關於 cookie 的說明

本網站使用瀏覽器紀錄 (Cookies) 來提供您最好的使用體驗,我們使用的 Cookie 也包括了第三方 Cookie。相關資訊請訪問我們的隱私權與 Cookie 政策。如果您選擇繼續瀏覽或關閉這個提示,便表示您已接受我們的網站使用條款。

搜尋結果Search Result

符合「GPU」新聞搜尋結果, 共 974 篇 ,以下為 1 - 24 篇 訂閱此列表,掌握最新動態
邊緣運算卓越之選,結合Intel Arc GPU之研揚多功能工業邊緣AI電腦MXM-ACMA-PUC

【臺北訊】專業物聯網及人工智慧邊緣運算平臺研發製造大廠—研揚科技(股票代碼: 6579),日前推出全新的MXM-ACMA-PUC系統產品。這款產品搭載第 13 代 英特爾Core™ 處理器與嵌入式 Intel® Arc™ GPU。融合了 CPU、GPU 和周邊設備支援,為智慧城市、自動光學檢測 (AOI) 和邊緣工作站解決方案提供最佳動能。研揚並非第一次推出Intel Arc系列GPU的產品,但是這是首款搭載這個GPU的系統,是款絕佳邊緣AI解決方案。 MXM-ACMA-PUC搭載第13代Intel® Core™處理器,最高功耗為35W,並預設搭載 16核心、24 執行緒的Intel® Core™ i7-13700TE 處理器。為了提升機器學習與 AI工作負載的管理效能,系統內建 MXM-ACMA 模組,可選配嵌入式 Intel® Arc™ A370E 或 Intel® Arc™ A350E GPU,非常適合提升AI工作負載的運算能力。 在介面方面,MXM-ACMA-PUC 配備了四個 LAN 連接埠,其中三個支援 2.5GbE 速度,一個支援 1GbE,此外還提供兩個 USB 3.2 Gen 2 埠和兩個具備 RS-232/422/485功能的COM連接埠。主板上提供一個 HDMI 2.0 顯示輸出,支援 3840 x 2160 解析度,刷新率為 60Hz,而內建的 GPU 則提供四個 DP++ 1.4 顯示選項,支援 1920 x 1080 解析度。 在擴展選項方面,MXM-ACMA-PUC提供多樣化選擇,包括 M.2 2230 E-Key、M.2 3052 B-Key 和 M.2 2280 M-Key 插槽,分別支援 Wi-Fi、5G 和 NVMe 儲存。這些功能,再加上其規格,MXM-ACMA-PUC的目標應用領域包括需要結合高 AI 性能與節能運行的應用領域,例如: 智慧城市和 AI 輔助自動光學檢測(AOI)。此外,該系統還可作為工業工作站的閘道設備。藉由高速介面配置、CPU 和 GPU 性能,以及高速 NVMe 儲存支援,MXM-ACMA-PUC可同時為四個邊緣工作站提供動力。 「MXM-ACMA-PUC提供高性能計算及反應能力,同時減少了延遲,特別適合機器視覺及邊緣AI運算的應用。也為了因應這類的高速運算需求,這款系統的機身兩側、上蓋和前後蓋都有大量的散熱孔設計,內部也搭載主動散熱器。讓高速運轉下所產生的熱能可以快速散熱。」研揚科技嵌入式板卡處產品經理黃怡臻表示。 更多MXM-ACMA-PUC的產品資訊,您可以上研揚官網查詢,也可以於官網訂購 ,或與研揚科技國內業務處02-89191234分機1142劉小姐聯繫 關於研揚科技 研揚科技集團(AAEON)是台灣專業物聯智能解決方案研發製造大廠,成立於1992年。研發製造並行銷全球IoT及AI邊緣運算解決方案,另有嵌入式電腦主板及系統、工業液晶顯示器、強固型平板電腦、工控機、網路安全設備以及相關配件等,提供OEM/ODM客戶及系統整合商完整且專業之軟硬體解決方案。同時,研揚科技有專屬團隊提供客製化服務,協助您從研發初期發想到產品製作、量產到售後服務,提供一貫之專業諮詢與服務,為您量身打造高品質產品。研揚科技目前提供多款AI邊緣運算產品及智慧城市、智慧零售及智慧製造等系統整合和解決方案。研揚是英特爾鈦金級會員,同時也是NVIDIA的菁英級夥伴(Elite partner)。欲瞭解更多詳細資訊請參考研揚科技官方網站

