親愛的客戶您好,因 2019/6/20(四)晚上18:00~24:00 台電施工工程,0800技術專線暫停服務,於 2019/6/21(五)上午09:30 恢復正常服務;服務暫停期間,如果有產品相關問題建議可來信 eSupport 技術支援(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 = []; });

作為 Blackwell 架構系列的旗艦產品,NVIDIA GB200 NVL72 連接 36 個基於ARM架構的 Grace CPU 和 72 個 Blackwell GPU 並採用全液冷設計。這一機架級配置透過最新的 NVLink 技術將所有節點相互串聯,實現「一個大 GPU」的效能。\n\n\n\t\t\t

此 NVIDIA GB200 NVL72單一機架的百萬兆級電腦,在大型語言模型推論效能上超越上一代 HGX H100 GPU 30 倍,在訓練效能上提升 4 倍,同時實現了降低25 倍能源消耗。憑藉無與倫比的互聯速度和能源效率,GB200 NVL72 成為 AI 和 HPC 工作負載的新基準。\n\t\t\t

\n\t\t\t\t
\n\t\t\t\t\t
\"\"\n\t\t\t\t\t

XN14-CB0-LA01 運算節點\n\n\n\t\t\t\t\t
    \n\t\t\t\t\t\t
  • 2 x GB200 Grace Blackwell 超級晶片\n\t\t\t\t\t\t
  • 2 x 384 GB HBM3e GPU 記憶體與 16TB/s 頻寬\n\t\t\t\t\t\t
  • 2 x 480 GB LPDDR5X CPU 記憶體與 152GB/s 頻寬\n\t\t\t\t\t\t
  • 8 x E1.S Gen5 NVMe 硬碟\n\t\t\t\t\t \n\t\t\t\t\n\t\t\t \n\t\t\t\n\t\t\n \n\t","Content":"[]"},{"Id":0,"ModelType":"NormalType","Title":"NVIDIA HGX™ B200:全球頂尖的人工智慧運算平台","Description":"NVIDIA HGX™ B200 採用先進的 Blackwell 架構。透過模組化的 NVIDIA NVSwitch™ 設計,與上一代相比,它在運算效能和能源效率上實現了超過10倍的提升。

    為應對前所未有的效能提升與熱功耗所衍生的散熱問題,技嘉推出兩種形式的伺服器選擇,包括 8U – 空氣冷卻和 4U – 液冷選項,確保相容於現有基礎架構。技嘉的高效能運算伺服器旨在提升效能與提供高可用性,包括:
    \n
      \n
    1. 全面支援全高擴展卡,能容納多組 DPU 和 SuperNIC。\n
    2. 後抽式 PCIe 擴充槽固定架與前抽式主機板 / GPU 托盤設計,簡化維護流程。\n
    3. 可熱插拔、完全冗餘的電源供應器,並提供多種電源連接埠選項,增加靈活性。\n \n\n

      \n \n
      其他功能,如支援 RDMA 的 PCIe 交換器設計,可實現 GPU 與 GPU 間的直接資料傳輸,確保技嘉伺服器在進行大規模加速運算負載時,擁有高度可靠的效能與效率。 \n \n \n
      \n
      \n\t\t
      \n\t\t\t\t
      \n\t\t\t\t\t
      \"Flexible\n\t\t\t\t\t

      NVIDIA HGX™ B200 8-GPU\n\t\t\t\t\t
        \n\t\t\t\t\t\t
      • 8 x NVIDIA Blackwell GPUs \n\t\t\t\t\t\t
      • 1,440GB of GPU 記憶體\n\t\t\t\t\t\t
      • 72 petaFLOPS 訓練效能\n\t\t\t\t\t\t
      • 144 petaFLOPS 推論效能\n\t\t\t\t\t\t
      • 1,800GB/s GPU-to-GPU 頻寬與 NVIDIA® NVLink™ 及 NVSwitch™\n\t\t\t\t\t\t \n\t\t\t\t\t\t \n\t\t\t\t\t \n\t\t\t\t\n\t\t\t\t
        \n\t\t\t\t\t
        \"Flexible\n\t\t\t\t\t

        G893 8U HPC/AI 伺服器\n\t\t\t\t\t
          \n\t\t\t\t\t\t
        • NVIDIA HGX™ B200 8-GPU\n\t\t\t\t\t\t
        • 雙 5th/4th Gen Intel® Xeon® 可擴展或 雙 AMD EPYC™ 9005/9004 系列處理器\n\t\t\t\t\t\t
        • 2 x 10Gb/s LAN \n\t\t\t\t\t\t
        • 8 x 2.5\" Gen5 NVMe 熱插拔硬碟槽\n\t\t\t\t\t\t
        • 12 x 全高半長 PCIe Gen5 x16 插槽\n\t\t\t\t\t\t
        • 12 x 3000W 80 PLUS 鈦金冗餘電源供應器\n\t\t\t\t\t \n\t\t\t\t\n\t\t\t
          \n\t\t\t\t\t
          \"Flexible\n\t\t\t\t\t

