Using GPU with Julia

I wanted to use Julia package CUDA.jl on computing clusters. But when I precompile the pkg, I got:

Error: CUDA.jl could not find an appropriate CUDA runtime to use.
This can have several reasons:
* you are using an unsupported platform: this version of CUDA.jl
  only supports Linux (x86_64, aarch64, ppc64le) and Windows (x86_64),
  while your platform was identified as 'something';
* you precompiled CUDA.jl in an environment where the CUDA driver
  was not available (i.e., a container, or an HPC login node).
  in that case, you need to specify which CUDA version to use
  by calling `CUDA.set_runtime_version!`;
* you requested use of a local CUDA toolkit, but not all
  required components were discovered. try running with
  JULIA_DEBUG=all in your environment for more details.

For more details, refer to the CUDA.jl documentation at
https://cuda.juliagpu.org/stable/installation/overview/
@ CUDA /software/opt/el_9/x86_64/julia/1.10.0/local/share/julia/packages/CUDA/rXson/src/initialization.jl:82

So I searched for this issue and got what I want from here.¹

For the same case as mine, one should first check the CUDA version number of your device by running nvidia-smi in BASH and/or using nvcc –version. For me, I used the latter and got

nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2022 NVIDIA Corporation
Built on Tue_May__3_18:49:52_PDT_2022
Cuda compilation tools, release 11.7, V11.7.64
Build cuda_11.7.r11.7/compiler.31294372_0

Then I run:

julia> using CUDA; CUDA.set_runtime_version!(v"11.7")

I got the same error but with additional info at the end:

Info: Configure the active project to use CUDA 11.7; please re-start Julia for this to take effect.

At last, I restarted Julia and typed ‘using CUDA’ again. Then the pkg precompiled successfully.

1. https://discourse.julialang.org/t/cuda-could-not-find-an-appropiate-cuda-runtime-to-use/97201/3 ↩