{"id":14713,"date":"2024-12-03T14:32:10","date_gmt":"2024-12-03T06:32:10","guid":{"rendered":"https:\/\/www.aisharenet.com\/?p=14713"},"modified":"2025-07-17T00:04:44","modified_gmt":"2025-07-16T16:04:44","slug":"gpustack","status":"publish","type":"post","link":"https:\/\/www.kdjingpai.com\/de\/gpustack\/","title":{"rendered":"GPUStack\uff1a\u7ba1\u7406GPU\u96c6\u7fa4\u8fd0\u884c\u5927\u8bed\u8a00\u6a21\u578b\uff0c\u5feb\u901f\u96c6\u6210LLMs\u5e38\u7528\u63a8\u7406\u670d\u52a1\u3002"},"content":{"rendered":"<p>GPUStack \u662f\u4e00\u4e2a\u5f00\u6e90\u7684 GPU \u96c6\u7fa4\u7ba1\u7406\u5de5\u5177\uff0c\u4e13\u4e3a\u8fd0\u884c\u5927\u8bed\u8a00\u6a21\u578b\uff08LLMs\uff09\u800c\u8bbe\u8ba1\u3002\u5b83\u652f\u6301\u591a\u79cd\u786c\u4ef6\uff0c\u5305\u62ec Apple MacBooks\u3001Windows PCs \u548c Linux \u670d\u52a1\u5668\uff0c\u80fd\u591f\u8f7b\u677e\u6269\u5c55 GPU \u6570\u91cf\u548c\u8282\u70b9\uff0c\u4ee5\u6ee1\u8db3\u4e0d\u65ad\u589e\u957f\u7684\u8ba1\u7b97\u9700\u6c42\u3002GPUStack \u63d0\u4f9b\u5206\u5e03\u5f0f\u63a8\u7406\u529f\u80fd\uff0c\u652f\u6301\u5355\u8282\u70b9\u591a GPU \u548c\u591a\u8282\u70b9\u63a8\u7406\u4e0e\u670d\u52a1\uff0c\u517c\u5bb9 OpenAI API\uff0c\u7b80\u5316\u7528\u6237\u548c API \u5bc6\u94a5\u7ba1\u7406\uff0c\u5e76\u5b9e\u65f6\u76d1\u63a7 GPU \u6027\u80fd\u548c\u5229\u7528\u7387\u3002\u5176\u8f7b\u91cf\u7ea7 Python \u5305\u8bbe\u8ba1\uff0c\u786e\u4fdd\u6700\u5c0f\u7684\u4f9d\u8d56\u548c\u64cd\u4f5c\u5f00\u9500\uff0c\u662f\u5f00\u53d1\u8005\u548c\u7814\u7a76\u4eba\u5458\u7684\u7406\u60f3\u9009\u62e9\u3002<\/p>\n<p><img decoding=\"async\" title=\"GPUStack\uff1a\u7ba1\u7406GPU\u96c6\u7fa4\u8fd0\u884c\u5927\u8bed\u8a00\u6a21\u578b\uff0c\u5feb\u901f\u96c6\u6210LLMs\u5e38\u7528\u63a8\u7406\u670d\u52a1\u3002-1\" src=\"https:\/\/www.kdjingpai.com\/wp-content\/uploads\/2024\/12\/649b4b30f19bcac.jpg\" alt=\"GPUStack\uff1a\u7ba1\u7406GPU\u96c6\u7fa4\u8fd0\u884c\u5927\u8bed\u8a00\u6a21\u578b\uff0c\u5feb\u901f\u96c6\u6210LLMs\u5e38\u7528\u63a8\u7406\u670d\u52a1\u3002-1\" \/><\/p>\n<p>&nbsp;<\/p>\n<h2>\u529f\u80fd\u5217\u8868<\/h2>\n<ul>\n<li>\u652f\u6301\u591a\u79cd\u786c\u4ef6\uff1a\u517c\u5bb9 Apple Metal\u3001NVIDIA CUDA\u3001Ascend CANN\u3001Moore Threads MUSA \u7b49\u3002<\/li>\n<li>\u5206\u5e03\u5f0f\u63a8\u7406\uff1a\u652f\u6301\u5355\u8282\u70b9\u591a GPU \u548c\u591a\u8282\u70b9\u63a8\u7406\u4e0e\u670d\u52a1\u3002<\/li>\n<li>\u591a\u79cd\u63a8\u7406\u540e\u7aef\uff1a\u652f\u6301 llama-box\uff08llama.cpp\uff09\u548c vLLM\u3002<\/li>\n<li>\u8f7b\u91cf\u7ea7 Python \u5305\uff1a\u6700\u5c0f\u4f9d\u8d56\u548c\u64cd\u4f5c\u5f00\u9500\u3002<\/li>\n<li>OpenAI \u517c\u5bb9 API\uff1a\u63d0\u4f9b\u4e0e OpenAI \u6807\u51c6\u517c\u5bb9\u7684 API \u670d\u52a1\u3002<\/li>\n<li>\u7528\u6237\u548c API \u5bc6\u94a5\u7ba1\u7406\uff1a\u7b80\u5316\u7528\u6237\u548c API \u5bc6\u94a5\u7ba1\u7406\u3002<\/li>\n<li>GPU \u6027\u80fd\u76d1\u63a7\uff1a\u5b9e\u65f6\u76d1\u63a7 GPU \u6027\u80fd\u548c\u5229\u7528\u7387\u3002<\/li>\n<li>\u4ee4\u724c\u4f7f\u7528\u548c\u901f\u7387\u76d1\u63a7\uff1a\u6709\u6548\u7ba1\u7406\u4ee4\u724c\u4f7f\u7528\u548c\u901f\u7387\u9650\u5236\u3002<\/li>\n<\/ul>\n<p>&nbsp;<\/p>\n<h2>\u4f7f\u7528\u5e2e\u52a9<\/h2>\n<h3>\u5b89\u88c5\u6d41\u7a0b<\/h3>\n<h4>Linux \u6216 MacOS<\/h4>\n<ol>\n<li>\u6253\u5f00\u7ec8\u7aef\u3002<\/li>\n<li>\u8fd0\u884c\u4ee5\u4e0b\u547d\u4ee4\u5b89\u88c5 GPUStack\uff1a<\/li>\n<\/ol>\n<pre><code>   curl -sfL https:\/\/get.gpustack.ai | sh -s -\r\n<\/code><\/pre>\n<ol start=\"3\">\n<li>\u5b89\u88c5\u5b8c\u6210\u540e\uff0cGPUStack \u5c06\u4f5c\u4e3a\u670d\u52a1\u8fd0\u884c\u5728 systemd \u6216 launchd \u7cfb\u7edf\u4e0a\u3002<\/li>\n<\/ol>\n<h4>Windows<\/h4>\n<ol>\n<li>\u4ee5\u7ba1\u7406\u5458\u8eab\u4efd\u8fd0\u884c PowerShell\uff08\u907f\u514d\u4f7f\u7528 PowerShell ISE\uff09\u3002<\/li>\n<li>\u8fd0\u884c\u4ee5\u4e0b\u547d\u4ee4\u5b89\u88c5 GPUStack\uff1a<\/li>\n<\/ol>\n<pre><code>   Invoke-Expression (Invoke-WebRequest -Uri \"https:\/\/get.gpustack.ai\" -UseBasicParsing).Content\r\n<\/code><\/pre>\n<h3>\u4f7f\u7528\u6307\u5357<\/h3>\n<h4>\u521d\u59cb\u8bbe\u7f6e<\/h4>\n<ol>\n<li>\u8bbf\u95ee GPUStack UI\uff1a\u5728\u6d4f\u89c8\u5668\u4e2d\u6253\u5f00 <code>http:\/\/myserver<\/code>\u3002<\/li>\n<li>\u4f7f\u7528\u9ed8\u8ba4\u7528\u6237\u540d <code>admin<\/code> \u548c\u521d\u59cb\u5bc6\u7801\u767b\u5f55\u3002\u83b7\u53d6\u521d\u59cb\u5bc6\u7801\u7684\u65b9\u6cd5\uff1a\n<ul>\n<li>Linux \u6216 MacOS\uff1a\u8fd0\u884c <code>cat \/var\/lib\/gpustack\/initial_admin_password<\/code>\u3002<\/li>\n<li>Windows\uff1a\u8fd0\u884c <code>Get-Content -Path \"$env:APPDATA\\gpustack\\initial_admin_password\" -Raw<\/code>\u3002<\/li>\n<\/ul>\n<\/li>\n<\/ol>\n<h4>\u521b\u5efa API \u5bc6\u94a5<\/h4>\n<ol>\n<li>\u767b\u5f55 GPUStack UI \u540e\uff0c\u70b9\u51fb\u5bfc\u822a\u83dc\u5355\u4e2d\u7684\u201cAPI Keys\u201d\u3002<\/li>\n<li>\u70b9\u51fb\u201cNew API Key\u201d\u6309\u94ae\uff0c\u586b\u5199\u540d\u79f0\u5e76\u4fdd\u5b58\u3002<\/li>\n<li>\u590d\u5236\u751f\u6210\u7684 API \u5bc6\u94a5\u5e76\u59a5\u5584\u4fdd\u5b58\uff08\u4ec5\u5728\u521b\u5efa\u65f6\u53ef\u89c1\uff09\u3002<\/li>\n<\/ol>\n<h4>\u4f7f\u7528 API<\/h4>\n<ol>\n<li>\u8bbe\u7f6e\u73af\u5883\u53d8\u91cf\uff1a<\/li>\n<\/ol>\n<pre><code>   export GPUSTACK_API_KEY=myapikey\r\n<\/code><\/pre>\n<ol start=\"2\">\n<li>\u4f7f\u7528 curl \u8bbf\u95ee OpenAI \u517c\u5bb9 API\uff1a<\/li>\n<\/ol>\n<pre><code>   curl http:\/\/myserver\/v1-openai\/chat\/completions \\\r\n-H \"Content-Type: application\/json\" \\\r\n-H \"Authorization: Bearer $GPUSTACK_API_KEY\" \\\r\n-d '{\r\n\"model\": \"llama3.2\",\r\n\"messages\": [\r\n{\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\r\n{\"role\": \"user\", \"content\": \"Hello!\"}\r\n],\r\n\"stream\": true\r\n}'\r\n<\/code><\/pre>\n<h3>\u8fd0\u884c\u548c\u804a\u5929<\/h3>\n<ol>\n<li>\u5728\u7ec8\u7aef\u4e2d\u8fd0\u884c\u4ee5\u4e0b\u547d\u4ee4\u4e0e llama3.2 \u6a21\u578b\u804a\u5929\uff1a<\/li>\n<\/ol>\n<pre><code>   gpustack chat llama3.2 \"tell me a joke.