{"id":61844,"date":"2024-12-24T09:44:05","date_gmt":"2024-12-24T01:44:05","guid":{"rendered":"http:\/\/www.biocloudservice.com\/wordpress\/?p=61844"},"modified":"2025-02-27T19:26:32","modified_gmt":"2025-02-27T11:26:32","slug":"%e2%98%86%e2%98%86slurm%e5%bf%85%e8%af%bb%e2%98%86%e2%98%86","status":"publish","type":"post","link":"http:\/\/www.biocloudservice.com\/wordpress\/?p=61844","title":{"rendered":"\u2606\u2606slurm\u5fc5\u8bfb\u2606\u2606"},"content":{"rendered":"\n

\u96c6\u7fa4\u4fe1\u606f(\u9010\u6b65\u589e\u52a0)<\/h1>\n\n\n\n
\n
\n
\u8282\u70b9<\/strong><\/th>CPU<\/strong><\/th>\u5185\u5b58<\/strong><\/th>\u663e\u5361<\/strong><\/th><\/tr><\/thead>
iyun105<\/td>72 \u6838\u5fc3<\/td>256 GB<\/td>NVIDIA Tesla P100 PCIe 16 GB(\u7ef4\u62a4\u4e2d)–\u5df2\u4e0b\u67b6<\/td><\/tr>
iyun103<\/td>72 \u6838\u5fc3<\/td>256 GB<\/td>NVIDIA Tesla P40 24 GB(\u7ef4\u62a4\u4e2d)–\u5df2\u4e0b\u67b6<\/td><\/tr>
iyun110<\/td>56 \u6838\u5fc3<\/td>128 GB<\/td>NVIDIA Tesla P100 PCIe 16 GB * 4(\u56db\u5f20\u5361)<\/td><\/tr><\/tbody><\/table><\/figure>\n<\/div>\n<\/div>\n\n\n\n

\u4e0a\u8ff0\u914d\u7f6e\u4e3a\u663e\u5361\u96c6\u7fa4\u8282\u70b9\u4fe1\u606f, \u670d\u52a1\u5668\u72ec\u4eab\u7528\u6237\u4e5f\u53ef\u4ee5\u5355\u72ec\u90e8\u7f72<\/strong>, \u72ec\u4eabslurm(\u663e\u5361\u9700\u989d\u5916\u9009\u79df)<\/h1>\n\n\n\n

\u901a\u77e5\u90ae\u7bb1\u4e3a: rainsc@yeah.net<\/p>\n\n\n\n

\u73af\u5883\u95ee\u9898\u4f7f\u7528conda\u89e3\u51b3 \u6545\u9700\u8981\u63d0\u524d\u6fc0\u6d3b\u73af\u5883\u540e\u8fd0\u884c\u4e0b\u9762\u811a\u672c \u811a\u672c\u540d\u79f0\u6700\u597dslurm\u7ed3\u5c3e \u4fbf\u4e8e\u533a\u5206 sh\u4e5f\u80fd\u8fd0\u884c <\/p>\n\n\n\n