          G4L3 4U HPC/AI 伺服器\n\t\t\t\t\t
            \n\t\t\t\t\t\t
          • 液冷 NVIDIA HGX™ B200 8-GPU\n\t\t\t\t\t\t
          • 雙 5th/4th Gen Intel® Xeon® 可擴展或 雙 AMD EPYC™ 9005/9004 系列處理器\n\t\t\t\t\t\t
          • 2 x 10Gb/s LAN \n\t\t\t\t\t\t
          • 8 x 2.5\" Gen5 NVMe 熱插拔硬碟槽\n\t\t\t\t\t\t
          • 12 x 全高半長 PCIe Gen5 x16 插槽\n\t\t\t\t\t\t
          • 8 x 3000W 80 PLUS 鈦金冗餘電源供應器\n\t\t\t\t\t \n\t\t\t\t\n\t\t \n\t\t\n\t","Content":"[]"}],"Name":"videos","GaCodeCategory":null,"IsH1Title":false},{"Id":2344,"HashNumber":0,"Title":"Blackwell 下一代人工智慧超級晶片","Description":"","IsShowGigabyteLogIcon":false,"BackGroundInfo":{"Id":2344,"Type":"None","Value":"","FontStyle":"Dark","CssStyleString":"","CssClassNameString":""},"ChildList":[{"Id":0,"ModelType":"UploadIcon","Title":"","Description":"","Content":"[{\"Title\":\"高效能運算\",\"Link\":\"\",\"TargetHtmlString\":\"\",\"HrefHtmlString\":\"\",\"Description\":\"在高效能運算 (HPC) 應用中,關鍵在於透過數值分析、模擬和運算來解決複雜問題以獲取洞見。雖然傳統上對加速器的依賴性低,但近年通用圖形處理器 (GPGPU) 的強大平行運算能力在近年來大幅推動了高效能運算的發展,混合配置亦成為現代超級計算機的重要趨勢。\",\"MediaInfo\":{\"Id\":7216,\"Description\":\"\",\"MediaPath\":\"//static.gigabyte.com/StaticFile/Image/Global/7c866bc491cc4d3ba3ee746e77eee6bf/ModelSectionChildItem/7216\",\"MediaPathYoutube\":\"\",\"MediaPathVideo\":\"\",\"MediaPathExt\":\"\",\"MediaPathWebP\":\"\",\"MediaAlt\":\"\",\"MediaType\":0,\"SeoAlt\":\"\",\"SeoTitle\":\"\",\"AngularSeoAlt\":\"\",\"AngularSeoTitle\":\"\"}},{\"Title\":\"人工智慧\",\"Link\":\"\",\"TargetHtmlString\":\"\",\"HrefHtmlString\":\"\",\"Description\":\"隨著人工智慧(AI)技術的迅速發展,從一般應用到不斷演進的深度學習,通用圖形處理器 (GPGPU) 已成為改變產業遊戲規則的關鍵技術。GPGPU 的平行處理能力能高效處理龐大數據集和複雜演算法,這對 AI 模型的訓練與部署至關重要。AI技術幫助各行各業在生產、營收與工作效率上實現更快速且更智慧的目標,成為產業升級的核心動力。\",\"MediaInfo\":{\"Id\":7217,\"Description\":\"\",\"MediaPath\":\"//static.gigabyte.com/StaticFile/Image/Global/7c866bc491cc4d3ba3ee746e77eee6bf/ModelSectionChildItem/7217\",\"MediaPathYoutube\":\"\",\"MediaPathVideo\":\"\",\"MediaPathExt\":\"\",\"MediaPathWebP\":\"\",\"MediaAlt\":\"\",\"MediaType\":0,\"SeoAlt\":\"\",\"SeoTitle\":\"\",\"AngularSeoAlt\":\"\",\"AngularSeoTitle\":\"\"}},{\"Title\":\"自然科學及工程模擬\",\"Link\":\"\",\"TargetHtmlString\":\"\",\"HrefHtmlString\":\"\",\"Description\":\"科學研究涵蓋從物理、化學到地理、生物科學等多元學科,對算力的需求極為廣泛。借助GPU加速運算叢集,仿真模擬和建模工作在平行處理能力的支援下實現飛躍發展。更快速的運算與精確的結果,使研究人員能有效分析大量數據,進行深入實驗,推動科學領域的創新與突破。\",\"MediaInfo\":{\"Id\":7218,\"Description\":\"\",\"MediaPath\":\"//static.gigabyte.com/StaticFile/Image/Global/7c866bc491cc4d3ba3ee746e77eee6bf/ModelSectionChildItem/7218\",\"MediaPathYoutube\":\"\",\"MediaPathVideo\":\"\",\"MediaPathExt\":\"\",\"MediaPathWebP\":\"\",\"MediaAlt\":\"\",\"MediaType\":0,\"SeoAlt\":\"\",\"SeoTitle\":\"\",\"AngularSeoAlt\":\"\",\"AngularSeoTitle\":\"\"}}]"}],"Name":"applications","GaCodeCategory":null,"IsH1Title":false},{"Id":2345,"HashNumber":0,"Title":"技嘉AI伺服器","Description":"","IsShowGigabyteLogIcon":false,"BackGroundInfo":{"Id":2345,"Type":"None","Value":"","FontStyle":"Dark","CssStyleString":"","CssClassNameString":""},"ChildList":[{"Id":0,"ModelType":"Product","Title":"","Description":"","Content":"[{\"Id\":7241,\"TableKey\":10082,\"Name\":\"G893-SD1-AAX5\",\"Link\":\"/tw/Enterprise/GPU-Server/G893-SD1-AAX5\",\"Title\":\"\",\"Description\":\"HPC/AI Server - 5th/4th Gen Intel® Xeon® Scalable - 8U DP NVIDIA HGX™ B200 8-GPU\",\"MediaInfo\":{\"Id\":7241,\"Description\":\"\",\"MediaPath\":\"//static.gigabyte.com/StaticFile/Image/Global/41889d347cf472a51bc922505a230efd/Product/43327\",\"MediaPathYoutube\":\"\",\"MediaPathVideo\":\"\",\"MediaPathExt\":\"\",\"MediaPathWebP\":\"\",\"MediaAlt\":\"\",\"MediaType\":0,\"SeoAlt\":\"\",\"SeoTitle\":\"\",\"AngularSeoAlt\":\"\",\"AngularSeoTitle\":\"\"},\"IsAiAuto\":false,\"ApplicationItem\":[{\"Id\":0,\"Url\":\"/tw/Enterprise/Server?fid=2260\",\"Name\":\"人工智慧平台\"},{\"Id\":0,\"Url\":\"/tw/Enterprise/Server?fid=2363\",\"Name\":\"人工智慧訓練伺服器\"},{\"Id\":0,\"Url\":\"/tw/Enterprise/Server?fid=2364\",\"Name\":\"人工智慧推論伺服器\"},{\"Id\":0,\"Url\":\"/tw/Enterprise/Server?fid=2262\",\"Name\":\"高效能運算伺服器\"}],\"IsQuickShip\":false,\"QuickShipUrl\":null,\"TitleForScoreWeight\":null},{\"Id\":7242,\"TableKey\":10083,\"Name\":\"G893-ZD1-AAX5\",\"Link\":\"/tw/Enterprise/GPU-Server/G893-ZD1-AAX5\",\"Title\":\"\",\"Description\":\"HPC/AI