Windows 11 で CUDA on WSL2 環境を構築する

#!/bin/bash
# Copyright (c) 2021 astherier
# This software is released under the MIT License.
# http://opensource.org/licenses/mit-license.php
#
#install-cuda-on-wsl.sh
#WSL2のUbuntu側で行うCUDA on WSLのセットアップスクリプト
#最終更新：2021/07/17
#https://astherier.com/blog/2021/07/windows11-cuda-on-wsl2-setup/

#CUDAリポジトリ名(DISTRO)とインストールするパッケージ名(APT_INSTALL)を指定してください。
#リポジトリ名を変更した場合、パッケージ名は明示的にCUDAツールキットを指定してください。
#(wsl-ubuntu以外のリポジトリから「cuda」をインストールすると、ディスプレイドライバーまでインストールしようとしてしまうため)
export DISTRO=wsl-ubuntu
export APT_INSTALL=cuda
#export DISTRO=ubuntu2004
#export APT_INSTALL=cuda-toolkit-11-4

#CUDAリポジトリの登録
sudo apt-key adv --fetch-keys http://developer.download.nvidia.com/compute/cuda/repos/${DISTRO}/x86_64/7fa2af80.pub
echo "deb http://developer.download.nvidia.com/compute/cuda/repos/${DISTRO}/x86_64 /" | sudo tee /etc/apt/sources.list.d/cuda.list
wget https://developer.download.nvidia.com/compute/cuda/repos/${DISTRO}/x86_64/cuda-${DISTRO}.pin
sudo mv cuda-${DISTRO}.pin /etc/apt/preferences.d/cuda-repository-pin-600

#CUDAツールキットのインストール
sudo apt update && sudo apt -y upgrade
sudo apt install -y ${APT_INSTALL}

#環境変数の設定
cat << 'EOS' >> ~/.profile

#Added by install-cuda-on-wsl.sh
#Ref: https://astherier.com/blog/2021/07/windows11-cuda-on-wsl2-setup/
export PATH=/usr/local/cuda-11.4/bin${PATH:+:${PATH}}
#Added: end

EOS

#サードパーティーライブラリのインストール（任意）
sudo apt install -y g++ freeglut3-dev build-essential libx11-dev libxmu-dev libxi-dev libglu1-mesa libglu1-mesa-dev

#サンプルプログラムのビルド（任意）
cd /usr/local/cuda/samples/1_Utilities/deviceQuery/
sudo make
cd /usr/local/cuda/samples/1_Utilities/bandwidthTest/
sudo make
cd /usr/local/cuda/samples/7_CUDALibraries/simpleCUBLAS/
sudo make
cd /usr/local/cuda/samples/7_CUDALibraries/simpleCUFFT/
sudo make

cd /usr/local/cuda/samples/bin/x86_64/linux/release/

echo "

CUDA on WSLのセットアップが終了しました。
いくつかサンプルをコンパイルしていますので、正常に動くか確認してください。
（例）
$ cd /usr/local/cuda/samples/bin/x86_64/linux/release/
$ ./deviceQuery

"

wget https://astherier.com/static/blog/2021-07-17/install-cuda-on-wsl.sh
source ./install-cuda-on-wsl.sh

export DISTRO=wsl-ubuntu
export APT_INSTALL=cuda

export DISTRO=ubuntu2004
export APT_INSTALL=cuda-toolkit-11-4

sudo apt-key adv --fetch-keys http://developer.download.nvidia.com/compute/cuda/repos/${DISTRO}/x86_64/7fa2af80.pub
echo "deb http://developer.download.nvidia.com/compute/cuda/repos/${DISTRO}/x86_64 /" | sudo tee /etc/apt/sources.list.d/cuda.list
wget https://developer.download.nvidia.com/compute/cuda/repos/${DISTRO}/x86_64/cuda-${DISTRO}.pin
sudo mv cuda-${DISTRO}.pin /etc/apt/preferences.d/cuda-repository-pin-600

sudo apt update && sudo apt -y upgrade
sudo apt install -y ${APT_INSTALL}

cat << 'EOS' >> ~/.profile

#Added by install-cuda-on-wsl.sh
#Ref: https://astherier.com/blog/2021/07/windows11-cuda-on-wsl2-setup/
export PATH=/usr/local/cuda-11.4/bin${PATH:+:${PATH}}
#Added: end

EOS

sudo apt install -y g++ freeglut3-dev build-essential libx11-dev libxmu-dev libxi-dev libglu1-mesa libglu1-mesa-dev

cd /usr/local/cuda/samples/1_Utilities/deviceQuery/
sudo make
cd /usr/local/cuda/samples/1_Utilities/bandwidthTest/
sudo make
cd /usr/local/cuda/samples/7_CUDALibraries/simpleCUBLAS/
sudo make
cd /usr/local/cuda/samples/7_CUDALibraries/simpleCUFFT/
sudo make

cd /usr/local/cuda/samples/
sudo make -j4

cd /usr/local/cuda/samples/bin/x86_64/linux/release/

$ ./deviceQuery
./deviceQuery Starting...

