{"id":1047,"date":"2026-07-28T00:00:00","date_gmt":"2026-07-27T16:00:00","guid":{"rendered":"https:\/\/localarch.ai\/?p=1047"},"modified":"2026-07-09T17:29:51","modified_gmt":"2026-07-09T09:29:51","slug":"virtualizing-local-ai-proxmox-docker-and-gpu-passthrough-best-practices","status":"publish","type":"post","link":"https:\/\/localarch.ai\/zh-hant\/virtualizing-local-ai-proxmox-docker-and-gpu-passthrough-best-practices\/","title":{"rendered":"\u865b\u64ec\u5316\u672c\u5730 AI\uff1aProxmox\u3001Docker \u8207 GPU \u76f4\u901a\u6700\u4f73\u5be6\u8e10"},"content":{"rendered":"

\u672c\u5730 AI <\/strong>\u865b\u64ec\u5316\uff1aProxmox<\/strong>\u3001Docker <\/strong>\u8207 GPU <\/strong>\u76f4\u901a\u6700\u4f73\u5be6\u8e10<\/strong><\/h2>\n

\u57fa\u790e\u8a2d\u65bd\u6f14\u9032\uff1a\u5f9e\u55ae\u4e00\u5de5\u4f5c\u8ca0\u8f09\u5230\u865b\u64ec\u5316 AI <\/strong>\u5be6\u9a57\u5ba4<\/strong><\/h2>\n

\u96a8\u8457\u6211\u5011\u8de8\u904e\u5e74\u4e2d\u4e26\u56de\u9867\u8fd1\u671f\u50cf ICML \u9019\u985e\u6703\u8b70\u7684\u8da8\u52e2\uff0c\u5728\u958b\u6e90 AI \u793e\u7fa4\u4e2d\u51fa\u73fe\u4e86\u4e00\u500b\u660e\u986f\u7684\u4e3b\u984c\uff1a\u672c\u5730\u57fa\u790e\u8a2d\u65bd\u7684\u5c08\u696d\u5316\u3002\u8a0e\u8ad6\u5df2\u7d93\u5f9e\u5728\u684c\u9762\u4e0a\u904b\u884c\u55ae\u4e00\u6a21\u578b\uff0c\u8f49\u5411\u5728\u55ae\u4e00\u4f3a\u670d\u5668\u4e0a\u591a\u500b\u7368\u7acb\u7684\u4eba\u5de5\u667a\u6167\u74b0\u5883<\/strong>\u3002\u9019\u5c31\u662f\u865b\u64ec\u5316\u7684\u529b\u91cf\u2013\u4e00\u7a2e\u5c07\u5f37\u5927\u5be6\u9ad4\u6a5f\u5668\u8f49\u8b8a\u70ba\u9748\u6d3b\u3001\u5b89\u5168\u4e14\u53ef\u64f4\u5c55\u7684 AI \u5be6\u9a57\u5ba4\u7684\u65b9\u6cd5\u3002\u5c0d\u65bc\u8a8d\u771f\u5c0d\u5f85\u672c\u5730 AI \u7684\u4f01\u696d\u8207\u7814\u7a76\u4eba\u54e1\u800c\u8a00\uff0c\u7cbe\u901a\u50cf Proxmox \u9019\u6a23\u7684\u88f8\u6a5f\u865b\u64ec\u5316\u5de5\u5177\u548c Docker \u7684\u5bb9\u5668\u5316\u6280\u8853\uff0c\u4e26\u901a\u904e GPU \u76f4\u901a\u9019\u9805\u95dc\u9375\u6280\u8853\u6a4b\u63a5\uff0c\u5df2\u4e0d\u518d\u662f\u9078\u9805\uff0c\u800c\u662f\u6210\u719f\u3001\u53ef\u751f\u7522\u672c\u5730 AI \u7b56\u7565\u7684\u57fa\u77f3\u3002<\/p>\n

\u865b\u64ec\u5316\u76f4\u63a5\u653e\u5927\u4e86\u672c\u5730\u4eba\u5de5\u667a\u6167\u7684\u6838\u5fc3\u512a\u52e2<\/u><\/strong>\uff1a\u5b83\u901a\u904e\u9694\u96e2\u5c08\u6848\u4f86\u589e\u5f37\u6578\u64da\u96b1\u79c1\u548c\u5b89\u5168\u6027<\/strong>\uff0c\u901a\u904e\u5141\u8a31\u52d5\u614b\u8cc7\u6e90\u914d\u7f6e\u4f86\u63d0\u5347\u9748\u6d3b\u6027\u548c\u53ef\u64f4\u5c55\u6027<\/strong>\uff0c\u4e26\u901a\u904e\u5efa\u7acb\u65bc\u958b\u6e90\u5e73\u53f0\u4e0a\u4f86\u978f\u56fa\u4f60\u64fa\u812b\u4f9b\u61c9\u5546\u9396\u5b9a\u7684\u81ea\u7531\u3002\u672c\u6587\u63d0\u4f9b\u4e86\u4e00\u500b\u69cb\u5efa\u9019\u7a2e\u865b\u64ec\u5316\u57fa\u790e\u7684\u5be6\u7528\u6307\u5357\uff0c\u78ba\u4fdd\u4f60\u7684\u786c\u9ad4\u6295\u8cc7\u767c\u63ee\u6700\u5927\u6548\u7528\uff0c\u4e26\u4e14\u4f60\u7684\u4eba\u5de5\u667a\u6167\u8a08\u756b\u53ef\u4ee5\u7121\u9650\u5236\u5730\u6210\u9577\u3002<\/p>\n

\u70ba\u4ec0\u9ebc\u8981\u865b\u64ec\u5316\u60a8\u7684\u672c\u5730 AI \u5806\u758a\uff1f<\/strong><\/h2>\n

\u5728\u6df1\u5165\u63a2\u7d22\u300c\u5982\u4f55\u300d\u4e4b\u524d\uff0c\u5148\u978f\u56fa\u300c\u70ba\u4f55\u300d\u3002\u865b\u64ec\u5316\u7684\u672c\u5730\u4eba\u5de5\u667a\u6167\u74b0\u5883\u63d0\u4f9b\u5be6\u969b\u7684\u597d\u8655\uff0c\u4ee5\u89e3\u6c7a\u5e38\u898b\u7684\u64f4\u5c55\u75db\u9ede\uff1a<\/p>\n