\u793a\u4f8b\u8c03\u7528 sbatch gpu_test103.slurm<\/p>\n\n\n\n

#!\/bin\/bash\n# \u4ee5\u4e0a\u884c\u6307\u5b9a\u4e86\u811a\u672c\u4f7f\u7528\u7684\u89e3\u91ca\u5668\u4e3a \/bin\/bash\n\n#SBATCH --job-name=gpu_test # \u8bbe\u7f6e\u4f5c\u4e1a\u540d\u79f0\u4e3a \"gpu_test\"\n#SBATCH --partition=gpu # \u6307\u5b9a\u4f5c\u4e1a\u63d0\u4ea4\u5230\u54ea\u4e2a\u5206\u533a\uff08\u6216\u961f\u5217\uff09\n#SBATCH --ntasks=1 # \u8bbe\u7f6e\u5e76\u884c\u4efb\u52a1\u7684\u6570\u91cf\u3002\u8fd9\u91cc\u662f 1\uff0c\u8868\u793a\u8fd9\u662f\u4e00\u4e2a\u5355\u4efb\u52a1\u4f5c\u4e1a\n#SBATCH --cpus-per-task=1 # \u4e3a\u6bcf\u4e2a\u4efb\u52a1\u5206\u914d\u7684 CPU \u6838\u5fc3\u6570\u3002\u8fd9\u91cc\u662f 1\uff0c\u8868\u793a\u6bcf\u4e2a\u4efb\u52a1\u53ea\u9700\u8981\u4e00\u4e2a CPU \u6838\u5fc3\n#SBATCH --gres=gpu:1 # \u8fd9\u91cc\u662f\u8bf7\u6c42 1 \u4e2a GPU \u8bbe\u5907\n#SBATCH --output=gpu_test_%j.out # \u8bbe\u7f6e\u6807\u51c6\u8f93\u51fa\u6587\u4ef6\u7684\u8def\u5f84\u548c\u540d\u79f0\u3002%j \u4f1a\u88ab\u66ff\u6362\u4e3a\u4f5c\u4e1a ID\uff0c\u786e\u4fdd\u6bcf\u4e2a\u4f5c\u4e1a\u7684\u8f93\u51fa\u6587\u4ef6\u662f\u552f\u4e00\u7684\n#SBATCH --error=gpu_test_%j.err # \u8bbe\u7f6e\u6807\u51c6\u9519\u8bef\u8f93\u51fa\u6587\u4ef6\u7684\u8def\u5f84\u548c\u540d\u79f0\u3002\n#SBATCH --mail-user=\u63a5\u6536\u901a\u77e5\u7684\u90ae\u7bb1\u53f7 # \u8bbe\u7f6e\u63a5\u6536\u4f5c\u4e1a\u72b6\u6001\u90ae\u4ef6\u901a\u77e5\u7684\u90ae\u7bb1\u5730\u5740\u3002\u5f53\u4f5c\u4e1a\u5f00\u59cb\u3001\u7ed3\u675f\u6216\u51fa\u9519\u65f6\uff0c\u4f1a\u5411\u8fd9\u4e2a\u90ae\u7bb1\u53d1\u9001\u901a\u77e5\n#SBATCH --mail-type=ALL #\u9700\u8981\u53d1\u9001\u7684\u90ae\u4ef6\u7c7b\u578b\n#SBATCH --nodelist=iyun110  # \u6307\u5b9a\u4f5c\u4e1a\u8fd0\u884c\u7684\u8282\u70b9iyun103  \u6216 iyun105\n\n# \u9700\u8981\u6267\u884c\u7684\u4ee3\u7801\u90e8\u5206\n\npython gpu_test.py  # \u6267\u884c\u4f60\u7684Python\u811a\u672c\n<\/code><\/pre>\n\n\n\n
# \u6587\u4ef6 gpu_test.py\n\nimport torch\nimport time\n\n# \u786e\u4fdd\u4f7f\u7528 GPU\ndevice = torch.device(\"cuda:0\" if torch.cuda.is_available() else \"cpu\")\n\n# \u6253\u5370\u4f7f\u7528\u7684\u8bbe\u5907\nprint(f\"Using device: {device}\")\n\n# \u521b\u5efa\u4e24\u4e2a\u968f\u673a\u77e9\u9635\na = torch.rand((10000, 10000), device=device)\nb = torch.rand((10000, 10000), device=device)\n\n# \u6d4b\u8bd5\u8fd0\u884c\u65f6\u95f4\nprint(\"Starting matrix multiplication...\")\nstart = time.time()\n\nfor i in range(1000):\n    if i % 100 == 0:  # \u6bcf100\u6b21\u8f93\u51fa\u4e00\u6b21\u65e5\u5fd7\n        print(f\"Iteration {i}...\")\n    c = torch.matmul(a, b)\n\ntorch.cuda.synchronize()  # \u786e\u4fdd\u540c\u6b65\nend = time.time()\n\n# \u8f93\u51fa\u8fd0\u884c\u65f6\u95f4\nprint(f\"Completed 1000 matrix multiplications.\")\nprint(f\"Time taken for 1000 matmul: {end - start} seconds\")\n<\/code><\/pre>\n\n\n\n
\u63d0\u4ea4\u793a\u4f8b<\/p>\n\n\n\n
# \u9996\u5148\u6fc0\u6d3b\u8f6f\u4ef6\u73af\u5883\nrainsc@iyun105:~$ conda activate pytorch_test\n(pytorch_test) rainsc@iyun105:~$ sbatch gpu_test103.slurm\nSubmitted batch job 234\n(pytorch_test) rainsc@iyun105:~$ sbatch gpu_test105.slurm\nSubmitted batch job 235<\/code><\/pre>\n\n\n\n
\u8282\u70b9\u7a7a\u95f2\u4f1a\u6709\u4efb\u52a1\u5f00\u59cb\u7684\u63d0\u793a<\/p>\n\n\n\n
\"\"<\/figure>\n\n\n\n
\u67e5\u770b\u961f\u5217<\/p>\n\n\n\n
(pytorch_test) rainsc@iyun105:~$ squeue\n             JOBID PARTITION     NAME     USER ST       TIME  NODES NODELIST(REASON)\n               234       gpu gpu_test   rainsc  R       0:13      1 iyun103\n               235       gpu gpu_test   rainsc  R       0:02      1 iyun105<\/code><\/pre>\n\n\n\n
\u67e5\u770b\u4efb\u52a1<\/p>\n\n\n\n
