GPU smb

NVIDIA_RTXモデルのGPU使用状態確認手順



NVIDIA_RTXモデルのGPU使用状態確認手順

目的

NVIDIAのグラフィックボードRTXモデルのGPU使用状態をコマンドで確認する

参考

https://zenn.dev/190ikp/articles/how_to_install_nvidia_drivers

環境

  • OS: ubuntu
  • グラボ:NVIDIA GeForce RTX 3060
  • ドライバ:cuda-drivers
    • CUDA(Compute Unified Device Architecture:クーダ)とは、NVIDIAが開発・提供している、GPU向けの汎用並列コンピューティングプラットフォームおよびプログラミングモデルです。

手順

CUDAドライバを配布するリポジトリを登録してaptアップデートする

Ubuntu 18.04 LTSのCUDAリポジトリの公開鍵をダウンロードする

admin@server01:~$ curl -fsSL https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64/3bf863cc.pub | sudo gpg --no-default-keyring --keyring /usr/share/keyrings/nvidia-cuda.gpg --import
gpg: keybox '/usr/share/keyrings/nvidia-cuda.gpg' created
gpg: directory '/root/.gnupg' created
gpg: /root/.gnupg/trustdb.gpg: trustdb created
gpg: key A4B469963BF863CC: public key "cudatools <cudatools@nvidia.com>" imported
gpg: Total number processed: 1
gpg:               imported: 1
admin@server01:~$

GPG公開鍵を登録する

admin@server01:~$ sudo apt-key adv --keyserver keyserver.ubuntu.com --recv-keys A4B469963BF863CC
Executing: /tmp/apt-key-gpghome.dAdlzICwYw/gpg.1.sh --keyserver keyserver.ubuntu.com --recv-keys A4B469963BF863CC
gpg: key A4B469963BF863CC: public key "cudatools <cudatools@nvidia.com>" imported
gpg: Total number processed: 1
gpg:               imported: 1

NVIDIA CUDA ToolkitのAPTリポジトリを/etc/apt/souces.list.d/cuda.listに追加する

admin@server01:~$ echo "deb https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64 /" | sudo tee /etc/apt/sources.list.d/cuda.list
deb https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64 /

admin@server01:~$ ll /etc/apt/sources.list.d
total 56
drwxr-xr-x 2 root root 4096 Nov 21 11:45 ./
drwxr-xr-x 7 root root 4096 Apr  9  2021 ../
-rw-r--r-- 1 root root  138 Nov 21 11:29 amdgpu.list
-rw-r--r-- 1 root root  138 Nov 21 11:29 amdgpu.list.save
-rw-r--r-- 1 root root  190 Nov 21 11:29 amdgpu-proprietary.list
-rw-r--r-- 1 root root  189 Nov 21 11:29 amdgpu-proprietary.list.save
-rw-r--r-- 1 root root   81 Nov 21 11:45 cuda.list
-rw-r--r-- 1 root root  142 Nov 21 11:29 graphics-drivers-ubuntu-ppa-focal.list
-rw-r--r-- 1 root root   78 Nov 21 11:29 rocm.list
-rw-r--r-- 1 root root   78 Nov 21 11:29 rocm.list.save
-rw-r--r-- 1 root root  136 Nov 21 11:29 ubuntu-x-swat-ubuntu-ppa-focal.list
-rw-r--r-- 1 root root  136 Nov 21 11:29 ubuntu-x-swat-ubuntu-ppa-focal.list.save
-rw-r--r-- 1 root root  116 Nov 21 11:29 zabbix.list
-rw-r--r-- 1 root root  116 Nov 21 11:29 zabbix.list.save

admin@server01:~$ cat /etc/apt/sources.list.d/cuda.list
deb https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64 /

aptアップデート

admin@server01:~$ sudo apt update
:
W: GPG error: https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64  InRelease: The following signatures couldn't be verified because the public key is not available: NO_PUBKEY A4B469963BF863CC
E: The repository 'https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64  InRelease' is not signed.
N: Updating from such a repository can't be done securely, and is therefore disabled by default.
N: See apt-secure(8) manpage for repository creation and user configuration details.
E: The repository 'http://ppa.launchpad.net/ubuntu-x-swat/ppa/ubuntu focal Release' does not have a Release file.
N: Updating from such a repository can't be done securely, and is therefore disabled by default.
N: See apt-secure(8) manpage for repository creation and user configuration details.

エラーが出るので、以下コマンド実行。–recv-keysにエラーになった鍵を指定する

admin@server01:~$ sudo apt-key adv --keyserver keyserver.ubuntu.com --recv-keys A4B469963BF863CC
Executing: /tmp/apt-key-gpghome.dAdlzICwYw/gpg.1.sh --keyserver keyserver.ubuntu.com --recv-keys A4B469963BF863CC
gpg: key A4B469963BF863CC: public key "cudatools <cudatools@nvidia.com>" imported
gpg: Total number processed: 1
gpg:               imported: 1

CUDAドライバをインストールする

admin@server01:~$ sudo apt install -y cuda-drivers
Reading package lists... Done
Building dependency tree
Reading state information... Done
The following additional packages will be installed:

正常性確認

nvidia-smiが使用できること

admin@server01:~$ nvidia-smi
Tue Nov 21 12:46:24 2023
+---------------------------------------------------------------------------------------+
| NVIDIA-SMI 535.129.03             Driver Version: 535.129.03   CUDA Version: 12.2     |
|-----------------------------------------+----------------------+----------------------+
| GPU  Name                 Persistence-M | Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp   Perf          Pwr:Usage/Cap |         Memory-Usage | GPU-Util  Compute M. |
|                                         |                      |               MIG M. |
|=========================================+======================+======================|
|   0  NVIDIA GeForce RTX 3060        Off | 00000000:0D:00.0 Off |                  N/A |
| 30%   32C    P0              29W / 170W |      1MiB / 12288MiB |      0%      Default |
|                                         |                      |                  N/A |
+-----------------------------------------+----------------------+----------------------+
The following additional packages will be installed:
+---------------------------------------------------------------------------------------+
| Processes:                                                                            |
|  GPU   GI   CI        PID   Type   Process name                            GPU Memory |
|        ID   ID                                                             Usage      |
|=======================================================================================|
|  No running processes found                                                           |
+---------------------------------------------------------------------------------------+

CATEGORIES & TAGS

GPU, smb,

Author:


comment

メールアドレスが公開されることはありません。 が付いている欄は必須項目です

関連記事

カテゴリー

むるし

フリーランスのインフラ系エンジニア。
備忘録で色々書いていきます。
お問い合わせは↓
mo-gyu@murci.net
LPIC303 Security
%d人のブロガーが「いいね」をつけました。