    \n
  • \u9694\u96e2\u8207\u53ef\u91cd\u73fe\u6027\uff1a<\/strong>\u5728\u5b8c\u5168\u7368\u7acb\u7684\u74b0\u5883\u4e2d\u904b\u884c\u4e0d\u540c\u7684 AI \u5c08\u6848\uff08\u4f8b\u5982\uff0c\u4e00\u500b\u7a69\u5b9a\u7684\u804a\u5929\u6a5f\u5668\u4eba\u3001\u4e00\u500b\u5be6\u9a57\u6027\u7684\u5716\u50cf\u751f\u6210\u5668\u3001\u4e00\u500b\u5fae\u8abf\u4efb\u52d9\uff09\u3002\u9019\u53ef\u4ee5\u9632\u6b62\u5eab\u885d\u7a81\uff0c\u5141\u8a31\u5b8c\u7f8e\u7684\u53ef\u91cd\u73fe\u6027\uff0c\u4e26\u7c21\u5316\u5099\u4efd\u548c\u9077\u79fb\u3002<\/li>\n
  • \u6700\u4f73\u786c\u9ad4\u5229\u7528\uff1a<\/strong>\u4e00\u53f0\u9ad8\u6027\u80fd\u7684 GPU \u4f3a\u670d\u5668\u53ef\u4ee5\u6258\u7ba1\u591a\u500b\u865b\u64ec\u6a5f\uff08VMs\uff09\u6216\u5bb9\u5668\uff0c\u6bcf\u500b\u904b\u884c\u4e0d\u540c\u7684\u6a21\u578b\u6216\u670d\u52d9\u4e0d\u540c\u7684\u5718\u968a\u3002\u9019\u5c07\u6602\u8cb4\u7684\u786c\u9ad4\u5f9e\u5076\u723e\u4f7f\u7528\u7684\u8cc7\u6e90\u8f49\u8b8a\u70ba\u6301\u7e8c\u88ab\u5229\u7528\u7684\u8cc7\u7522\u3002<\/li>\n
  • \u589e\u5f37\u7684\u5b89\u5168\u6027\u8207\u7ba1\u7406\uff1a<\/strong>\u900f\u904e\u9694\u96e2\u5de5\u4f5c\u8ca0\u8f09\uff0c\u60a8\u53ef\u4ee5\u63a7\u5236\u6f5b\u5728\u7684\u5b89\u5168\u554f\u984c\u3002\u5b83\u9084\u7c21\u5316\u4e86\u7528\u6236\u5b58\u53d6\u63a7\u5236\uff0c\u4e26\u5141\u8a31\u60a8\u5728\u9032\u884c\u91cd\u5927\u66f4\u65b0\u4e4b\u524d\u5c0d\u865b\u64ec\u6a5f\u9032\u884c\u5feb\u7167\uff0c\u4ee5\u4fbf\u8f15\u9b06\u56de\u6efe\u3002<\/li>\n
  • \u958b\u767c\u5230\u751f\u7522\u6d41\u7a0b\uff1a<\/strong>\u4f7f\u7528\u8f15\u91cf\u7d1a\u7684 Docker \u5bb9\u5668\u9032\u884c\u958b\u767c\u548c\u6e2c\u8a66\uff0c\u7136\u5f8c\u5c07\u76f8\u540c\u7684\u5bb9\u5668\u90e8\u7f72\u5230\u66f4\u5f37\u5927\u7684 Proxmox \u865b\u64ec\u6a5f\u4e0a\u9032\u884c\u751f\u7522\uff0c\u78ba\u4fdd\u6574\u500b\u751f\u547d\u5468\u671f\u7684\u4e00\u81f4\u6027\u3002<\/li>\n<\/ul>\n

    <\/h2>\n

    \u6838\u5fc3\u6280\u8853 1<\/strong>\uff1aProxmox VE <\/strong>\u2013 <\/strong>\u88f8\u91d1\u5c6c\u865b\u64ec\u6a5f\u7ba1\u7406\u7a0b\u5f0f<\/strong><\/h2>\n

    Proxmox \u865b\u64ec\u74b0\u5883\uff08VE\uff09\u662f\u4e00\u500b\u5f37\u5927\u53c8\u958b\u6e90\u7684\u5e73\u53f0\uff0c\u8b93\u4f60\u53ef\u4ee5\u628a\u5be6\u9ad4\u4f3a\u670d\u5668\u8b8a\u6210\u4e00\u500b\u8d85\u7d1a\u7ba1\u7406\u5668\uff0c\u7ba1\u7406\u591a\u500b\u865b\u64ec\u6a5f\u548c Linux \u5bb9\u5668\uff08LXC\uff09\u3002<\/p>\n

    \u70ba\u4ec0\u9ebc\u9078\u64c7 Proxmox <\/strong>\u4f86\u505a AI<\/strong>\uff1f<\/strong><\/p>\n

    \u5b83\u8b93\u4f60\u5b8c\u5168\u638c\u63a7\u5e95\u5c64\u786c\u9ad4\u3002\u4f60\u53ef\u4ee5\u5efa\u7acb\u5206\u914d\u597d CPU \u6838\u5fc3\u3001\u8a18\u61b6\u9ad4\uff0c\u6700\u91cd\u8981\u7684\u662f\u53ef\u4ee5\u900f\u904e\u76f4\u901a\u65b9\u5f0f\u76f4\u63a5\u4f7f\u7528 GPU \u8cc7\u6e90\u7684\u5c08\u7528\u865b\u64ec\u6a5f\u3002\u9019\u5c0d\u65bc\u5efa\u7acb\u6301\u4e45\u3001\u529f\u80fd\u5b8c\u6574\u7684\u74b0\u5883\u4f86\u8aaa\u975e\u5e38\u7406\u60f3\uff0c\u50cf\u662f\u9577\u6642\u9593\u904b\u884c\u7684\u6a21\u578b\u4f3a\u670d\u5668\u6216\u8907\u96dc\u7684 AI \u6d41\u7a0b\u3002<\/p>\n

    \u6700\u4f73\u5be6\u8e10\u8a2d\u7f6e\uff1a<\/strong><\/p>\n

      \n
    1. \u76f4\u63a5\u5b89\u88dd\u5728\u88f8\u6a5f\u4e0a\uff1a<\/strong>\u5c07 Proxmox \u76f4\u63a5\u5b89\u88dd\u5728\u4f60\u7684\u4f3a\u670d\u5668\u786c\u9ad4\u4e0a\uff0c\u800c\u4e0d\u662f\u73fe\u6709\u7684\u4f5c\u696d\u7cfb\u7d71\u4e2d\uff0c\u9019\u6a23\u53ef\u4ee5\u7372\u5f97\u6700\u4f73\u6548\u80fd\u548c\u7a69\u5b9a\u6027\u3002<\/li>\n
    2. \u8a2d\u5b9a\u5132\u5b58\u6c60\uff1a<\/strong>\u70ba\u4f60\u7684 AI \u865b\u64ec\u6a5f\u8a2d\u7f6e\u5c08\u7528\u7684\u5feb\u901f\u5132\u5b58\uff08\u4f8b\u5982 NVMe \u5132\u5b58\u6c60\uff09\uff0c\u4ee5\u78ba\u4fdd\u6a21\u578b\u52a0\u8f09\u548c\u6aa2\u67e5\u9ede\u4fdd\u5b58\u8fc5\u901f\u3002\u5099\u4efd\u5247\u4f7f\u7528\u8f03\u6162\u3001\u5bb9\u91cf\u8f03\u5927\u7684\u5132\u5b58\u3002<\/li>\n
    3. \u7db2\u8def\u9694\u96e2\uff1a<\/strong>\u70ba\u4f60\u7684 AI \u5de5\u4f5c\u8ca0\u8f09\u5efa\u7acb\u4e00\u500b\u7368\u7acb\u7684\u865b\u64ec\u7db2\u8def\uff08VLAN\uff09\uff0c\u4ee5\u7ba1\u7406\u6d41\u91cf\u4e26\u63d0\u5347\u5b89\u5168\u6027\u3002<\/li>\n<\/ol>\n

