https://esupport.gigabyte.com/),我們將於上班日後儘速為您處理。","HolidayStart":"","HolidayEnd":"2020/02/10 23:58:00","IsShowOnHomePage":"no","IsShowOnSuppotHome":"yes"}; //Holiday Notice 新版
window.gbtGlobal = {
Razor: {
CountryCode: 'tw',
FormLang: 'TW',
UrlCountryCode: '/tw',
MainFileUploadPathAjax: '/tw/Ajax', //ModuleHeader.js:Header Search 使用
IsB2B: ("False".toLowerCase() === "true"),
IsCN: false,
IsProduction: true,
IsRTL: ("False".toLowerCase() === "true"),
DefaultOnErrorImage: 'https://www.gigabyte.com/Images/Assets/Layout/OnError-Default-Image/OnError-Default_1200x630.png?v=202404301033',
DefaultOnErrorImageSquare: 'https://www.gigabyte.com/Images/Assets/Layout/OnError-Default-Image/OnError-Default_1100x1100.png?v=202404301033',
HolidayInfo : razorHolidayInfo,
},
IsNewIPadOs: CheckIsNewVersionIPad(),
IsMobile: CheckIsMobile()
};
window.CountryCode = "tw";
window.UrlCountryCode = "/tw";
window.GTMCategory = '';
window.GTMAction = '';
window.GTMLabel = '';
window.envConfig = 'globalazurevm';
$(function() {
//Holiday Notice 新版
window.chinaReplaceWord = [];
});
為了在效能、效率、穩定性和擴展性之間達到絕佳平衡,技嘉推出多款支援加速器的伺服器產品,以因應人工智慧時代,需要GPU強大算力的各種應用場景。針對此款搭載 Intel® Gaudi® 3 AI 加速器的G系列伺服器新成員,其擁有8U機箱空間,提供優化的散熱能力藉此充分釋放運算效能。這也是技嘉首款採用8U機身設計的氣冷解方案伺服器,可無縫部署於標準氣冷基礎架構資料中心。\n \n
\n
\n \n
\n
\n \n \n
技嘉最佳化且經過驗證的 – GIGAPOD 人工智慧資料中心解決方案亦為以 Ethernet 網路為核心的可擴充解決方案;支援最新的 Intel® Gaudi® 3 AI 加速器伺服器產品,由於其採開放式軟體與業界標準 Ethernet 網路,在佈署時可大量減少驗證的前置作業時程。於 GIGAPOD 資料中心解決方案中,單一機櫃將配置四台搭載 Intel® Gaudi® 3 AI 加速器伺服器以及後門熱交換器(RDHx),提供最高密度的算力與最佳化的空間部署。\n 想了解有關 GIGAPOD 的更多信息,請參閱:GIGAPOD 整合式人工智慧資料中心解决方案\n","Content":"[]"}],"Name":"overview","GaCodeCategory":null},{"Id":2278,"HashNumber":0,"Title":"專為真實世界對 AI 的需求設計","Description":"","IsShowGigabyteLogIcon":false,"BackGroundInfo":{"Id":2278,"Type":"None","Value":"","FontStyle":"Dark","CssStyleString":"","CssClassNameString":""},"ChildList":[{"Id":0,"ModelType":"AdvantageIcon","Title":"","Description":"","Content":"[{\"Title\":\"輕鬆採用\",\"Description\":\"輕鬆啟用新模型或轉移現有代碼至Intel Gaudi軟體,專為提供生成式 AI 效能、效率而設計,提升開發者的使用便利性與生產力\",\"MediaInfo\":{\"Id\":6871,\"Description\":\"\",\"MediaPath\":\"//static.gigabyte.com/StaticFile/Image/Global/40505512b7117629a096580e60df226d/Feature/6315\",\"MediaPathYoutube\":\"\",\"MediaPathVideo\":\"\",\"MediaPathExt\":\"\",\"MediaPathWebP\":\"\",\"MediaAlt\":\"Support\",\"MediaType\":0,\"SeoAlt\":\"alt=\\\"Support\\\"\",\"SeoTitle\":\"title=\\\"Support\\\"\",\"AngularSeoAlt\":\"ng-alt=\\\"Support\\\"\",\"AngularSeoTitle\":\"ng-title=\\\"Support\\\"\"}},{\"Title\":\"輕鬆擴充\",\"Description\":\"使用已擁有的網路基礎架構,24 個 200 Gigabit 乙太網路(RoCE)連接埠整合至 Intel Gaudi 3 AI 加速器,即使是最大、最複雜的部署也能以具成本效益的方式進行擴展。