Server - AMD EPYC™ 9005/9004 - 8U DP NVIDIA HGX™ B200 8-GPU\",\"MediaInfo\":{\"Id\":7242,\"Description\":\"\",\"MediaPath\":\"//static.gigabyte.com/StaticFile/Image/Global/648f5ea9e92ae445279d5eca853dbddd/Product/43273\",\"MediaPathYoutube\":\"\",\"MediaPathVideo\":\"\",\"MediaPathExt\":\"\",\"MediaPathWebP\":\"\",\"MediaAlt\":\"\",\"MediaType\":0,\"SeoAlt\":\"\",\"SeoTitle\":\"\",\"AngularSeoAlt\":\"\",\"AngularSeoTitle\":\"\"},\"IsAiAuto\":false,\"ApplicationItem\":[{\"Id\":0,\"Url\":\"/tw/Enterprise/Server?fid=2260\",\"Name\":\"人工智慧平台\"},{\"Id\":0,\"Url\":\"/tw/Enterprise/Server?fid=2363\",\"Name\":\"人工智慧訓練伺服器\"},{\"Id\":0,\"Url\":\"/tw/Enterprise/Server?fid=2364\",\"Name\":\"人工智慧推論伺服器\"},{\"Id\":0,\"Url\":\"/tw/Enterprise/Server?fid=2262\",\"Name\":\"高效能運算伺服器\"}],\"IsQuickShip\":false,\"QuickShipUrl\":null,\"TitleForScoreWeight\":null}]"}],"Name":"products","GaCodeCategory":null,"IsH1Title":false},{"Id":2346,"HashNumber":0,"Title":"","Description":"","IsShowGigabyteLogIcon":false,"BackGroundInfo":{"Id":2346,"Type":"None","Value":"","FontStyle":"Dark","CssStyleString":"","CssClassNameString":""},"ChildList":[{"Id":0,"ModelType":"PartnerIcon","Title":"","Description":"","Content":"[{\"Link\":\"https://www.nvidia.com/en-us/\",\"MediaInfo\":{\"Id\":7219,\"Description\":\"\",\"MediaPath\":\"//static.gigabyte.com/StaticFile/Image/Global/40505512b7117629a096580e60df226d/Feature/6335\",\"MediaPathYoutube\":\"\",\"MediaPathVideo\":\"\",\"MediaPathExt\":\"\",\"MediaPathWebP\":\"\",\"MediaAlt\":\"NVIDIA\",\"MediaType\":0,\"SeoAlt\":\"alt=\\\"NVIDIA\\\"\",\"SeoTitle\":\"title=\\\"NVIDIA\\\"\",\"AngularSeoAlt\":\"ng-alt=\\\"NVIDIA\\\"\",\"AngularSeoTitle\":\"ng-title=\\\"NVIDIA\\\"\"}}]"}],"Name":"partners","GaCodeCategory":null,"IsH1Title":false},{"Id":2349,"HashNumber":0,"Title":"了解更多AI秘辛","Description":"","IsShowGigabyteLogIcon":false,"BackGroundInfo":{"Id":2349,"Type":"None","Value":"","FontStyle":"Dark","CssStyleString":"","CssClassNameString":""},"ChildList":[{"Id":0,"ModelType":"Resource","Title":"","Description":"","Content":"[{\"Id\":7269,\"TypeName\":\"News\",\"TypeLink\":\"/tw/Press/News\",\"MediaInfo\":{\"Id\":7269,\"Description\":\"\",\"MediaPath\":\"//static.gigabyte.com/StaticFile/Image/tw/4831c4b6c9889c841fc63ab228e3fffb/NewsThumbnail/2244\",\"MediaPathYoutube\":\"\",\"MediaPathVideo\":\"\",\"MediaPathExt\":\"\",\"MediaPathWebP\":\"\",\"MediaAlt\":\"\",\"MediaType\":0,\"SeoAlt\":\"\",\"SeoTitle\":\"\",\"AngularSeoAlt\":\"\",\"AngularSeoTitle\":\"\"},\"Title\":\"NVIDIA Blackwell 產品線再擴展\\nGIGABYTE 推出全新 NVIDIA HGX™ B200 平台伺服器\",\"Description\":\"GIGABYTE NVIDIA Blackwell 架構解決方案助力百億億次級運算發展\",\"ButtonName\":\"Start To Read\",\"Link\":\"/tw/Press/News/2244\",\"TableKey\":\"2244\",\"IsAiAuto\":false,\"TitleForScoreWeight\":null},{\"Id\":7243,\"TypeName\":\"Solution\",\"TypeLink\":\"/tw/Solutions\",\"MediaInfo\":{\"Id\":7243,\"Description\":\"\",\"MediaPath\":\"//static.gigabyte.com/StaticFile/Image/Global/40505512b7117629a096580e60df226d/SolutionMain/233\",\"MediaPathYoutube\":\"\",\"MediaPathVideo\":\"\",\"MediaPathExt\":\"\",\"MediaPathWebP\":\"\",\"MediaAlt\":\"\",\"MediaType\":0,\"SeoAlt\":\"\",\"SeoTitle\":\"\",\"AngularSeoAlt\":\"\",\"AngularSeoTitle\":\"\"},\"Title\":\"GIGAPOD - AI 資料中心解決方案\",\"Description\":\"AI伺服器機櫃打造強大GPU運算叢集,GIGAPOD將AI超級運算再進化。