Too Yuen Min

This works. ```julia julia> import VideoIO julia> using PyCall julia> py""" import cv2 """ ```

This fixes the issue. The main branch works fine with multi-threaded CSV reading. Thanks!

The versions of the dependencies: ``` (jl_yrusdV) pkg> st --manifest Status `C:\Users\Admin\AppData\Local\Temp\jl_yrusdV\Manifest.toml` [992eb4ea] CondaPkg v0.2.18 [692b3bcd] JLLWrappers v1.4.1 [0f8b85d8] JSON3 v1.12.0 [0b3b1443] MicroMamba v0.1.12 [69de0a69] Parsers v2.5.10 [fa939f87] Pidfile v1.3.0...

Xref: https://forums.developer.nvidia.com/t/error-when-using-torch-compile-on-jetson-orin-nx/297421

The torch version is 2.30 from https://developer.download.nvidia.com/compute/redist/jp/v60dp/pytorch/.

> This is tricky to fix. > > IIRC one needs to perform a `cas` loop for < 4 byte atomics. > > Atomix should use https://github.com/JuliaConcurrent/Atomix.jl/blob/e60c518e3ffd2c9d4e96104f16f2a970a69e4289/lib/AtomixCUDA/src/AtomixCUDA.jl#L38 > > Which...

The error still occurs after adding and loading `AtomixCUDA`. ```julia (jl_OV6Zim) pkg> st Status `/tmp/jl_OV6Zim/Project.toml` [a9b6321e] Atomix v0.1.0 [6171a885] AtomixCUDA v0.1.0-DEV `https://github.com/JuliaConcurrent/Atomix.jl#main:lib/AtomixCUDA` [052768ef] CUDA v5.4.2 [63c18a36] KernelAbstractions v0.9.22 julia> using...

`channels = ["nvidia/label/cuda-11.7.1"]` yields the correct cuda-nvcc version. For the following CondaPkg.toml ``` channels = ["nvidia/label/cuda-11.7.1", "conda-forge"] [deps.cuda-toolkit] channel = "nvidia/label/cuda-11.7.1" version = "11.7.1" ``` `conda up` installs the wrong...