\",\"MediaInfo\":{\"Id\":6872,\"Description\":\"\",\"MediaPath\":\"//static.gigabyte.com/StaticFile/Image/Global/40505512b7117629a096580e60df226d/Feature/6266\",\"MediaPathYoutube\":\"\",\"MediaPathVideo\":\"\",\"MediaPathExt\":\"\",\"MediaPathWebP\":\"\",\"MediaAlt\":\"connect\",\"MediaType\":0,\"SeoAlt\":\"alt=\\\"connect\\\"\",\"SeoTitle\":\"title=\\\"connect\\\"\",\"AngularSeoAlt\":\"ng-alt=\\\"connect\\\"\",\"AngularSeoTitle\":\"ng-title=\\\"connect\\\"\"}},{\"Title\":\"靈活且強大的運算能力\",\"Description\":\"由兩個互連晶片與以 8 個矩陣乘法引擎 (MME) 和 64 Tensor處理器核心 (TPC)所組成,可於各式工作負載中提供最佳效能。\",\"MediaInfo\":{\"Id\":6873,\"Description\":\"\",\"MediaPath\":\"//static.gigabyte.com/StaticFile/Image/Global/40505512b7117629a096580e60df226d/Feature/6280\",\"MediaPathYoutube\":\"\",\"MediaPathVideo\":\"\",\"MediaPathExt\":\"\",\"MediaPathWebP\":\"\",\"MediaAlt\":\"high_efficiency\",\"MediaType\":0,\"SeoAlt\":\"alt=\\\"high_efficiency\\\"\",\"SeoTitle\":\"title=\\\"high_efficiency\\\"\",\"AngularSeoAlt\":\"ng-alt=\\\"high_efficiency\\\"\",\"AngularSeoTitle\":\"ng-title=\\\"high_efficiency\\\"\"}},{\"Title\":\"高效記憶體密集型運算\",\"Description\":\"內建128GB HBM和96MB的L2快取,有效解決AI訓練和推理工作中常見記憶體瓶頸,加速大型語言模型(LLM)記憶體密集型應用。\",\"MediaInfo\":{\"Id\":6874,\"Description\":\"\",\"MediaPath\":\"//static.gigabyte.com/StaticFile/Image/Global/40505512b7117629a096580e60df226d/Feature/6278\",\"MediaPathYoutube\":\"\",\"MediaPathVideo\":\"\",\"MediaPathExt\":\"\",\"MediaPathWebP\":\"\",\"MediaAlt\":\"gpu\",\"MediaType\":0,\"SeoAlt\":\"alt=\\\"gpu\\\"\",\"SeoTitle\":\"title=\\\"gpu\\\"\",\"AngularSeoAlt\":\"ng-alt=\\\"gpu\\\"\",\"AngularSeoTitle\":\"ng-title=\\\"gpu\\\"\"}}]"}],"Name":"advantage","GaCodeCategory":null},{"Id":2279,"HashNumber":0,"Title":"Intel Gaudi 3 AI 加速器規格","Description":"\n
Model |
Intel® Gaudi® 3 Accelerator |
BF16/FP8 MME TFOPs |
1835 |
BF16 Vector TFLOPs |
28.7 |
MME Units |
8 |
TPC Units |
64 |
HBM Capacity |
128 GB |
HBM Bandwidth |
3.7 TB/s |
On-die SRAM Capacity |
96 MB |
On-die SRAM Bandwidth |
12.8 TB/s |
Networking |
1200 GB/s bidirectional |
Host Interface |
PCIe Gen5 x16 |
Media |
14 Decoders |