文章來源 : AAEON 發表時間 : 瀏覽次數 : 5863 加入收藏 :
HKBNES x Suanova Technology Launch Large-scale AI Computing Resource Platform Powered by METAX GPU Stacks for Enterprise Transformation

HONG KONG, Nov. 26, 2024 /PRNewswire/ -- HKBN Enterprise Solutions (HKBNES), in its ongoing commitment to enrich the local technology landscape, has teamed up with Suanova Technology to become its first distributor of computing power services in Hong Kong. This partnership will introduce a more diversified array of cloud services and provide access to artificial intelligence (AI) computing resources powered by METAX's GPU (Graphics Processing Unit) stacks, empowering enterprise clients to transform and fully maximize their digital capabilities. Amid its burgeoning emergence, AI computing has evolved into an essential competitive asset that every enterprise must harness. According to market research firm IDC[1], AI has officially become a central component of global corporate strategies. Estimates suggest that by 2028, global spending on AI technology will reach $749 billion (approximately HKD 5.83 trillion), a significant increase from the expected $227 billion (approximately HKD 1.76 trillion) in 2025. As AI computing rapidly advances, the demand for related resources is on the rise. A leading provider of compute services, tech development and industrial investment, Suanova Technology operates the "Fengshou No. 1" Intelligent Computing Center in Shanghai and the "Fengshou No. 2" Intelligent Computing Center in Hong Kong, both capable of delivering up to 1,000 PetaFLOPs of computing power. By utilizing its cutting-edge Fengshou series powered by METAX GPUs, Suanova equips enterprise clients with autonomous computing resource scheduling platforms to address a diverse range of application scenarios, including fulfilling the computational requirements for training large language models (LLM).  Dr Denis Yip, HKBN President and Group Chief Operating Officer, said: "As a pioneering force and Hong Kong's first telco to launch 25Gbps broadband service, HKBN has consistently pushed the boundaries to supercharge digital transformation and innovation for enterprises. Our collaboration with Suanova Technology brings powerful AI computing resources to Hong Kong, providing enterprises with breakthrough compute speed and flexibility. This not only meets the local market's pressing need for high-performance computing power but it also opens new avenues for innovative applications across industries, ushering in a new era of opportunities." Gu Meng, President of Suanova Technology, said: "Our goal is to deliver professional, autonomous, and innovative services for enterprises in mainland China, the Greater Bay Area, and Hong Kong, fostering growth and facilitating smart transformations for a shared future. Our collaboration with HKBN Enterprise Solutions will significantly shorten the distance for enterprises to access AI computing resources, making cutting-edge technology services readily available. We expect this partnership to set a new standard across the industry, driving many more enterprises to expedite their digitalization journeys, and unlock limitless possibilities." Image #1: (From the left) Dr Denis Yip, President and COO of HKBN Group, Samuel Hui, HKBN Co-Owner and Chief Strategy Officer, Enterprise Solutions, Gu Meng, CEO of Suanova Technology and Douglas Fangm, Chairman of Yeebo (International Holdings) Limited.   [1] IDC Unveils 2025 FutureScapes: Worldwide IT Industry Predictions. Retrieved from https://www.idc.com/getdoc.jsp?containerId=prUS52691924 About HKBN Group Headquartered in Hong Kong with operations spanning across Hong Kong, Macau and mainland China, HKBN Group ("the Group" or "HKBN") is a leading integrated telecommunications and technology services provider. The Group's extensive tri-carrier fibre infrastructure covers around 2.6 million residential homes and 8,200 commercial buildings and facilities across Hong Kong, offering comprehensive one-stop Information and Communications Technology ("ICT") solutions and Infinite-play bundles to both individual and enterprise customers. Committed to creating a lasting positive impact to wherever it operates, HKBN embraces a core purpose to "Make our Home a Better Place to Live" and has received a highest possible rating of AAA in MSCI's 2024 ESG Ratings assessment in environment, society and governance. The Group is managed by hundreds of Co-Owners (supervisory and management level Talents in the Group) who invested their savings to buy shares of HKBN Ltd. (SEHK stock code: 1310). For more information about HKBN, please visit https://www.hkbn.net/group/. About HKBN Enterprise Solutions  HKBN Enterprise Solutions ("HKBNES") is the arm of HKBN. As a leading system integration ICT services provider, it focuses on enterprise solutions development. Comprises a professional team and profound experience, together with HKBN's unique tri-carrier network, HKBNES provides enterprises with one-stop comprehensive digital transformation solutions, including cloud and data centres, cyber security, smart and digital solutions, and more. For more information about HKBNES, please visit  www.hkbnes.com/http://www.hkbnes.net/web/en/.  About Suanova Technology Suanova Technology Limited (hereinafter referred to as "Suanova") is a wholly-owned subsidiary of Hong Kong listed company Yeebo (International Holdings) Limited. Its business covers three major business sectors, including: computing power and cloud operations, computing power technology development and computing power industry investment. As a domestic cloud computing pioneer, Suanova aims to provide enterprises with professional, independent and innovative one-stop services. Our products mainly include: cloud, AI computing power, large devices, system integration, and GPU chips. Suanova has set up branches in Hong Kong and Shanghai to provide customers with better localized services. Our vision is to help companies grow, support intelligent transformation, and create a better future.