\",\"ButtonName\":\"Start To Read\",\"Link\":\"/tw/Industry-Solutions/giga-pod-as-a-service\",\"TableKey\":\"233\",\"IsAiAuto\":false,\"TitleForScoreWeight\":null},{\"Id\":7244,\"TypeName\":\"Insight\",\"TypeLink\":\"/tw/Article\",\"MediaInfo\":{\"Id\":7244,\"Description\":\"\",\"MediaPath\":\"//static.gigabyte.com/StaticFile/Image/Global/110b8f1e732061b768a82694295a940b/Article/225\",\"MediaPathYoutube\":\"\",\"MediaPathVideo\":\"\",\"MediaPathExt\":\"\",\"MediaPathWebP\":\"\",\"MediaAlt\":\"\",\"MediaType\":0,\"SeoAlt\":\"\",\"SeoTitle\":\"\",\"AngularSeoAlt\":\"\",\"AngularSeoTitle\":\"\"},\"Title\":\"如何透過GIGAPOD一站式服務,加速AI時代的全面革新\",\"Description\":\"\",\"ButtonName\":\"Start To Read\",\"Link\":\"/tw/Article/how-gigapod-provides-a-one-stop-service-accelerating-a-comprehensive-ai-revolution\",\"TableKey\":\"225\",\"IsAiAuto\":false,\"TitleForScoreWeight\":null},{\"Id\":7245,\"TypeName\":\"Video\",\"TypeLink\":\"https://www.youtube.com/user/GIGABYTEchannel\",\"MediaInfo\":{\"Id\":7245,\"Description\":\"\",\"MediaPath\":\"//static.gigabyte.com/StaticFile/Image/Global/6766558b20c73763b15651ffd0aa8964/VideoThumbnail/1157\",\"MediaPathYoutube\":\"https://www.youtube.com/embed/vQeOXnNwDd0\",\"MediaPathVideo\":\"\",\"MediaPathExt\":\"\",\"MediaPathWebP\":\"\",\"MediaAlt\":\"\",\"MediaType\":1,\"SeoAlt\":\"\",\"SeoTitle\":\"\",\"AngularSeoAlt\":\"\",\"AngularSeoTitle\":\"\"},\"Title\":\"GIGAPOD, the Turnkey AI Supercomputing Solution\",\"Description\":\"GIGAPOD is so much more than just an AI computing product. It is an AI ecosystem platform that excels in parallel processing, serving as a one-stop solution for data centers transitioning to AI factories that run deep learning models at scale. 💪🦾\\nThis is also our rack-as-a-service where we are approached to build an entire data center, to accommodate tens or hundreds of racks with top-tier computer performance and cooling. \\nLearn more: https://gbte.tech/2y3N\\nDesign a data center or supercomputing cluster with GIGAPOD: https://gbte.tech/inquiry \\n\\n#GIGABYTE #UpgradeYourLife \\n#GIGABYTEgroup #GigaComputing #GIGABYTEServer #GIGABYTEai #serversolutions #AIserver #AI #GenerativeAI #LLM #cluster #supercomputing #GIGAPOD \\n\\n--\\nFind us on social media\\nhttps://www.linkedin.com/company/gigabyte/\\nhttps://www.facebook.com/GIGABYTEofficial\\nhttps://www.instagram.com/gigabyte_official\\n\\nAnd visit our website for more info.\\nhttps://www.gigabyte.com/\",\"ButtonName\":\"Start To Play\",\"Link\":\"\",\"TableKey\":\"1157\",\"IsAiAuto\":false,\"TitleForScoreWeight\":null},{\"Id\":7247,\"TypeName\":\"Video\",\"TypeLink\":\"https://www.youtube.com/user/GIGABYTEchannel\",\"MediaInfo\":{\"Id\":7247,\"Description\":\"\",\"MediaPath\":\"//static.gigabyte.com/StaticFile/Image/Global/1a2ae3930f6119559d54ffa4dad72a4a/VideoThumbnail/1071\",\"MediaPathYoutube\":\"https://www.youtube.com/embed/2Q7S-CbnAAY\",\"MediaPathVideo\":\"\",\"MediaPathExt\":\"\",\"MediaPathWebP\":\"\",\"MediaAlt\":\"\",\"MediaType\":1,\"SeoAlt\":\"\",\"SeoTitle\":\"\",\"AngularSeoAlt\":\"\",\"AngularSeoTitle\":\"\"},\"Title\":\"GIGAPOD: The Future of AI Computing in Data Centers\",\"Description\":\"The \\\"GIGAPOD\\\" scales seamlessly from G593 HGX H100 GPU servers to full racks, achieving this expansion by carefully ...