 CUDA Device Query (Runtime API) version (CUDART static linking)

Detected 1 CUDA Capable device(s)

Device 0: "NVIDIA GeForce GTX 960M"
  CUDA Driver Version / Runtime Version          11.4 / 11.4
  CUDA Capability Major/Minor version number:    5.0
  Total amount of global memory:                 4096 MBytes (4294967296 bytes)
  (005) Multiprocessors, (128) CUDA Cores/MP:    640 CUDA Cores
  GPU Max Clock rate:                            1176 MHz (1.18 GHz)
  Memory Clock rate:                             2505 Mhz
  Memory Bus Width:                              128-bit
  L2 Cache Size:                                 2097152 bytes
  Maximum Texture Dimension Size (x,y,z)         1D=(65536), 2D=(65536, 65536), 3D=(4096, 4096, 4096)
  Maximum Layered 1D Texture Size, (num) layers  1D=(16384), 2048 layers
  Maximum Layered 2D Texture Size, (num) layers  2D=(16384, 16384), 2048 layers
  Total amount of constant memory:               65536 bytes
  Total amount of shared memory per block:       49152 bytes
  Total shared memory per multiprocessor:        65536 bytes
  Total number of registers available per block: 65536
  Warp size:                                     32
  Maximum number of threads per multiprocessor:  2048
  Maximum number of threads per block:           1024
  Max dimension size of a thread block (x,y,z): (1024, 1024, 64)
  Max dimension size of a grid size    (x,y,z): (2147483647, 65535, 65535)
  Maximum memory pitch:                          2147483647 bytes
  Texture alignment:                             512 bytes
  Concurrent copy and kernel execution:          Yes with 4 copy engine(s)
  Run time limit on kernels:                     Yes
  Integrated GPU sharing Host Memory:            No
  Support host page-locked memory mapping:       Yes
  Alignment requirement for Surfaces:            Yes
  Device has ECC support:                        Disabled
  Device supports Unified Addressing (UVA):      Yes
  Device supports Managed Memory:                Yes
  Device supports Compute Preemption:            No
  Supports Cooperative Kernel Launch:            No
  Supports MultiDevice Co-op Kernel Launch:      No
  Device PCI Domain ID / Bus ID / location ID:   0 / 2 / 0
  Compute Mode:
     < Default (multiple host threads can use ::cudaSetDevice() with device simultaneously) >

deviceQuery, CUDA Driver = CUDART, CUDA Driver Version = 11.4, CUDA Runtime Version = 11.4, NumDevs = 1
Result = PASS

$ ./deviceQuery
./deviceQuery Starting...

 CUDA Device Query (Runtime API) version (CUDART static linking)

cudaGetDeviceCount returned 35
-> CUDA driver version is insufficient for CUDA runtime version
Result = FAIL

$ ./bandwidthTest
[CUDA Bandwidth Test] - Starting...
Running on...

 Device 0: NVIDIA GeForce GTX 960M
 Quick Mode

 Host to Device Bandwidth, 1 Device(s)
 PINNED Memory Transfers
   Transfer Size (Bytes)        Bandwidth(GB/s)
   32000000                     6.2

 Device to Host Bandwidth, 1 Device(s)
 PINNED Memory Transfers
   Transfer Size (Bytes)        Bandwidth(GB/s)
   32000000                     6.4

 Device to Device Bandwidth, 1 Device(s)
 PINNED Memory Transfers
   Transfer Size (Bytes)        Bandwidth(GB/s)
   32000000                     70.0

Result = PASS

NOTE: The CUDA Samples are not meant for performance measurements. Results may vary when GPU Boost is enabled.

$ ./simpleCUBLAS
GPU Device 0: "Maxwell" with compute capability 5.0

simpleCUBLAS test running..
simpleCUBLAS test passed.

$ ./simpleCUFFT
[simpleCUFFT] is starting...
GPU Device 0: "Maxwell" with compute capability 5.0

Temporary buffer size 448 bytes
Transforming signal cufftExecC2C
Launching ComplexPointwiseMulAndScale<<< >>>
Transforming signal back cufftExecC2C

CUDA error at bodysystemcuda_impl.h:186 code=304(cudaErrorOperatingSystem) "cudaGraphicsGLRegisterBuffer(&m_pGRes[i], m_pbo[i], cudaGraphicsMapFlagsNone)"