------------------
=> => # ERROR: Failed building wheel for pycuda
=> => # Running setup.py clean for pycuda

2

Hello @user84330 and welcome to the NVIDIA developer forums!

Did you make sure you have compatible versions of python and dependent libraries as well as kernel modules in your docker container? That error message seems more related to a python library mismatch than related to pyCUDA specifically.

I hope it is ok for you if I move this topic to the correct CUDA category. That way it will be seen by more people with CUDA setup experience.

1 Like
3

How do i know the compatible versions of python and dependent libraries as well as kernel modules in your docker container ?
Environment
Docker nvidia/cuda:9.0-devel-ubuntu16.04
python 3.7
pycuda version 2018.1.1
Cuda version 9

I have also asked here without any luck