\",\"ButtonName\":\"Start To Play\",\"Link\":\"\",\"TableKey\":\"1071\",\"IsAiAuto\":false,\"TitleForScoreWeight\":null},{\"Id\":7246,\"TypeName\":\"Solution\",\"TypeLink\":\"/tw/Solutions\",\"MediaInfo\":{\"Id\":7246,\"Description\":\"\",\"MediaPath\":\"//static.gigabyte.com/StaticFile/Image/Global/40505512b7117629a096580e60df226d/SolutionMain/234\",\"MediaPathYoutube\":\"\",\"MediaPathVideo\":\"\",\"MediaPathExt\":\"\",\"MediaPathWebP\":\"\",\"MediaAlt\":\"\",\"MediaType\":0,\"SeoAlt\":\"\",\"SeoTitle\":\"\",\"AngularSeoAlt\":\"\",\"AngularSeoTitle\":\"\"},\"Title\":\"NVIDIA Grace™ CPU 超級晶片和 GH200 Grace Hopper 超級晶片\",\"Description\":\"推動資料中心的人工智慧與高效能運算──突破性的CPU超級晶片設計\",\"ButtonName\":\"Start To Read\",\"Link\":\"/tw/Industry-Solutions/nvidia-superchip\",\"TableKey\":\"234\",\"IsAiAuto\":false,\"TitleForScoreWeight\":null}]"}],"Name":"resources","GaCodeCategory":null,"IsH1Title":false}],"ContactUsInfoChild":{"BaseInfo":{"ShowContactInfo":false,"ContactUsCountryList":["Afghanistan","Albania","Algeria","Andorra","Antigua","Argentina","Armenia","Australia","Austria","Azerbaijan","Bahamas","Bahrain","Bangladesh","Barbados","Belarus","Belgium","Belize","Bermuda","Bhutan","Bolivia","Bosnia & Herzegovina","Brazil","Brunei","Bulgaria","Burkina Faso","Cambodia","Canada","Canada","Central African Republic","Chile","China","Colombia","Costa Rica","Croatia","Cuba","Cyprus","Czech","Czech Republic","Denmark","Dominica","Dominican Republic","Dubai","East Timor","Ecuador","Egypt","El Salvador","Estonia","European Union","Fiji","Finland","France","Georgia","Germany","Greece","Grenada","Guatemala","Guinea","Guyana","Haiti","Honduras","Hong Kong","Hungary","Iceland","India","Indonesia","Iraq","Ireland","Israel","Italy","Jamaica","Japan","Jordan","Kazakhstan","Kiribati","Korea","Kuwait","Kyrgyzstan","Laos","Latin America","Latvia","Lebanon","Libya","Liechtenstein","Lithuania","Luxembourg","Macedonia","Madagascar","Malaysia","Maldives","Malta","Mauritius","Mexico","Middle East","Moldova","Monaco","Mongolia","Montenegro","Morocco","Myanmar (Burma)","Nauru","Nepal","Netherlands","New Zealand","Nicaragua","Nigeria","North America ","Norway","Oman","Pakistan","Palestine","Panama","Papua New Guinea","Paraguay","Peru","Philippines","Poland","Portugal","Puerto Rico","Qatar","Romania","Russia","Saint Lucia","San Marino","Saudi Arabia","Serbia","Singapore","Slovak Republic","Slovenia","South Africa","Southeast Asia","Spain","Sri Lanka","Sudan","Sweden","Switzerland","Syria","Taiwan","Tajikistan","Thailand","Trinidad and Tobago","Tunisia","Turkey","Turkmenistan","U.A.E.","U.K.","U.S.A.","Ukraine","Uruguay","Uzbekistan","Venezuela","Vietnam","Yemen"],"Email":"","Id":""},"ClassNameButton":[],"Title":"","SalesFromButtonShowText":"業務洽詢","GaCodeCategory":"","IsHaveNewsletterButton":false,"InterestItemWithEmail":[],"IsShowB2bSnsText":false,"IsHaveCover":true},"ModelTypeMenuHaveEmailSales":{"ModuleMenuList":[{"Id":2342,"KeyName":"advantage","ShowContent":"Advantage"},{"Id":2344,"KeyName":"applications","ShowContent":"Applications"},{"Id":2345,"KeyName":"products","ShowContent":"Products"},{"Id":2346,"KeyName":"partners","ShowContent":"Partners"},{"Id":2349,"KeyName":"resources","ShowContent":"Resources"}],"IsSetCenter":true,"IsHaveEmailSales":false,"ShowEmailSalesText":"Email Sales"}}; //異動須同步確認 MiniSite 內容頁 window.RazorDataProductInfo = { QuickShippingText: '快速達交產品', IsShowQuickShipping: true, ItemListInfo: [{"Id":0,"ModelType":"Product","Title":"","Description":"","Content":"[{\"Id\":7241,\"TableKey\":10082,\"Name\":\"G893-SD1-AAX5\",\"Link\":\"/tw/Enterprise/GPU-Server/G893-SD1-AAX5\",\"Title\":\"\",\"Description\":\"HPC/AI Server - 5th/4th Gen Intel® Xeon® Scalable - 8U DP NVIDIA HGX™ B200 8-GPU\",\"MediaInfo\":{\"Id\":7241,\"Description\":\"\",\"MediaPath\":\"//static.gigabyte.com/StaticFile/Image/Global/41889d347cf472a51bc922505a230efd/Product/43327\",\"MediaPathYoutube\":\"\",\"MediaPathVideo\":\"\",\"MediaPathExt\":\"\",\"MediaPathWebP\":\"\",\"MediaAlt\":\"\",\"MediaType\":0,\"SeoAlt\":\"\",\"SeoTitle\":\"\",\"AngularSeoAlt\":\"\",\"AngularSeoTitle\":\"\"},\"IsAiAuto\":false,\"ApplicationItem\":[{\"Id\":0,\"Url\":\"/tw/Enterprise/Server?fid=2260\",\"Name\":\"人工智慧平台\"},{\"Id\":0,\"Url\":\"/tw/Enterprise/Server?fid=2363\",\"Name\":\"人工智慧訓練伺服器\"},{\"Id\":0,\"Url\":\"/tw/Enterprise/Server?fid=2364\",\"Name\":\"人工智慧推論伺服器\"},{\"Id\":0,\"Url\":\"/tw/Enterprise/Server?fid=2262\",\"Name\":\"高效能運算伺服器\"}],\"IsQuickShip\":false,\"QuickShipUrl\":null,\"TitleForScoreWeight\":null},{\"Id\":7242,\"TableKey\":10083,\"Name\":\"G893-ZD1-AAX5\",\"Link\":\"/tw/Enterprise/GPU-Server/G893-ZD1-AAX5\",\"Title\":\"\",\"Description\":\"HPC/AI