文章來源 : PR Newswire 美通社 發表時間 : 瀏覽次數 : 160 加入收藏 :
神雲科技整合最新CPU與GPU技術,SC24展示AI/HPC伺服器新品

亞特蘭大2024年11月20日 /美通社/ -- 業界一流的伺服器設計暨製造商,神達控股股份有限公司(股票代號:3706)子公司神雲科技股份有限公司(MiTAC Computing Technology Corp.),今日於美國亞特蘭大 SC24 展會自家攤位#2543展示了一系列全新伺服器。這些伺服器整合了最新的 AMD EPYC™ 9005 系列處理器、AMD Instinct™ MI325X 加速器、Intel® Xeon® 6處理器與專業級GPU,為 HPC 和 AI 工作負載提供卓越的優化效能。 神雲科技整合最新CPU與GPU技術,SC24展示AI/HPC伺服器新品 為 AI 驅動的資料中心工作負載提供領先效能與密度 搭載 AMD EPYC 9005 系列處理器,神雲科技新一代伺服器專為高效能 AI 運算工作負載進行優化設計,提升資料中心的效能與運算密度。在這次的 SC24 大會,神雲科技重點展示了兩款優秀的 AI/HPC 伺服器產品:首先是MiTAC G8825Z5,一款 8U雙路 AI 伺服器,支援最新 AMD Instinct MI325X 加速器,並可搭載最高 6TB 的  DDR5 6000 記憶體,提供 8 個熱插拔 NVMe U.2 托盤,非常適用於大規模 AI 與 HPC 基礎架構的建置需求。其次是MiTAC TYAN TN85-B8261,一款2U雙路GPU伺服器,專為HPC和深度學習應用設計,支援最多 4 張雙寬 GPU 卡、24 個 DDR5 RDIMM 插槽,並擁有 8 個熱插拔 NVMe U.2托盤。而在雲端伺服器方面,神雲科技展出MiTAC TYAN GC68C-B8056,這款1U單路伺服器能搭載 24 個 DDR5 DIMM 插槽與 12 個熱插拔NVMe U.2托盤,為雲端環境提供高效能與靈活的擴展性。此外,神雲科技還展出專為高 IOPS NVMe資料儲存設計的MiTAC TYAN TS70A-B8056 2U 單路伺服器,以及採用2U 4 節點架構設計的 MiTAC M2810Z5 單路伺服器,MiTAC M2810Z5 的每個節點最高可支援3,072GB的RDIMM DDR5 6000 記憶體與 4 個熱插拔 E1.S 硬碟,為高密度雲端伺服器的優異解決方案。 滿足AI與加速需求:可擴展效能提升投資效益並降低功耗 SC24大會上,神雲科技也展出基於Intel Xeon 6處理器,並具備優異性能和效率的次世代伺服器產品;首先是MiTAC R2520G6,這款2U雙路的通用型伺服器專為記憶體密集運算設計,支援高達8TB的DDR5記憶體、5個PCIe 5.0 x16插槽並提供 NVMe U.2和E1.s 靈活的儲存媒介選項。接著是MiTAC R2513G6,這台2U單路儲存伺服器支援24個3.5寸SATA硬碟和2個2.5寸NVMe硬碟,並提供 8 個 DDR5 記憶體插槽,同時內建 SAS RAID 卡實現高效可靠的資料儲存管理,適合高密度儲存應用場合。