      <\/h2>\n

      \u6838\u5fc3\u6280\u8853 2<\/strong>\uff1aDocker <\/strong>\u2013 <\/strong>\u5bb9\u5668\u5316\u5f15\u64ce<\/strong><\/h2>\n

      Docker \u5c07\u61c9\u7528\u7a0b\u5f0f\u548c\u5b83\u7684\u4f9d\u8cf4\u6253\u5305\u6210\u4e00\u500b\u6a19\u6e96\u5316\u3001\u53ef\u651c\u5e36\u7684\u55ae\u4f4d\uff0c\u7a31\u70ba\u5bb9\u5668\u3002\u5c0d AI \u4f86\u8aaa\uff0c\u9019\u610f\u5473\u8457\u4f60\u53ef\u4ee5\u6709\u4e00\u500b\u5bb9\u5668\u88dd\u8457 Ollama\uff0c\u53e6\u4e00\u500b\u88dd\u8457 AutoGPT\uff0c\u518d\u6709\u7b2c\u4e09\u500b\u88dd\u8457\u7279\u5b9a\u7684 PyTorch \u8a13\u7df4\u74b0\u5883\u2013\u5168\u90e8\u5728\u540c\u4e00\u53f0\u4e3b\u6a5f\u4e0a\u7368\u7acb\u904b\u884c\u3002<\/p>\n

      \u70ba\u4ec0\u9ebc\u7528 Docker <\/strong>\u505a AI<\/strong>\uff1f<\/strong><\/p>\n

      \u5b83\u6548\u7387\u8d85\u9ad8\u53c8\u5f88\u5feb\u3002\u5bb9\u5668\u5171\u4eab\u4e3b\u6a5f\u7684\u6838\u5fc3\uff0c\u6240\u4ee5\u5e7e\u79d2\u9418\u5c31\u80fd\u555f\u52d5\uff0c\u800c\u4e14\u958b\u92b7\u975e\u5e38\u5c0f\u3002Docker Hub \u4e0a\u9f90\u5927\u7684 AI\/ML \u751f\u614b\u7cfb\u7d71\u63d0\u4f9b\u5e7e\u4e4e\u6240\u6709\u6846\u67b6\u548c\u5de5\u5177\u7684\u9810\u5efa\u6620\u50cf\uff0c\u5927\u5927\u52a0\u5feb\u4e86\u8a2d\u7f6e\u901f\u5ea6\u3002<\/p>\n

      \u6700\u4f73\u5be6\u8e10\u8a2d\u7f6e\uff1a<\/strong><\/p>\n

        \n
      1. \u4f7f\u7528\u5b98\u65b9\/<\/strong>\u793e\u7fa4\u6620\u50cf\u6a94\uff1a<\/strong>\u5f9e\u5df2\u9a57\u8b49\u7684\u6620\u50cf\u6a94\u958b\u59cb\uff0c\u4f8b\u5982 ollama\/ollama\u3001nvcr.io\/nvidia\/pytorch \u6216 langchain\/langchain\uff0c\u78ba\u4fdd\u6709\u7a69\u5b9a\u7684\u57fa\u790e\u3002<\/li>\n
      2. \u4f7f\u7528\u5377\u4f86\u6301\u4e45\u5316\u6578\u64da\uff1a<\/strong>\u5343\u842c\u4e0d\u8981\u628a\u6a21\u578b\u3001\u6578\u64da\u96c6\u6216\u8cc7\u6599\u5eab\u5b58\u653e\u5728\u5bb9\u5668\u88e1\u3002\u4f7f\u7528 Docker \u5377\u6216\u7d81\u5b9a\u639b\u8f09\u4f86\u639b\u8f09\u76ee\u9304\uff08\u4f8b\u5982\uff1a-v \/your\/models:\/root\/.ollama\/models\uff09\u3002<\/li>\n
      3. \u70ba\u7de8\u6392\u800c\u64b0\u5beb\uff1a<\/strong>\u4f7f\u7528 docker-compose.yml \u6587\u4ef6\u4f86\u5b9a\u7fa9\u548c\u904b\u884c\u591a\u5bb9\u5668\u61c9\u7528\u7a0b\u5f0f\u3002\u9019\u975e\u5e38\u9069\u5408\u7528\u4e00\u500b\u6307\u4ee4\u5c31\u555f\u52d5\u5b8c\u6574\u7684\u74b0\u5883\uff0c\u4f8b\u5982\u4e00\u500b Postgres \u8cc7\u6599\u5eab\u3001\u4e00\u500b Redis \u5feb\u53d6\uff0c\u9084\u6709\u4e00\u500b Ollama API \u5bb9\u5668\u3002<\/li>\n<\/ol>\n

        \u95dc\u9375\u6a4b\u6a11\uff1aGPU <\/strong>\u76f4\u901a (VFIO)<\/strong><\/h2>\n

        \u9019\u662f\u9ad8\u6548\u80fd\u865b\u64ec\u5316 AI \u7684\u95dc\u9375\u3002GPU \u76f4\u901a\u5141\u8a31\u4f60\u628a\u5be6\u9ad4 GPU \u5c08\u9580\u7d66\u67d0\u500b\u865b\u64ec\u6a5f\u5668\u4f7f\u7528\uff0c\u6216\u8005\u8b93 Docker \u5bb9\u5668\u4e5f\u80fd\u7528\u5230\uff0c\u8b93\u5b83\u5011\u7684\u6548\u80fd\u5e7e\u4e4e\u9054\u5230\u672c\u6a5f\u6c34\u5e73\u3002<\/p>\n

        \u65b9\u6cd5 A<\/strong>\uff1a\u76f4\u63a5\u900f\u50b3\u5230 Proxmox <\/strong>\u865b\u64ec\u6a5f<\/strong><\/p>\n

        \u9019\u8b93\u4e00\u500b\u865b\u64ec\u6a5f\u53ef\u4ee5\u7368\u5360\u4e26\u5b8c\u5168\u63a7\u5236 GPU\u3002\u5c0d\u65bc\u5c08\u7528\u7684\u9ad8\u6548\u80fd AI \u5de5\u4f5c\u8ca0\u8f09\u4f86\u8aaa\uff0c\u9019\u662f\u6700\u597d\u7684\u9078\u64c7\u3002<\/p>\n