Server - AMD EPYC™ 9005/9004 - 8U DP NVIDIA HGX™ B200 8-GPU\",\"MediaInfo\":{\"Id\":7242,\"Description\":\"\",\"MediaPath\":\"//static.gigabyte.com/StaticFile/Image/Global/648f5ea9e92ae445279d5eca853dbddd/Product/43273\",\"MediaPathYoutube\":\"\",\"MediaPathVideo\":\"\",\"MediaPathExt\":\"\",\"MediaPathWebP\":\"\",\"MediaAlt\":\"\",\"MediaType\":0,\"SeoAlt\":\"\",\"SeoTitle\":\"\",\"AngularSeoAlt\":\"\",\"AngularSeoTitle\":\"\"},\"IsAiAuto\":false,\"ApplicationItem\":[{\"Id\":0,\"Url\":\"/tw/Enterprise/Server?fid=2260\",\"Name\":\"人工智慧平台\"},{\"Id\":0,\"Url\":\"/tw/Enterprise/Server?fid=2363\",\"Name\":\"人工智慧訓練伺服器\"},{\"Id\":0,\"Url\":\"/tw/Enterprise/Server?fid=2364\",\"Name\":\"人工智慧推論伺服器\"},{\"Id\":0,\"Url\":\"/tw/Enterprise/Server?fid=2262\",\"Name\":\"高效能運算伺服器\"}],\"IsQuickShip\":false,\"QuickShipUrl\":null,\"TitleForScoreWeight\":null}]"}], } $(function() { if (!window.gbtGlobal.Razor.IsProduction) { console.log(window.dataJsonAll); } window.isCheckOldIEVersionRedirect = true; TriggerGoTopAndHide(undefined, true); //setSnsButtonHoverClick(); window.backgroundMediaDiv = 'backgroundMedia'; //for Background setBackgroundCenterRwd($("." + window.backgroundMediaDiv)); $(window).resize(function() { setBackgroundCenterRwd($("." + window.backgroundMediaDiv)); checkContentFlowType(); }); setSpeakerImgOnclick(); SetGIGABYTECiAnimate('section-tab-forum'); $(document).ajaxSuccess(function () { transferGigabyteTextToUpperCase(); }); transferGigabyteTextToUpperCase(); }); var SetGIGABYTECiAnimate = function(parentClassName) { window.CiAnimateRun = {}; window.CiAnimateStyle = {}; var ciImageDom = $(""); var ciTextDom = $('.' + parentClassName).find('.title-text'); ciTextDom.each(function() { var thisTextDom = $(this); var isSectionModelDarkStyle = (thisTextDom.parents('.sectionModelDarkStyle').length > 0) ? true : false; if (thisTextDom.text().indexOf('GIGABYTE') >= 0) { thisTextDom.html(thisTextDom.text() .replace("GIGABYTE", ' ' + ciImageDom.prop('outerHTML') + ' ')); window.CiAnimateRun[thisTextDom.text()] = false; window.CiAnimateStyle[thisTextDom.text()] = (isSectionModelDarkStyle === true) ? "ci-wt" : "ci-bk"; } }); $(window).scroll(function() { var screenHeight = $(window).height(); //視窗高度 var scrollTop = $(window).scrollTop(); //scroll bar 移動高度 var visibleOffsetTop = screenHeight + scrollTop; //目前物件可視位置 var visibleDeltaTop = 0; ciTextDom.find('.ci-animate').each(function() { var thisImgDom = $(this); var thisTextDom = thisImgDom.parent(); var thisTextDomOffset = thisTextDom.offset(); var thisTextDomVisibleRange = thisTextDomOffset.top + screenHeight; //console.log(visibleOffsetTop - thisTextDomOffset.top, thisTextDom.text(), window.CiAnimateRun[thisTextDom.text()]); var blnTextDomVisible = (visibleOffsetTop - thisTextDomOffset.top) >= visibleDeltaTop; blnTextDomVisible = blnTextDomVisible && (visibleOffsetTop >= thisTextDomOffset.top) && (visibleOffsetTop < thisTextDomVisibleRange); if (blnTextDomVisible) { if (window.CiAnimateRun[thisTextDom.text()] === false) { setTimeout(function() { thisImgDom.attr('data', "/Images/" + window.CiAnimateStyle[thisTextDom.text()] + ".svg?" + Math.random()); }, 350); window.CiAnimateRun[thisTextDom.text()] = true; } } else { if (window.CiAnimateRun[thisTextDom.text()] === true) { $(this).attr('data', "/Images/ci-g.svg"); window.CiAnimateRun[thisTextDom.text()] = false; } } }); }); };