(pytorch_test) rainsc@iyun105:~$ sacct\nJobID           JobName  Partition    Account  AllocCPUS      State ExitCode\n------------ ---------- ---------- ---------- ---------- ---------- --------\n235            gpu_test        gpu                     2    RUNNING      0:0\n235.batch         batch                                2    RUNNING      0:0<\/code><\/pre>\n\n\n\n
conda\u5b89\u88c5<\/strong><\/p>\n\n\n\n
1. \u5b89\u88c5 Miniforge<\/h3>\n\n\n\n
\u6309\u7167\u987a\u5e8f \u4f9d\u6b21\u6267\u884c<\/h4>\n\n\n\n
 # \u4e0b\u8f7d\u5e76\u5b89\u88c5 Miniforge
 wget http:\/\/iyun70.com:8011\/softs\/Miniforge3-Linux-x86_64.sh &&
 chmod +x Miniforge3-Linux-x86_64.sh &&
 .\/Miniforge3-Linux-x86_64.sh -b -f -p ~\/.miniconda &&
 ~\/.miniconda\/bin\/mamba init<\/pre>\n\n\n\n
 # \u91cd\u65b0\u52a0\u8f7d Bash
 bash<\/pre>\n\n\n\n
 # \u5173\u95ed\u81ea\u52a8\u6fc0\u6d3b\u57fa\u7840\u73af\u5883
 conda config --set auto_activate_base false<\/pre>\n\n\n\n
 # \u79fb\u9664 Mamba \u542f\u52a8\u6a2a\u5e45
 mamba activate base
 mamba env config vars set MAMBA_NO_BANNER=1
 mamba deactivate
 mamba activate base<\/pre>\n\n\n\n
 # \u5220\u9664\u5b89\u88c5\u811a\u672c
 rm -rf Miniforge3-Linux-x86_64.sh<\/pre>\n\n\n\n
2. \u914d\u7f6e Conda \u6e90<\/h3>\n\n\n\n
 echo \"channels:\n   - defaults\n show_channel_urls: true\n default_channels:\n   - https:\/\/mirrors.tuna.tsinghua.edu.cn\/anaconda\/pkgs\/main\n   - https:\/\/mirrors.tuna.tsinghua.edu.cn\/anaconda\/pkgs\/r\n   - https:\/\/mirrors.tuna.tsinghua.edu.cn\/anaconda\/pkgs\/msys2\n custom_channels:\n   conda-forge: https:\/\/mirrors.tuna.tsinghua.edu.cn\/anaconda\/cloud\n   msys2: https:\/\/mirrors.tuna.tsinghua.edu.cn\/anaconda\/cloud\n   bioconda: https:\/\/mirrors.tuna.tsinghua.edu.cn\/anaconda\/cloud\n   menpo: https:\/\/mirrors.tuna.tsinghua.edu.cn\/anaconda\/cloud\n   pytorch: https:\/\/mirrors.tuna.tsinghua.edu.cn\/anaconda\/cloud\n   pytorch-lts: https:\/\/mirrors.tuna.tsinghua.edu.cn\/anaconda\/cloud\n   simpleitk: https:\/\/mirrors.tuna.tsinghua.edu.cn\/anaconda\/cloud\n   deepmodeling: https:\/\/mirrors.tuna.tsinghua.edu.cn\/anaconda\/cloud\" > ~\/.condarc<\/pre>\n\n\n\n
3. \u914d\u7f6e Pip \u6e90<\/h3>\n\n\n\n
 # \u8bbe\u7f6e Pip \u6e90
 pip config set global.index-url https:\/\/pypi.tuna.tsinghua.edu.cn\/simple
 # \u66f4\u65b0\u5230\u6700\u65b0  \u9632\u6b62\u517c\u5bb9\u6027\u95ee\u9898\u6216\u7f3a\u5e93\u7f3a\u5305\u7684\u95ee\u9898
 conda update conda -y
 mamba update mamba -y<\/pre>\n\n\n\n
<\/p>\n","protected":false},"excerpt":{"rendered":"
\u96c6\u7fa4\u4fe1\u606f(\u9010\u6b65\u589e\u52a0) \u8282\u70b9 CPU \u5185\u5b58 \u663e\u5361 iyun105 72 \u6838\u5fc3 256 GB NVIDIA Tes […]<\/p>\n","protected":false},"author":58,"featured_media":0,"comment_status":"closed","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"jetpack_post_was_ever_published":false,"_jetpack_newsletter_access":""},"categories":[1],"tags":[],"jetpack_featured_media_url":"","_links":{"self":[{"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/61844"}],"collection":[{"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=\/wp\/v2\/users\/58"}],"replies":[{"embeddable":true,"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=61844"}],"version-history":[{"count":11,"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/61844\/revisions"}],"predecessor-version":[{"id":61899,"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/61844\/revisions\/61899"}],"wp:attachment":[{"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=61844"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=61844"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=61844"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}