        \u5be6\u4f5c\u6b65\u9a5f\uff1a<\/strong><\/p>\n

          \n
        1. \u555f\u7528 IOMMU<\/strong>\uff1a<\/strong>\u5728\u4e3b\u6a5f BIOS\/UEFI \u548c Proxmox \u6838\u5fc3\u88e1\uff08\u900f\u904e\u7de8\u8f2f \/etc\/default\/grub\uff09\u3002<\/li>\n
        2. \u8b58\u5225 GPU<\/strong>\uff1a<\/strong>\u4f7f\u7528 lspci -nn | grep -i vga \u627e\u5230\u4f60\u7684 GPU ID\uff08\u4f8b\u5982\uff0cNVIDIA \u986f\u5361\u7684 ID \u662f 10de:2231\uff09\u3002<\/li>\n
        3. \u5c01\u9396\u9a45\u52d5\u7a0b\u5f0f\uff1a<\/strong>\u900f\u904e\u5c07\u5176 ID \u52a0\u5165 VFIO \u6a21\u7d44\u5217\u8868\uff08\/etc\/modprobe.d\/vfio.conf\uff09\u4f86\u9632\u6b62\u4e3b\u6a5f\u4f7f\u7528 GPU\u3002<\/li>\n
        4. \u9644\u52a0\u5230\u865b\u64ec\u6a5f\uff1a<\/strong>\u5728 Proxmox \u7db2\u9801\u4ecb\u9762\u4e2d\uff0c\u5c07 PCI \u88dd\u7f6e\u6dfb\u52a0\u5230\u4f60\u60f3\u8981\u7684\u865b\u64ec\u6a5f\u3002\u52fe\u9078\u300cAll Functions\u300d\u548c\u300cROM-Bar\u300d\u9078\u9805\u3002<\/li>\n<\/ol>\n

          \u65b9\u6cd5 B<\/strong>\uff1a\u8207 Docker <\/strong>\u5bb9\u5668\u5171\u7528 GPU<\/strong>\uff08NVIDIA<\/strong>\uff09<\/strong><\/p>\n

          \u5c0d\u65bc\u4ee5\u5bb9\u5668\u70ba\u57fa\u790e\u7684\u5de5\u4f5c\u6d41\u7a0b\uff0cNVIDIA \u5bb9\u5668\u5de5\u5177\u7d44\u662f\u5fc5\u5099\u7684\u3002<\/p>\n

          \u5be6\u65bd\u6b65\u9a5f\uff1a<\/strong><\/p>\n

            \n
          1. \u5b89\u88dd\u9a45\u52d5\u7a0b\u5f0f\uff1a<\/strong>\u78ba\u4fdd\u4e3b\u6a5f\u7cfb\u7d71\u4e0a\u5b89\u88dd\u4e86\u6700\u65b0\u7684 NVIDIA \u9a45\u52d5\u7a0b\u5f0f\u3002<\/li>\n
          2. \u5b89\u88dd\u5de5\u5177\u5305\uff1a<\/strong>\u6309\u7167 NVIDIA \u7684\u6307\u5357\u5b89\u88dd nvidia-container-toolkit\u3002<\/li>\n
          3. \u6e2c\u8a66\u904b\u884c\u6642\uff1a<\/strong>\u900f\u904e\u7de8\u8f2f \/etc\/docker\/daemon.json\uff0c\u5c07 Docker \u8a2d\u5b9a\u70ba\u4f7f\u7528 nvidia \u904b\u884c\u6642\u3002<\/li>\n
          4. \u4f7f\u7528 GPU <\/strong>\u904b\u884c\uff1a<\/strong>\u555f\u52d5\u5bb9\u5668\u6642\u52a0\u4e0a –gpus all \u53c3\u6578\uff1adocker run –gpus all ollama\/ollama serve\u3002<\/li>\n<\/ol>\n

            \u9078\u64c7\u4f60\u7684\u9053\u8def\uff1a\u5be6\u7528\u6c7a\u7b56\u6307\u5357<\/strong><\/h2>\n\n\n\n\n\n\n\n
            \u4f7f\u7528\u60c5\u5883\u8207\u76ee\u6a19<\/td>\n\u5efa\u8b70\u7684\u4e3b\u8981\u65b9\u6cd5<\/td>\n\u95dc\u9375\u914d\u7f6e\u8207\u7406\u7531<\/td>\n<\/tr>\n<\/thead>\n
            \u591a\u5718\u968a\/AIServer\u00a0 <\/strong><\/p>\n

            \u70ba\u4e0d\u540c\u90e8\u9580\u6216\u9805\u76ee\u63d0\u4f9b\u7368\u7acb\u5b8c\u6574\u7684\u74b0\u5883\u3002<\/td>\n

            		\u5e36 GPU <\/strong>\u76f4\u901a\u7684 Proxmox VE<\/strong><\/td>\n\u5c07\u6574\u500b GPU \u5c08\u9580\u5206\u914d\u7d66\u7279\u5b9a\u7684\u865b\u64ec\u6a5f\u3002\u63d0\u4f9b\u5f37\u5927\u7684\u9694\u96e2\u6027\u3001\u5b8c\u6574\u7684\u4f5c\u696d\u7cfb\u7d71\u5ba2\u88fd\u5316\uff0c\u5c0d\u50b3\u7d71 IT \u5718\u968a\u4f86\u8aaa\u4e5f\u66f4\u5bb9\u6613\u7ba1\u7406\u3002<\/td>\n<\/tr>\n
            \u958b\u767c\u8207\u5be6\u9a57<\/strong><\/p>\n

            \u5feb\u901f\u6e2c\u8a66\u4e0d\u540c\u7684\u6a21\u578b\u3001\u6846\u67b6\u548c\u5de5\u5177\u93c8\u3002<\/td>\n

            		\u5e36\u6709 NVIDIA <\/strong>\u57f7\u884c\u74b0\u5883\u7684 Docker<\/strong><\/td>\n\u4f7f\u7528 Docker Compose \u4f86\u5b9a\u7fa9\u5806\u758a\u3002\u6700\u5927\u5316\u786c\u9ad4\u5171\u4eab\u548c\u6548\u7387\uff1b\u5bb9\u5668\u53ef\u7acb\u5373\u5f9e\u9810\u5148\u5efa\u7acb\u7684\u6620\u50cf\u555f\u52d5\u3002<\/td>\n<\/tr>\n
            \u6df7\u5408\u751f\u7522\u5be6\u9a57\u5ba4<\/strong><\/p>\n

            \u7d50\u5408\u7a69\u5b9a\u3001\u9577\u671f\u904b\u884c\u7684\u670d\u52d9\u8207\u77ed\u66ab\u7684\u7814\u7a76\u9805\u76ee\u3002<\/td>\n

