{"id":5485,"date":"2024-03-14T09:22:38","date_gmt":"2024-03-14T01:22:38","guid":{"rendered":"https:\/\/www.1ai.net\/?p=5485"},"modified":"2024-03-14T09:22:38","modified_gmt":"2024-03-14T01:22:38","slug":"meta-%e6%96%b0%e5%bb%ba%e4%b8%a4%e5%ba%a7%e6%95%b0%e6%8d%ae%e4%b8%ad%e5%bf%83%e9%9b%86%e7%be%a4%ef%bc%9a%e5%86%85%e5%90%ab%e8%b6%85-4-9-%e4%b8%87%e5%9d%97%e8%8b%b1%e4%bc%9f%e8%be%be-h100-gpu%ef%bc%8c","status":"publish","type":"post","link":"https:\/\/www.1ai.net\/en\/5485.html","title":{"rendered":"Meta builds two new data center clusters: containing more than 49,000 NVIDIA H100 GPUs, dedicated to training Llama3"},"content":{"rendered":"<p data-vmark=\"4f04\"><a href=\"https:\/\/www.1ai.net\/en\/tag\/meta\" title=\"[View articles tagged with [Meta]]\" target=\"_blank\" >Meta<\/a> The company unveiled two new data center clusters via an official press release on Local 12, which the company is looking to<a href=\"https:\/\/www.1ai.net\/en\/tag\/%e8%8b%b1%e4%bc%9f%e8%be%be\" title=\"Look at the article with the label\" target=\"_blank\" >Nvidia<\/a>of <a href=\"https:\/\/www.1ai.net\/en\/tag\/gpu\" title=\"_OTHER ORGANISER\" target=\"_blank\" >GPU<\/a>that stand out in AI-focused development.<\/p>\n<p data-vmark=\"f3bf\">The sole purpose of the two data centers is said to be AI research and the development of large language models in consumer-specific application areas (IT House note: including sound or image recognition).<span class=\"accentTextColor\">Each cluster contains 24,576 NVIDIA H100 AI GPUs, which will be used for their own big language models <a href=\"https:\/\/www.1ai.net\/en\/tag\/llama-3\" title=\"[See articles with [Llama 3] labels]\" target=\"_blank\" >Llama 3<\/a> enhancement<\/span>.<\/p>\n<p data-vmark=\"b997\"><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-5486\" title=\"aa432ce2-e4a5-4cf3-aa13-11750c8ef6a5\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2024\/03\/aa432ce2-e4a5-4cf3-aa13-11750c8ef6a5.jpg\" alt=\"aa432ce2-e4a5-4cf3-aa13-11750c8ef6a5\" width=\"1999\" height=\"1432\" \/><\/p>\n<p data-vmark=\"e9ae\">Both new data center clusters feature 400Gbps interconnections, with one cluster using Meta's proprietary fabric solution based on the Arista 7800 and the other using NVIDIA's Quantum2 InfiniBand fabric to ensure a seamless interconnect experience.<\/p>\n<p data-vmark=\"a820\">In addition, the cluster is based on Meta's own open GPU Grand Teton AI platform, which leverages the capabilities of modern gas pedals by increasing host-to-GPU bandwidth and compute power.<\/p>\n<p data-vmark=\"e31c\">Meta officials say that the efficiency of the high-performance network fabric and key storage decisions of these clusters, combined with the H100 GPUs in each cluster, can support larger, more complex models, paving the way for advances in general-purpose AI product development and AI research.<\/p>\n<p data-vmark=\"c702\">Meta CEO Zuckerberg announced that the company is building a massive infrastructure. \"The prediction is that by the end of this year, we will have about 350,000 NVIDIA H100 accelerator cards, which is the computing power equivalent of 600,000 H100s if you count other GPUs.\"<\/p>","protected":false},"excerpt":{"rendered":"<p>Meta, which announced two new data center clusters via an official press release on 12 local time, is looking to differentiate itself in AI-focused development with NVIDIA's GPUs. The sole purpose of the two data centers is said to be AI research and development of large language models in consumer-specific applications (IT House note: including sound or image recognition), and each cluster contains 24,576 NVIDIA H100 AI GPUs, which will be used for the training of its own large language model, Llama 3. Both new data center clusters have 400Gbps interconnectivity, with one cluster utilizing Meta's own Arista 7800-based Fabr<\/p>","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[146],"tags":[415,1671,297,239],"collection":[],"class_list":["post-5485","post","type-post","status-publish","format-standard","hentry","category-news","tag-gpu","tag-llama-3","tag-meta","tag-239"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/5485","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/comments?post=5485"}],"version-history":[{"count":0,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/5485\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/media?parent=5485"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/categories?post=5485"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/tags?post=5485"},{"taxonomy":"collection","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/collection?post=5485"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}