最後是MiTAC G4520G6,這是款專為平行式AI工作負載打造的4U雙路伺服器,支援多達8張雙插槽GPU卡、8TB DDR5 記憶體,並配備11個 PCIe 5.0 x16 插槽和 8 個熱插拔 U.2硬碟,是高性能運算和 AI 訓練的理想選擇。 此外,神雲科技還展示了英特爾 D50DNP 系列(Denali Pass)中的 D50DNP1MHCPLC 計算模組,這款1U半寬模組可支援兩顆功耗最高達 385W 的第5代/第4代Intel® Xeon® 可擴充處理器或Intel® Xeon® CPU Max 系列,採用直接液冷技術,具備16個DDR5 DIMM 記憶體插槽,最高可達5600 MT/s(搭配第 5 代Intel Xeon可擴展處理器)資料傳輸頻寬、2個半高半長 PCIe Gen 5 插槽以及多達 2 個 NVMe/SATA M.2 硬碟,為高密度計算應用提供了靈活與高效的解決方案。 除了前述介紹的產品,神雲科技在此次SC24大會也展示了其多元化的產品線,包括支援 AMD EPYC 9004/8004/4004 系列處理器、第5代/第4代Intel Xeon 可擴充處理器、Intel® Xeon® E-2400處理器以及專業級GPU的多款伺服器產品,提供符合現代用戶在性能與預算方面需要的各式解決方案。 關於神雲 神雲科技股份有限公司(MiTAC Computing Technology Corporation) 隸屬于神達控股股份有限公司(MiTAC Holdings Corp., TSE:3706),擁有超過 30 年的伺服器設計與製造經驗,為雲端服務供應商、人工智慧/高性能運算,以及邊緣運算等不同的應用環境提供靈活的定制化供應模式。自 2023 年 7 月起,MiTAC開始承接英特爾資料中心解決方案事業群(DSG)伺服器產品之銷售與後續世代產品之規劃設計,為下一代資料中心設備提供先進的解決方案。MiTAC產品陣容涵蓋 TYAN 伺服器、ORAN 伺服器、高性能 AI 伺服器與資料中心產品,以豐富的產品範疇,提供給客戶全方位的企業級解決方案。 神雲原以 TYAN 品牌經營伺服器通路業務,於2023年7月起開始以 MiTAC 品牌經營 Intel DSG 伺服器產品之銷售。而為強化通路規劃、發揮品牌管理的綜效,並凝聚品牌形象,自2024年10月起,神雲整合 TYAN 與 MiTAC 兩個品牌,統一以 MiTAC 作為產品的品牌名稱。 神雲科技網站: http://www.mitaccomputing.com/ Intel, the Intel logo, and other marks are trademarks of Intel Corporation or its subsidiaries.  AMD, the AMD Arrow logo, AMD Instinct, EPYC, and combinations thereof are trademarks of Advanced Micro Devices, Inc.  Other names are for informational purposes only and may be trademarks of their respective owners.