            		Proxmox <\/strong>\u8207 Docker <\/strong>\u7d50\u5408<\/strong><\/td>\n\u5728\u88f8\u6a5f\u4e0a\u904b\u884c Proxmox\u3002\u5728\u4e00\u500b\u5e36\u6709 GPU \u76f4\u901a\u7684\u5f37\u529b\u865b\u64ec\u6a5f\u88e1\uff0c\u904b\u884c Docker \u4f86\u7ba1\u7406\u591a\u500b AI \u61c9\u7528\u5bb9\u5668\u3002\u9019\u589e\u52a0\u4e86\u4e00\u5c64\u7ba1\u7406\u548c\u9694\u96e2\u3002<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

            \u89e3\u6c7a\u5e38\u898b\u7684\u865b\u64ec\u5316\u554f\u984c<\/strong><\/h2>\n
              \n
            • \u5728 VM\/<\/strong>\u5bb9\u5668\u4e2d\u51fa\u73fe\u300c\u672a\u5075\u6e2c\u5230 GPU<\/strong>\u300d\uff1a<\/strong>\u9019\u5e7e\u4e4e\u7e3d\u662f\u76f4\u901a\u914d\u7f6e\u7684\u554f\u984c\u3002\u518d\u6aa2\u67e5\u4e00\u4e0b IOMMU \u7fa4\u7d44\uff0c\u78ba\u4fdd\u4e3b\u6a5f\u9a45\u52d5\u5df2\u6b63\u78ba\u9ed1\u540d\u55ae\u5316\uff0c\u4e26\u78ba\u8a8d PCI \u88dd\u7f6e\u7684\u6240\u6709\u529f\u80fd\u90fd\u5df2\u6b63\u78ba\u9023\u63a5\u3002<\/li>\n
            • \u5bb9\u5668\u6548\u80fd\u5dee\uff1a<\/strong>\u78ba\u4fdd\u4f60\u6b63\u5728\u4f7f\u7528 nvidia runtime \u548c –gpus \u6a19\u8a8c\u3002\u53e6\u5916\uff0c\u78ba\u8a8d\u4f60\u7684\u4e3b\u6a5f\u64c1\u6709\u8207\u5bb9\u5668\u61c9\u7528\u7a0b\u5f0f\u6240\u9700 CUDA \u7248\u672c\u76f8\u7b26\u7684 GPU \u9a45\u52d5\u7a0b\u5f0f\u7248\u672c<\/li>\n
            • Proxmox <\/strong>\u4e3b\u6a5f\u5728\u555f\u52d5\u865b\u64ec\u6a5f\u6642\u5d29\u6f70\uff1a<\/strong>\u901a\u5e38\u548c\u6709\u554f\u984c\u7684 ROM-Bar \u8655\u7406\u6709\u95dc\u3002\u53ef\u4ee5\u8a66\u8457\u5728\u672a\u52fe\u9078\u300cROM-Bar\u300d\u9078\u9805\u4e0b\u555f\u52d5\u865b\u64ec\u6a5f\uff0c\u6216\u8005\u4f7f\u7528\u7279\u5b9a\u7684 GPU ROM \u6587\u4ef6\u4f86\u63d0\u9ad8\u7a69\u5b9a\u6027\u3002<\/li>\n<\/ul>\n

              \u7d50\u8ad6\uff1a\u5efa\u7acb\u4f60\u7684\u4e3b\u6b0a AI <\/strong>\u96f2<\/strong><\/h2>\n

              \u6b63\u5982\u6700\u8fd1\u7684\u6703\u8b70\u6240\u5f37\u8abf\u7684\uff0c\u672a\u4f86\u53ef\u64f4\u5c55\u7684\u958b\u6e90 AI \u5c07\u662f\u6a21\u7d44\u5316\u548c\u8edf\u9ad4\u5b9a\u7fa9\u7684\u3002\u4f7f\u7528 Proxmox \u548c Docker \u865b\u64ec\u5316\u4f60\u7684\u672c\u5730 AI \u57fa\u790e\u8a2d\u65bd\uff0c\u662f\u6253\u9020\u81ea\u6709\u4e3b\u6b0a<\/strong> AI <\/strong>\u96f2<\/strong>\u7aef\u7684\u6c7a\u5b9a\u6027\u4e00\u6b65\u2013\u4e00\u500b\u9748\u6d3b\u3001\u79c1\u5bc6\u4e14\u5177\u6210\u672c\u6548\u76ca\u7684\u5e73\u53f0\uff0c\u9ad4\u73fe\u4e86 LocalArch.ai <\/u>\u6240\u5021\u5c0e\u7684\u6578\u64da\u63a7\u5236\u548c\u9577\u671f\u50f9\u503c\u539f\u5247\u3002<\/p>\n

              \u9019\u7a2e\u65b9\u5f0f\u80fd\u8b93\u4f60\u7684\u6295\u8cc7\u5177\u5099\u672a\u4f86\u4fdd\u969c\u3002\u9700\u8981\u6e2c\u8a66\u65b0\u7684\u6a21\u578b\u55ce\uff1f\u958b\u4e00\u500b Docker \u5bb9\u5668\u5c31\u884c\u3002\u9700\u8981\u7d66\u65b0\u7684\u8cc7\u6599\u79d1\u5b78\u5718\u968a\u63d0\u4f9b\u5c08\u5c6c\u74b0\u5883\u55ce\uff1f\u5728 Proxmox \u88e1\u8907\u88fd\u4e00\u500b VM \u7bc4\u672c\u5c31\u597d\u3002\u638c\u63e1\u9019\u4e9b\u865b\u64ec\u5316\u548c\u76f4\u901a\u6280\u8853\u5f8c\uff0c\u4f60\u5c31\u4e0d\u53ea\u662f\u55ae\u7d14\u4f7f\u7528 AI \u5de5\u5177\uff0c\u800c\u662f\u80fd\u6210\u70ba\u4e00\u500b\u80fd\u6253\u9020\u5f48\u6027\u3001\u53ef\u64f4\u5c55 AI \u751f\u614b\u7cfb\u7d71\u7684\u67b6\u69cb\u5e2b\uff0c\u8b93\u5b83\u96a8\u8457\u4f60\u7684\u91ce\u5fc3\u7121\u7e2b\u6210\u9577\u3002<\/p>\n

              \u8a2d\u8a08\u548c\u5be6\u73fe\u4e00\u500b\u7a69\u5065\u7684\u865b\u64ec\u5316\u672c\u5730 AI <\/em>\u57fa\u790e\u8a2d\u65bd\u9700\u8981\u7d30\u5fc3\u7684\u898f\u5283\u548c\u5c08\u696d\u77e5\u8b58\u3002LocalArch.ai<\/u> <\/em>\u5718\u968a\u5229\u7528\u5728 Proxmox<\/em>\u3001Docker <\/em>\u548c GPU <\/em>\u914d\u7f6e\u4e0a\u7684\u6df1\u539a\u7d93\u9a57\uff0c\u70ba\u5ba2\u6236\u6253\u9020\u91cf\u8eab\u5b9a\u5236\u3001\u53ef\u6295\u5165\u751f\u7522\u7684\u672c\u5730 AI <\/em>\u5e73\u53f0\u3002\u806f\u7d61\u6211\u5011<\/u>\uff0c\u5c07\u4f60\u7684\u786c\u9ad4\u8f49\u8b8a\u6210\u5c08\u5c6c\u7684 AI <\/em>\u96f2\u7aef\u5427\u3002<\/em><\/p>\n","protected":false},"excerpt":{"rendered":"

