{"id":80455,"date":"2024-09-02T12:58:21","date_gmt":"2024-09-02T04:58:21","guid":{"rendered":"https:\/\/www.ioiotimes.com\/?p=80455"},"modified":"2024-09-02T12:58:21","modified_gmt":"2024-09-02T04:58:21","slug":"amd%e6%96%bcmlperf%e5%9f%ba%e6%ba%96%e6%b8%ac%e8%a9%a6%e5%b1%95%e7%8f%be%e5%85%a8%e6%96%b0ai%e9%a0%98%e5%85%88%e5%9c%b0%e4%bd%8d%ef%bc%8c%e6%8f%ad%e6%9b%89-amd-instinct-mi300x%e5%8a%a0%e9%80%9f","status":"publish","type":"post","link":"https:\/\/www.ioiotimes.com\/?p=80455","title":{"rendered":"AMD\u65bcMLPerf\u57fa\u6e96\u6e2c\u8a66\u5c55\u73fe\u5168\u65b0AI\u9818\u5148\u5730\u4f4d\uff0c\u63ed\u66c9 AMD Instinct MI300X\u52a0\u901f\u5668\u4e0a\u7684MLPerf\u7d50\u679c"},"content":{"rendered":"\n

AMD \u5f88\u9ad8\u8208\u5730\u767c\u8868\u9996\u6b21 MLPerf \u57fa\u6e96\u6e2c\u8a66\u4e4b\u7d50\u679c\uff0c\u4e26\u5728\u66f4\u65b0\u7684MLCommons\u7db2\u7ad9<\/a>\u4e0a\u5c55\u793a\u642d\u8f09 AMD Instinct\u2122 MI300X GPU \u9032\u884c\u6e2c\u8a66\u7684\u7d50\u679c\u3002<\/p>\n\n\n\n

MLPerf \u6709\u591a\u7a2e\u57fa\u6e96\u6e2c\u8a66\u985e\u578b\uff0c\u5305\u62ec MLPerf Inference: Datacenter<\/a>\u3001MLPerf Training<\/a> \u4ee5\u53ca\u65bc 2024 \u5e74\u63a8\u51fa\u7684\u5c16\u7aef\u5927\u578b\u751f\u6210\u5f0f\u4eba\u5de5\u667a\u6167\uff08AI\uff09\u8a9e\u8a00\u6a21\u578b Llama 2 70B<\/a>\u3002AMD \u9078\u64c7\u4e86\u53d7\u6b61\u8fce\u4e14\u5177\u6709\u91cd\u8981\u610f\u7fa9\u7684\u5927\u578b\u8a9e\u8a00\u6a21\u578b\uff08LLM\uff09Llama 2 70B \u9032\u884c\u9996\u6b21 MLPerf \u57fa\u6e96\u6e2c\u8a66\u3002<\/p>\n\n\n\n

\"\"
\u25b2Showcasing performance of CPU-GPU combination for AI workload<\/figcaption><\/figure>\n\n\n\n

<\/p>\n\n\n\n

AMD \u5728 MLPerf Inference v4.1<\/a>\uff08MLPerf 4.1 \u63a8\u8ad6\uff09\u4e2d\u70ba Llama 2 70B \u63d0\u51fa 3 \u9805\u6210\u679c\u4e26\u5c55\u793a\u51fa\uff1a<\/p>\n\n\n\n

\u00b7\u00a0\u00a0\u00a0\u00a0\u5e02\u5834\u4e0a\u6700\u9069\u5408\u7528\u65bc AI \u4efb\u52d9\u7684 AMD CPU \u548c GPU \u7d44\u5408\u3002
\u00b7\u00a0\u00a0\u00a0\u00a0MI300X \u64c1\u6709 192GB \u7684\u5927\u5bb9\u91cf\u8a18\u61b6\u9ad4\uff0c\u8b93\u55ae\u4e00 MI300X GPU \u53ef\u57f7\u884c\u6574\u500b Llama 2 70B \u6a21\u578b\uff0c\u800c\u5927\u591a\u6578\u7af6\u722d\u5c0d\u624b\u7684 GPU \u5247\u9700\u8981\u5c07\u8a72\u6a21\u578b\u62c6\u5206\u81f3\u591a\u500b\u52a0\u901f\u5668\u4e0a\u57f7\u884c\u3002
\u00b7\u00a0\u00a0\u00a0\u00a0AMD \u65b0\u4e00\u4ee3 CPU \u5982\u4f55\u63d0\u5347 AI \u4efb\u52d9\u7684\u6548\u80fd\u3002<\/p>\n\n\n\n