\"\r\n<\/code><\/pre>\n<ol start=\"2\">\n<li>\u5728 GPUStack UI \u4e2d\u70b9\u51fb\u201cPlayground\u201d\u8fdb\u884c\u4ea4\u4e92\u3002<\/li>\n<\/ol>\n<h3>\u76d1\u63a7\u548c\u7ba1\u7406<\/h3>\n<ol>\n<li>\u5b9e\u65f6\u76d1\u63a7 GPU \u6027\u80fd\u548c\u5229\u7528\u7387\u3002<\/li>\n<li>\u7ba1\u7406\u7528\u6237\u548c API \u5bc6\u94a5\uff0c\u8ddf\u8e2a\u4ee4\u724c\u4f7f\u7528\u548c\u901f\u7387\u3002<\/li>\n<\/ol>\n<h3>\u652f\u6301\u7684\u6a21\u578b\u548c\u5e73\u53f0<\/h3>\n<ul>\n<li>\u652f\u6301\u7684\u6a21\u578b\uff1aLLaMA\u3001Mistral 7B\u3001Mixtral MoE\u3001Falcon\u3001Baichuan\u3001Yi\u3001Deepseek\u3001Qwen\u3001Phi\u3001Grok-1 \u7b49\u3002<\/li>\n<li>\u652f\u6301\u7684\u591a\u6a21\u6001\u6a21\u578b\uff1aLlama3.2-Vision\u3001Pixtral\u3001Qwen2-VL\u3001LLaVA\u3001InternVL2 \u7b49\u3002<\/li>\n<li>\u652f\u6301\u7684\u5e73\u53f0\uff1aMacOS\u3001Linux\u3001Windows\u3002<\/li>\n<li>\u652f\u6301\u7684\u52a0\u901f\u5668\uff1aApple Metal\u3001NVIDIA CUDA\u3001Ascend CANN\u3001Moore Threads MUSA\uff0c\u672a\u6765\u8ba1\u5212\u652f\u6301 AMD ROCm\u3001Intel oneAPI\u3001Qualcomm AI Engine\u3002<\/li>\n<\/ul>\n<h3>\u6587\u6863\u548c\u793e\u533a<\/h3>\n<ul>\n<li>\u5b98\u65b9\u6587\u6863\uff1a\u8bbf\u95ee <a href=\"https:\/\/docs.gpustack.ai\/\">GPUStack \u6587\u6863<\/a> \u83b7\u53d6\u5b8c\u6574\u7684\u4f7f\u7528\u6307\u5357\u548c API \u6587\u6863\u3002<\/li>\n<li>\u8d21\u732e\u6307\u5357\uff1a\u9605\u8bfb <a href=\"https:\/\/github.com\/gpustack\/gpustack\">\u8d21\u732e\u6307\u5357<\/a> \u4e86\u89e3\u5982\u4f55\u4e3a GPUStack \u505a\u51fa\u8d21\u732e\u3002<\/li>\n<\/ul>\n","protected":false},"excerpt":{"rendered":"<p>GPUStack \u662f\u4e00\u4e2a\u5f00\u6e90\u7684 GPU \u96c6\u7fa4\u7ba1\u7406\u5de5\u5177\uff0c\u4e13\u4e3a\u8fd0\u884c\u5927\u8bed\u8a00\u6a21\u578b\uff08LLMs\uff09\u800c\u8bbe\u8ba1\u3002\u5b83\u652f\u6301\u591a\u79cd\u786c\u4ef6\uff0c\u5305\u62ec Apple MacBooks\u3001Windows PCs \u548c Linux \u670d\u52a1\u5668\uff0c\u80fd\u591f\u8f7b\u677e\u6269\u5c55 GPU \u6570\u91cf\u548c\u8282\u70b9\uff0c\u4ee5\u6ee1\u8db3\u4e0d\u65ad\u589e\u957f&#8230;<\/p>\n","protected":false},"author":1,"featured_media":32591,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[425,496,20],"tags":[208],"class_list":["post-14713","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai-professional","category-ai-basic-services","category-tool","tag-aikaifangfuwu"],"_links":{"self":[{"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/posts\/14713","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/comments?post=14713"}],"version-history":[{"count":0,"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/posts\/14713\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/media\/32591"}],"wp:attachment":[{"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/media?parent=14713"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/categories?post=14713"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/tags?post=14713"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}