              本地 AI 虛擬化：Proxmox& […]<\/p>\n","protected":false},"author":2,"featured_media":1048,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_jetpack_newsletter_access":"","_jetpack_dont_email_post_to_subs":false,"_jetpack_newsletter_tier_id":0,"_jetpack_memberships_contains_paywalled_content":false,"_jetpack_memberships_contains_paid_content":false,"footnotes":""},"categories":[78],"tags":[320,56,82,295,59,296,321],"class_list":["post-1047","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-blog","tag-architecture","tag-docker","tag-gpu","tag-local-ai","tag-proxmox","tag--ai","tag-321"],"yoast_head":"\n\u865b\u64ec\u5316\u672c\u5730 AI\uff1aProxmox\u3001Docker \u8207 GPU \u76f4\u901a\u6700\u4f73\u5be6\u8e10 -<\/title>\n<meta name=\"description\" content=\"\u5c07 GPU \u786c\u9ad4\u5206\u4eab\u7d66\u865b\u64ec\u5316\u7bc0\u9ede\u548c\uff0f\u6216 Docker\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/localarch.ai\/zh-hant\/virtualizing-local-ai-proxmox-docker-and-gpu-passthrough-best-practices\/\" \/>\n<meta property=\"og:locale\" content=\"zh_TW\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"\u865b\u64ec\u5316\u672c\u5730 AI\uff1aProxmox\u3001Docker \u8207 GPU \u76f4\u901a\u6700\u4f73\u5be6\u8e10 -\" \/>\n<meta property=\"og:description\" content=\"\u5c07 GPU \u786c\u9ad4\u5206\u4eab\u7d66\u865b\u64ec\u5316\u7bc0\u9ede\u548c\uff0f\u6216 Docker\" \/>\n<meta property=\"og:url\" content=\"https:\/\/localarch.ai\/zh-hant\/virtualizing-local-ai-proxmox-docker-and-gpu-passthrough-best-practices\/\" \/>\n<meta property=\"article:published_time\" content=\"2026-07-27T16:00:00+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/localarch.ai\/wp-content\/uploads\/2026\/07\/GPU_Virtualization-1024x575.png\" \/>\n\t<meta property=\"og:image:width\" content=\"1024\" \/>\n\t<meta property=\"og:image:height\" content=\"575\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/png\" \/>\n<meta name=\"author\" content=\"webeditor\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"\u4f5c\u8005:\" \/>\n\t<meta name=\"twitter:data1\" content=\"webeditor\" \/>\n\t<meta name=\"twitter:label2\" content=\"\u9810\u4f30\u95b1\u8b80\u6642\u9593\" \/>\n\t<meta name=\"twitter:data2\" content=\"2 \u5206\u9418\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\\\/\\\/localarch.ai\\\/zh-hant\\\/virtualizing-local-ai-proxmox-docker-and-gpu-passthrough-best-practices\\\/#article\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/localarch.ai\\\/zh-hant\\\/virtualizing-local-ai-proxmox-docker-and-gpu-passthrough-best-practices\\\/\"},\"author\":{\"name\":\"webeditor\",\"@id\":\"https:\\\/\\\/localarch.ai\\\/zh-hant\\\/#\\\/schema\\\/person\\\/91a141c7253ea42b1a40cd04c042b399\"},\"headline\":\"\u865b\u64ec\u5316\u672c\u5730 AI\uff1aProxmox\u3001Docker \u8207 GPU \u76f4\u901a\u6700\u4f73\u5be6\u8e10\",\"datePublished\":\"2026-07-27T16:00:00+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\\\/\\\/localarch.ai\\\/zh-hant\\\/virtualizing-local-ai-proxmox-docker-and-gpu-passthrough-best-practices\\\/\"},\"wordCount\":188,\"commentCount\":0,\"publisher\":{\"@id\":\"https:\\\/\\\/localarch.ai\\\/zh-hant\\\/#organization\"},\"image\":{\"@id\":\"https:\\\/\\\/localarch.ai\\\/zh-hant\\\/virtualizing-local-ai-proxmox-docker-and-gpu-passthrough-best-practices\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/localarch.ai\\\/wp-content\\\/uploads\\\/2026\\\/07\\\/GPU_Virtualization-scaled.png\",\"keywords\":[\"Architecture\",\"Docker\",\"GPU\",\"Local AI\",\"Proxmox\",\"\u672c\u5730 AI\",\"\u67b6\u69cb\"],\"articleSection\":[\"Blog\"],\"inLanguage\":\"zh-TW\",\"potentialAction\":[{\"@type\":\"CommentAction\",\"name\":\"Comment\",\"target\":[\"https:\\\/\\\/localarch.ai\\\/zh-hant\\\/virtualizing-local-ai-proxmox-docker-and-gpu-passthrough-best-practices\\\/#respond\"]}]},{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/localarch.ai\\\/zh-hant\\\/virtualizing-local-ai-proxmox-docker-and-gpu-passthrough-best-practices\\\/\",\"url\":\"https:\\\/\\\/localarch.ai\\\/zh-hant\\\/virtualizing-local-ai-proxmox-docker-and-gpu-passthrough-best-practices\\\/\",\"name\":\"\u865b\u64ec\u5316\u672c\u5730 AI\uff1aProxmox\u3001Docker \u8207 GPU \u76f4\u901a\u6700\u4f73\u5be6\u8e10 -\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/localarch.ai\\\/zh-hant\\\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\\\/\\\/localarch.ai\\\/zh-hant\\\/virtualizing-local-ai-proxmox-docker-and-gpu-passthrough-best-practices\\\/#primaryimage\"},\"image\":{\"@id\":\"https:\\\/\\\/localarch.ai\\\/zh-hant\\\/virtualizing-local-ai-proxmox-docker-and-gpu-passthrough-best-practices\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/localarch.ai\\\/wp-content\\\/uploads\\\/2026\\\/07\\\/GPU_Virtualization-scaled.png\",\"datePublished\":\"2026-07-27T16:00:00+00:00\",\"description\":\"\u5c07 GPU \u786c\u9ad4\u5206\u4eab\u7d66\u865b\u64ec\u5316\u7bc0\u9ede\u548c\uff0f\u6216 