文章來源 : PR Newswire 美通社 發表時間 : 瀏覽次數 : 176 加入收藏 :
MiTAC Computing Unveils New AI/HPC-Optimized Servers with Advanced CPU and GPU Integration at SC24

ATLANTA, Nov. 20, 2024 /PRNewswire/ -- MiTAC Computing Technology Corporation, an industry-leading server platform design manufacturer and a subsidiary of MiTAC Holdings Corporation (TSE:3706), is unveiling its new server lineup at SC24, booth #2543, in Atlanta, Georgia. MiTAC Computing's servers integrate the latest AMD EPYC™ 9005 Series CPUs, AMD Instinct™ MI325X GPU accelerators, Intel® Xeon® 6 processors, and professional GPUs to deliver enhanced performance optimized for HPC and AI workloads. MiTAC Computing Unveils New AI/HPC-Optimized Servers with Advanced CPU and GPU Integration at SC24 Leading Performance and Density for AI-Driven Data Center Workloads MiTAC Computing's new servers, powered by AMD EPYC 9005 Series CPUs, are optimized for high-performance AI workloads. At SC24, MiTAC highlights two standout AI/HPC products: the 8U dual-socket MiTAC G8825Z5, featuring AMD Instinct MI325X GPU accelerators, up to 6TB of DDR5 6000 memory, and eight hot-swap U.2 drive trays, ideal for large-scale AI/HPC setups; and the 2U dual-socket MiTAC TYAN TN85-B8261, designed for HPC and deep learning applications with support for up to four dual-slot GPUs, twenty-four DDR5 RDIMM slots, and eight hot-swap NVMe U.2 drives. For mainstream cloud applications, MiTAC offers the 1U single-socket MiTAC TYAN GC68C-B8056, with twenty-four DDR5 DIMM slots and twelve tool-less 2.5-inch NVMe U.2 hot-swap bays. Also featured is the 2U single-socket MiTAC TYAN TS70A-B8056, designed for high-IOPS NVMe storage, and the 2U 4-node single-socket MiTAC M2810Z5, supporting up to 3,072 GB of DDR5 6000 RDIMM memory and four easy-swap E1.S drives per node. Meeting AI and Acceleration Needs: Scalable Performance to Enhance Investment and Save Power At SC24, MiTAC Computing presents its new Intel Xeon 6-based servers, offering trusted performance and exceptional efficiency. First is the 2U dual-socket general-purpose compute server, the MiTAC R2520G6, built for in-memory computing with up to 8TB of DDR5 memory, five PCIe 5.0 x16 slots, and flexible storage options for U.2 and E1.s SSDs. Next is the 2U single-socket storage server, MiTAC R2513G6, which supports twenty-four 3.5" SATA drives and two 2.5" NVMe drives, along with eight DDR5 memory slots with an integrated SAS RAID card for storage management. Finally, the 4U dual-socket MiTAC G4520G6 GPU server supports parallel AI workloads with up to eight GPUs, 8TB of DDR5 memory, eleven PCIe 5.0 x16 slots, and eight hot-swap U.2 drives. In addition, MiTAC Computing showcases one of the Intel® Server D50DNP Family (Denali Pass) SKUs, the D50DNP1MHCPLC, which supports up to 385W with two 5th/4th Gen Intel® Xeon® Scalable processors or Intel® Xeon® CPU Max Series. This 1U half-width compute module features direct liquid cooling, 8+8 DDR5 DIMMs up to 5600 MT/s (5th Gen), two low-profile PCIe Gen 5 slots, and up to two NVMe/SATA M.2 drives. Along with its previously showcased products, MiTAC Computing introduces a diverse server lineup at SC24, featuring AMD EPYC 9004/8004/4004 series processors, 5th/4th Gen Intel Xeon Scalable processors, Intel® Xeon® E-2400 processors and professional GPUs, providing solutions to meet the performance and budget needs of modern users. About MiTAC Computing Technology Corporation MiTAC Computing Technology Corporation, a subsidiary of MiTAC Holdings Corp. (TSE:3706), specializes in cloud, AI/HPC and edge computing solutions and has over 30 years of design and manufacturing expertise. With a strong focus on large-scale data centers, MiTAC offers flexible and customized solutions for various systems and applications. Our product lineup includes TYAN servers, ORAN servers, high-performance AI servers, and other data center products. Originally, MiTAC operated the server channel business under the TYAN brand and began selling Intel DSG server products under the MiTAC brand in July 2023. As of October 2024, we have integrated the TYAN and MiTAC brands to streamline our operations and enhance brand management. MiTAC now serves as the sole brand name for all our products. MiTAC Computing Technology Corporation website:http://www.mitaccomputing.com/ Intel, the Intel logo, and other marks are trademarks of Intel Corporation or its subsidiaries. AMD, the AMD Arrow logo, AMD Instinct, EPYC, and combinations thereof are trademarks of Advanced Micro Devices, Inc. Other names are for informational purposes only and may be trademarks of their respective owners.