<\/p>\n\n\n\n

\"\"
\u25b2Single GPU Running the Entire Llama 2 70B Model<\/figcaption><\/figure>\n\n\n\n

<\/p>\n\n\n\n

\"\"
\u25b2LLaMa\u00a03.1 (405B) Estimated Memory Requirements vs Available GPU Memory (Source \u2013 Artificial Analysis<\/a>)<\/figcaption><\/figure>\n\n\n\n

<\/p>\n\n\n\n

\u5728 MLPerf 4.1 \u63a8\u8ad6\u4e2d\uff0cAMD \u7684\u7d50\u679c\u5c55\u73fe\u51fa\u642d\u8f09 ROCm \u7684 MI300X \u70ba Llama 2 70B \u7b49\u5927\u578b LLM \u63d0\u4f9b\u5353\u8d8a\u7684\u63a8\u8ad6\u6548\u80fd\uff0c\u4e14\u8207\u5176\u4ed6\u696d\u754c\u9818\u5148\u8005\u5982 Nvidia H100 \u5728 Llama 2 70B LLM \u6a21\u578b\u4e0a\u7684\u6e2c\u8a66\u7d50\u679c\u4e0d\u5206\u8ed2\u8f0a\u3002<\/p>\n\n\n\n

\u6b32\u4e86\u89e3\u66f4\u591a\u76f8\u95dc\u8cc7\u8a0a\uff0c\u8acb\u53c3\u95b1 AMD\u90e8\u843d\u683c\u6587\u7ae0<\/a>\u3002<\/p>\n\n\n\n

<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

<\/p>\n\n\n\n

<\/p>\n\n\n\n

<\/p>\n\n\n\n

<\/p>\n\n\n\n


\ud83d\udfe6\u73fe\u5728\u5c31\u52a0\u5165 ioioTIMES \u81c9\u66f8\u7c89\u7d72\u5718<\/a> \u66f4\u591a\u4e92\u52d5\u3001\u66f4\u591a\u597d\u5eb7\u650f\u62b5\u52a0!!<\/strong>
\ud83d\udfe6\u6211\u5011\u6709LINE TODAY<\/a>\u983b\u9053\u4e86\uff0c\u5feb\u4f86\u8ffd\u8e2a\u6211\u5011\u5427!!–\u6700\u65b0\u79d1\u6280\u65b0\u805e \u76e1\u5728\u4f60\u624b<\/strong><\/h4>\n","protected":false},"excerpt":{"rendered":"

AMD \u5f88\u9ad8\u8208\u5730\u767c\u8868\u9996\u6b21 MLPer<\/p>\n","protected":false},"author":3,"featured_media":80459,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"rank_math_lock_modified_date":false,"footnotes":""},"categories":[13],"tags":[35,181,580,9710,6360],"class_list":["post-80455","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-news","tag-ai","tag-amd","tag-focus","tag-instinct-mi300x","tag-mlperf"],"_links":{"self":[{"href":"https:\/\/www.ioiotimes.com\/index.php?rest_route=\/wp\/v2\/posts\/80455"}],"collection":[{"href":"https:\/\/www.ioiotimes.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.ioiotimes.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.ioiotimes.com\/index.php?rest_route=\/wp\/v2\/users\/3"}],"replies":[{"embeddable":true,"href":"https:\/\/www.ioiotimes.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=80455"}],"version-history":[{"count":2,"href":"https:\/\/www.ioiotimes.com\/index.php?rest_route=\/wp\/v2\/posts\/80455\/revisions"}],"predecessor-version":[{"id":80464,"href":"https:\/\/www.ioiotimes.com\/index.php?rest_route=\/wp\/v2\/posts\/80455\/revisions\/80464"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.ioiotimes.com\/index.php?rest_route=\/wp\/v2\/media\/80459"}],"wp:attachment":[{"href":"https:\/\/www.ioiotimes.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=80455"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.ioiotimes.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=80455"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.ioiotimes.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=80455"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}