Docker\",\"breadcrumb\":{\"@id\":\"https:\\\/\\\/localarch.ai\\\/zh-hant\\\/virtualizing-local-ai-proxmox-docker-and-gpu-passthrough-best-practices\\\/#breadcrumb\"},\"inLanguage\":\"zh-TW\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\\\/\\\/localarch.ai\\\/zh-hant\\\/virtualizing-local-ai-proxmox-docker-and-gpu-passthrough-best-practices\\\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"zh-TW\",\"@id\":\"https:\\\/\\\/localarch.ai\\\/zh-hant\\\/virtualizing-local-ai-proxmox-docker-and-gpu-passthrough-best-practices\\\/#primaryimage\",\"url\":\"https:\\\/\\\/localarch.ai\\\/wp-content\\\/uploads\\\/2026\\\/07\\\/GPU_Virtualization-scaled.png\",\"contentUrl\":\"https:\\\/\\\/localarch.ai\\\/wp-content\\\/uploads\\\/2026\\\/07\\\/GPU_Virtualization-scaled.png\",\"width\":2560,\"height\":1438},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/localarch.ai\\\/zh-hant\\\/virtualizing-local-ai-proxmox-docker-and-gpu-passthrough-best-practices\\\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\\\/\\\/localarch.ai\\\/zh-hant\\\/home-localarch-ai\\\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"\u865b\u64ec\u5316\u672c\u5730 AI\uff1aProxmox\u3001Docker \u8207 GPU \u76f4\u901a\u6700\u4f73\u5be6\u8e10\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\\\/\\\/localarch.ai\\\/zh-hant\\\/#website\",\"url\":\"https:\\\/\\\/localarch.ai\\\/zh-hant\\\/\",\"name\":\"Local AI Architecture\",\"description\":\"Local AI\\\/ IT Architecture\",\"publisher\":{\"@id\":\"https:\\\/\\\/localarch.ai\\\/zh-hant\\\/#organization\"},\"alternateName\":\"LocalArch AI\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\\\/\\\/localarch.ai\\\/zh-hant\\\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"zh-TW\"},{\"@type\":\"Organization\",\"@id\":\"https:\\\/\\\/localarch.ai\\\/zh-hant\\\/#organization\",\"name\":\"Local AI Architecture\",\"url\":\"https:\\\/\\\/localarch.ai\\\/zh-hant\\\/\",\"logo\":{\"@type\":\"ImageObject\",\"inLanguage\":\"zh-TW\",\"@id\":\"https:\\\/\\\/localarch.ai\\\/zh-hant\\\/#\\\/schema\\\/logo\\\/image\\\/\",\"url\":\"https:\\\/\\\/localarch.ai\\\/wp-content\\\/uploads\\\/2025\\\/09\\\/cropped-LocalArch-AI-logo.png\",\"contentUrl\":\"https:\\\/\\\/localarch.ai\\\/wp-content\\\/uploads\\\/2025\\\/09\\\/cropped-LocalArch-AI-logo.png\",\"width\":400,\"height\":119,\"caption\":\"Local AI Architecture\"},\"image\":{\"@id\":\"https:\\\/\\\/localarch.ai\\\/zh-hant\\\/#\\\/schema\\\/logo\\\/image\\\/\"}},{\"@type\":\"Person\",\"@id\":\"https:\\\/\\\/localarch.ai\\\/zh-hant\\\/#\\\/schema\\\/person\\\/91a141c7253ea42b1a40cd04c042b399\",\"name\":\"webeditor\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"zh-TW\",\"@id\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/4ecc7bad18fce62b20524b26668563f37907995e1838ca8a29a5cb6c98262cee?s=96&d=mm&r=g\",\"url\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/4ecc7bad18fce62b20524b26668563f37907995e1838ca8a29a5cb6c98262cee?s=96&d=mm&r=g\",\"contentUrl\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/4ecc7bad18fce62b20524b26668563f37907995e1838ca8a29a5cb6c98262cee?s=96&d=mm&r=g\",\"caption\":\"webeditor\"},\"url\":\"https:\\\/\\\/localarch.ai\\\/zh-hant\\\/author\\\/laeditor\\\/\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"\u865b\u64ec\u5316\u672c\u5730 AI\uff1aProxmox\u3001Docker \u8207 GPU \u76f4\u901a\u6700\u4f73\u5be6\u8e10 -","description":"\u5c07 GPU \u786c\u9ad4\u5206\u4eab\u7d66\u865b\u64ec\u5316\u7bc0\u9ede\u548c\uff0f\u6216 Docker","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/localarch.ai\/zh-hant\/virtualizing-local-ai-proxmox-docker-and-gpu-passthrough-best-practices\/","og_locale":"zh_TW","og_type":"article","og_title":"\u865b\u64ec\u5316\u672c\u5730 AI\uff1aProxmox\u3001Docker \u8207 GPU \u76f4\u901a\u6700\u4f73\u5be6\u8e10 -","og_description":"\u5c07 GPU \u786c\u9ad4\u5206\u4eab\u7d66\u865b\u64ec\u5316\u7bc0\u9ede\u548c\uff0f\u6216 Docker","og_url":"https:\/\/localarch.ai\/zh-hant\/virtualizing-local-ai-proxmox-docker-and-gpu-passthrough-best-practices\/","article_published_time":"2026-07-27T16:00:00+00:00","og_image":[{"width":1024,"height":575,"url":"https:\/\/localarch.ai\/wp-content\/uploads\/2026\/07\/GPU_Virtualization-1024x575.png","type":"image\/png"}],"author":"webeditor","twitter_card":"summary_large_image","twitter_misc":{"\u4f5c\u8005:":"webeditor","\u9810\u4f30\u95b1\u8b80\u6642\u9593":"2 \u5206\u9418"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/localarch.ai\/zh-hant\/virtualizing-local-ai-proxmox-docker-and-gpu-passthrough-best-practices\/#article","isPartOf":{"@id":"https:\/\/localarch.ai\/zh-hant\/virtualizing-local-ai-proxmox-docker-and-gpu-passthrough-best-practices\/"},"author":{"name":"webeditor","@id":"https:\/\/localarch.ai\/zh-hant\/#\/schema\/person\/91a141c7253ea42b1a40cd04c042b399"},"headline":"\u865b\u64ec\u5316\u672c\u5730 AI\uff1aProxmox\u3001Docker \u8207 GPU \u76f4\u901a\u6700\u4f73\u5be6\u8e10","datePublished":"2026-07-27T16:00:00+00:00","mainEntityOfPage":{"@id":"https:\/\/localarch.ai\/zh-hant\/virtualizing-local-ai-proxmox-docker-and-gpu-passthrough-best-practices\/"},"wordCount":188,"commentCount":0,"publisher":{"@id":"https:\/\/localarch.ai\/zh-hant\/#organization"},"image":{"@id":"https:\/\/localarch.ai\/zh-hant\/virtualizing-local-ai-proxmox-docker-and-gpu-passthrough-best-practices\/#primaryimage"},"thumbnailUrl":"https:\/\/localarch.ai\/wp-content\/uploads\/2026\/07\/GPU_Virtualization-scaled.png","keywords":["Architecture","Docker","GPU","Local AI","Proxmox","\u672c\u5730 