            引領 AI 新時代:技嘉 NVIDIA Blackwell 架構解決方案

            探索技嘉針對 NVIDIA Blackwell 架構優化的解決方案,提供無與倫比的擴展性、靈活性與效能,專為加速 AI、高效能運算 (HPC) 以及 GPU 驅動的工作負載而設計,打破加速運算和生成式人工智慧面臨的阻礙,為生成式人工智慧開創全新篇章。

            GPU加速時代

            專用圖形元件的發展歷程始於 1970 年代,並最終催生了GPU 這一顛覆性的創新。經過半個世紀的演進,GPU 已從簡單的電路發展為獨立的板卡,進一步演化為嵌入於高密度計算基礎架構中的模組。隨著生成式人工智慧和大型訓練模型的應用日益普及,GPU強大的算力在各產業中扮演著不可或缺的角色,也進一步推動了AI工廠的廣泛部署與運營。

            如今,加速器引領各式產業與人們工作、日常生活革新:從修復影片至提升解析度,到氣候地理模擬預測,再到驅動 ChatGPT 生成式 AI 模型的大規模使用。隨著對 GPU 計算資源需求的指數級增長,技嘉與產業領導者攜手合作,實現超越摩爾定律的效能提升。作為 GPU創新領導者,NVIDIA 一直致力於突破極限,現在更推出新世代 Blackwell 架構,推動產業革新進入全面生產階段。

            突破加速運算與生成式 AI 的界限

            在 NVIDIA Hopper 架構取得巨大成功的基礎上,Blackwell 架構更進一步,為因應AI 模型日益成長的複雜度和參數量而設計。採用台積電 4 奈米製程技術 (4NP),Blackwell GPU 集成了數十億個晶體管,並導入了更快速、更高頻寬的 NVIDIA® NVLink™ 和第二代 Transformer Engine 等創新技術。提供了比起前一代倍數等級的效能提升,讓 Blackwell 為生成式人工智慧開創全新篇章。

            NVIDIA Blackwell ArchitectureBlackwell 架構 GPU

            包含2080億個電晶體,採用台積電 4NP 製程製造

            第二代 Transformer Engine

            實現4位元浮點 (FP4) AI,效能加倍

            第五代 NVLink 和 NVLink Switch

            1.8 TB/s GPU-GPU 互連頻寬

            可靠性、可用性、可維護性 (RAS) 引擎

            100% 系統內自我檢測

            安全的人工智慧

            高效能加密與TEE

            解壓縮引擎

            800 GB/s

            技嘉致力於提供靈活且可擴展的解決方案

            Short TTM短上市週期以實現敏捷部署

            技嘉與產業領導廠商密切合作,快速推出新一代產品,全面因應快速成長的運算需求。技嘉洞悉產業趨勢與用戶需求,憑藉對企業級產品領域的豐富經驗,透過模組化設計降低成本;簡化設計流程並實現彈性客製的可能性,旨在讓用戶即時採用最新一代的技術與效能優勢。