AI","\u67b6\u69cb"],"articleSection":["Blog"],"inLanguage":"zh-TW","potentialAction":[{"@type":"CommentAction","name":"Comment","target":["https:\/\/localarch.ai\/zh-hant\/virtualizing-local-ai-proxmox-docker-and-gpu-passthrough-best-practices\/#respond"]}]},{"@type":"WebPage","@id":"https:\/\/localarch.ai\/zh-hant\/virtualizing-local-ai-proxmox-docker-and-gpu-passthrough-best-practices\/","url":"https:\/\/localarch.ai\/zh-hant\/virtualizing-local-ai-proxmox-docker-and-gpu-passthrough-best-practices\/","name":"\u865b\u64ec\u5316\u672c\u5730 AI\uff1aProxmox\u3001Docker \u8207 GPU \u76f4\u901a\u6700\u4f73\u5be6\u8e10 -","isPartOf":{"@id":"https:\/\/localarch.ai\/zh-hant\/#website"},"primaryImageOfPage":{"@id":"https:\/\/localarch.ai\/zh-hant\/virtualizing-local-ai-proxmox-docker-and-gpu-passthrough-best-practices\/#primaryimage"},"image":{"@id":"https:\/\/localarch.ai\/zh-hant\/virtualizing-local-ai-proxmox-docker-and-gpu-passthrough-best-practices\/#primaryimage"},"thumbnailUrl":"https:\/\/localarch.ai\/wp-content\/uploads\/2026\/07\/GPU_Virtualization-scaled.png","datePublished":"2026-07-27T16:00:00+00:00","description":"\u5c07 GPU \u786c\u9ad4\u5206\u4eab\u7d66\u865b\u64ec\u5316\u7bc0\u9ede\u548c\uff0f\u6216 Docker","breadcrumb":{"@id":"https:\/\/localarch.ai\/zh-hant\/virtualizing-local-ai-proxmox-docker-and-gpu-passthrough-best-practices\/#breadcrumb"},"inLanguage":"zh-TW","potentialAction":[{"@type":"ReadAction","target":["https:\/\/localarch.ai\/zh-hant\/virtualizing-local-ai-proxmox-docker-and-gpu-passthrough-best-practices\/"]}]},{"@type":"ImageObject","inLanguage":"zh-TW","@id":"https:\/\/localarch.ai\/zh-hant\/virtualizing-local-ai-proxmox-docker-and-gpu-passthrough-best-practices\/#primaryimage","url":"https:\/\/localarch.ai\/wp-content\/uploads\/2026\/07\/GPU_Virtualization-scaled.png","contentUrl":"https:\/\/localarch.ai\/wp-content\/uploads\/2026\/07\/GPU_Virtualization-scaled.png","width":2560,"height":1438},{"@type":"BreadcrumbList","@id":"https:\/\/localarch.ai\/zh-hant\/virtualizing-local-ai-proxmox-docker-and-gpu-passthrough-best-practices\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/localarch.ai\/zh-hant\/home-localarch-ai\/"},{"@type":"ListItem","position":2,"name":"\u865b\u64ec\u5316\u672c\u5730 AI\uff1aProxmox\u3001Docker \u8207 GPU \u76f4\u901a\u6700\u4f73\u5be6\u8e10"}]},{"@type":"WebSite","@id":"https:\/\/localarch.ai\/zh-hant\/#website","url":"https:\/\/localarch.ai\/zh-hant\/","name":"Local AI Architecture","description":"Local AI\/ IT Architecture","publisher":{"@id":"https:\/\/localarch.ai\/zh-hant\/#organization"},"alternateName":"LocalArch AI","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/localarch.ai\/zh-hant\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"zh-TW"},{"@type":"Organization","@id":"https:\/\/localarch.ai\/zh-hant\/#organization","name":"Local AI Architecture","url":"https:\/\/localarch.ai\/zh-hant\/","logo":{"@type":"ImageObject","inLanguage":"zh-TW","@id":"https:\/\/localarch.ai\/zh-hant\/#\/schema\/logo\/image\/","url":"https:\/\/localarch.ai\/wp-content\/uploads\/2025\/09\/cropped-LocalArch-AI-logo.png","contentUrl":"https:\/\/localarch.ai\/wp-content\/uploads\/2025\/09\/cropped-LocalArch-AI-logo.png","width":400,"height":119,"caption":"Local AI Architecture"},"image":{"@id":"https:\/\/localarch.ai\/zh-hant\/#\/schema\/logo\/image\/"}},{"@type":"Person","@id":"https:\/\/localarch.ai\/zh-hant\/#\/schema\/person\/91a141c7253ea42b1a40cd04c042b399","name":"webeditor","image":{"@type":"ImageObject","inLanguage":"zh-TW","@id":"https:\/\/secure.gravatar.com\/avatar\/4ecc7bad18fce62b20524b26668563f37907995e1838ca8a29a5cb6c98262cee?s=96&d=mm&r=g","url":"https:\/\/secure.gravatar.com\/avatar\/4ecc7bad18fce62b20524b26668563f37907995e1838ca8a29a5cb6c98262cee?s=96&d=mm&r=g","contentUrl":"https:\/\/secure.gravatar.com\/avatar\/4ecc7bad18fce62b20524b26668563f37907995e1838ca8a29a5cb6c98262cee?s=96&d=mm&r=g","caption":"webeditor"},"url":"https:\/\/localarch.ai\/zh-hant\/author\/laeditor\/"}]}},"jetpack_featured_media_url":"https:\/\/localarch.ai\/wp-content\/uploads\/2026\/07\/GPU_Virtualization-scaled.png","jetpack_sharing_enabled":true,"_links":{"self":[{"href":"https:\/\/localarch.ai\/zh-hant\/wp-json\/wp\/v2\/posts\/1047","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/localarch.ai\/zh-hant\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/localarch.ai\/zh-hant\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/localarch.ai\/zh-hant\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/localarch.ai\/zh-hant\/wp-json\/wp\/v2\/comments?post=1047"}],"version-history":[{"count":3,"href":"https:\/\/localarch.ai\/zh-hant\/wp-json\/wp\/v2\/posts\/1047\/revisions"}],"predecessor-version":[{"id":1051,"href":"https:\/\/localarch.ai\/zh-hant\/wp-json\/wp\/v2\/posts\/1047\/revisions\/1051"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/localarch.ai\/zh-hant\/wp-json\/wp\/v2\/media\/1048"}],"wp:attachment":[{"href":"https:\/\/localarch.ai\/zh-hant\/wp-json\/wp\/v2\/media?parent=1047"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/localarch.ai\/zh-hant\/wp-json\/wp\/v2\/categories?post=1047"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/localarch.ai\/zh-hant\/wp-json\/wp\/v2\/tags?post=1047"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}