            Flexible Scalability靈活擴充,適應多元場景

            為滿足用戶對運算需求與日俱增的基礎建設升級需求,技嘉伺服器產品配備充足擴充插槽,強化節點間連結效能;特別在GPU應用方面,確保巨量資料能在GPU之間實現高效傳輸。

            完整的一站式服務

            隨著計算架構規模擴展,且應用日益朝向專業領域發展,最佳的系統配置對於實現高效能與高效率至關重要。技嘉提供完整的一站式服務協助用戶打造AI工廠並協助快速啟動維運,服務範圍涵蓋專業諮詢、量身訂製的環境架構規劃、資料中心佈署到認證驗收以及後續服務。透過技嘉擁有的 Know-How,確保用戶獲得完全符合其營運要求和績效目標的端到端解決方案。

            欲了解更多技嘉的一站式解決方案:GIGAPOD – AI 資料中心解決方案

            NVIDIA GB200 NVL72:機架級設計的巔峰之作

            NVIDIA GB200 NVL7

            作為 Blackwell 架構系列的旗艦產品,NVIDIA GB200 NVL72 連接 36 個基於ARM架構的 Grace CPU 和 72 個 Blackwell GPU 並採用全液冷設計。這一機架級配置透過最新的 NVLink 技術將所有節點相互串聯,實現「一個大 GPU」的效能。

            此 NVIDIA GB200 NVL72單一機架的百萬兆級電腦,在大型語言模型推論效能上超越上一代 HGX H100 GPU 30 倍,在訓練效能上提升 4 倍,同時實現了降低25 倍能源消耗。憑藉無與倫比的互聯速度和能源效率,GB200 NVL72 成為 AI 和 HPC 工作負載的新基準。

            XN14-CB0-LA01 運算節點

            • 2 x GB200 Grace Blackwell 超級晶片
            • 2 x 384 GB HBM3e GPU 記憶體與 16TB/s 頻寬
            • 2 x 480 GB LPDDR5X CPU 記憶體與 152GB/s 頻寬
            • 8 x E1.S Gen5 NVMe 硬碟

            NVIDIA HGX™ B200:全球頂尖的人工智慧運算平台

            NVIDIA HGX™ B200 採用先進的 Blackwell 架構。透過模組化的 NVIDIA NVSwitch™ 設計,與上一代相比,它在運算效能和能源效率上實現了超過10倍的提升。

            為應對前所未有的效能提升與熱功耗所衍生的散熱問題,技嘉推出兩種形式的伺服器選擇,包括 8U – 空氣冷卻和 4U – 液冷選項,確保相容於現有基礎架構。技嘉的高效能運算伺服器旨在提升效能與提供高可用性,包括:
            1. 全面支援全高擴展卡,能容納多組 DPU 和 SuperNIC。
            2. 後抽式 PCIe 擴充槽固定架與前抽式主機板 / GPU 托盤設計,簡化維護流程。
            3. 可熱插拔、完全冗餘的電源供應器,並提供多種電源連接埠選項,增加靈活性。

            其他功能,如支援 RDMA 的 PCIe 交換器設計,可實現 GPU 與 GPU 間的直接資料傳輸,確保技嘉伺服器在進行大規模加速運算負載時,擁有高度可靠的效能與效率。
            Flexible Scalability

            NVIDIA HGX™ B200 8-GPU

            • 8 x NVIDIA Blackwell GPUs
            • 1,440GB of GPU 記憶體
            • 72 petaFLOPS 訓練效能
            • 144 petaFLOPS 推論效能
            • 1,800GB/s GPU-to-GPU 頻寬與 NVIDIA® NVLink™ 及 NVSwitch™
            Flexible Scalability

            G893 8U HPC/AI 伺服器

            • NVIDIA HGX™ B200 8-GPU
            • 雙 5th/4th Gen Intel® Xeon® 可擴展或 雙 AMD EPYC™ 9005/9004 系列處理器
            • 2 x 10Gb/s LAN
            • 8 x 2.5" Gen5 NVMe 熱插拔硬碟槽
            • 12 x 全高半長 PCIe Gen5 x16 插槽
            • 12 x 3000W 80 PLUS 鈦金冗餘電源供應器
            Flexible Scalability

            G4L3 4U HPC/AI 伺服器

            • 液冷 NVIDIA HGX™ B200 8-GPU
            • 雙 5th/4th Gen Intel® Xeon® 可擴展或 雙 AMD EPYC™ 9005/9004 系列處理器
            • 2 x 10Gb/s LAN
            • 8 x 2.5" Gen5 NVMe 熱插拔硬碟槽
            • 12 x 全高半長 PCIe Gen5 x16 插槽
            • 8 x 3000W 80 PLUS 鈦金冗餘電源供應器

            技嘉AI伺服器