文章來源 : PR Newswire 美通社 發表時間 : 瀏覽次數 : 150 加入收藏 :
MiTAC Computing Unveils New AI/HPC-Optimized Servers with Advanced CPU and GPU Integration at SC24

ATLANTA, Nov. 20, 2024 /PRNewswire/ -- MiTAC Computing Technology Corporation, an industry-leading server platform design manufacturer and a subsidiary of MiTAC Holdings Corporation (TSE:3706), is unveiling its new server lineup at SC24, booth #2543, in Atlanta, Georgia. MiTAC Computing's servers integrate the latest AMD EPYC™ 9005 Series CPUs, AMD Instinct™ MI325X GPU accelerators, Intel® Xeon® 6 processors, and professional GPUs to deliver enhanced performance optimized for HPC and AI workloads. MiTAC Computing Unveils New AI/HPC-Optimized Servers with Advanced CPU and GPU Integration at SC24 Leading Performance and Density for AI-Driven Data Center Workloads MiTAC Computing's new servers, powered by AMD EPYC 9005 Series CPUs, are optimized for high-performance AI workloads. At SC24, MiTAC highlights two standout AI/HPC products: the 8U dual-socket MiTAC G8825Z5, featuring AMD Instinct MI325X GPU accelerators, up to 6TB of DDR5 6000 memory, and eight hot-swap U.2 drive trays, ideal for large-scale AI/HPC setups; and the 2U dual-socket MiTAC TYAN TN85-B8261, designed for HPC and deep learning applications with support for up to four dual-slot GPUs, twenty-four DDR5 RDIMM slots, and eight hot-swap NVMe U.2 drives. For mainstream cloud applications, MiTAC offers the 1U single-socket MiTAC TYAN GC68C-B8056, with twenty-four DDR5 DIMM slots and twelve tool-less 2.5-inch NVMe U.2 hot-swap bays. Also featured is the 2U single-socket MiTAC TYAN TS70A-B8056, designed for high-IOPS NVMe storage, and the 2U 4-node single-socket MiTAC M2810Z5, supporting up to 3,072 GB of DDR5 6000 RDIMM memory and four easy-swap E1.S drives per node. Meeting AI and Acceleration Needs: Scalable Performance to Enhance Investment and Save Power At SC24, MiTAC Computing presents its new Intel Xeon 6-based servers, offering trusted performance and exceptional efficiency. First is the 2U dual-socket general-purpose compute server, the MiTAC R2520G6, built for in-memory computing with up to 8TB of DDR5 memory, five PCIe 5.0 x16 slots, and flexible storage options for U.2 and E1.s SSDs. Next is the 2U single-socket storage server, MiTAC R2513G6, which supports twenty-four 3.5" SATA drives and two 2.5" NVMe drives, along with eight DDR5 memory slots with an integrated SAS RAID card for storage management. Finally, the 4U dual-socket MiTAC G4520G6 GPU server supports parallel AI workloads with up to eight GPUs, 8TB of DDR5 memory, eleven PCIe 5.0 x16 slots, and eight hot-swap U.2 drives. In addition, MiTAC Computing showcases one of the Intel® Server D50DNP Family (Denali Pass) SKUs, the D50DNP1MHCPLC, which supports up to 385W with two 5th/4th Gen Intel® Xeon® Scalable processors or Intel® Xeon® CPU Max Series. This 1U half-width compute module features direct liquid cooling, 8+8 DDR5 DIMMs up to 5600 MT/s (5th Gen), two low-profile PCIe Gen 5 slots, and up to two NVMe/SATA M.2 drives. Along with its previously showcased products, MiTAC Computing introduces a diverse server lineup at SC24, featuring AMD EPYC 9004/8004/4004 series processors, 5th/4th Gen Intel Xeon Scalable processors, Intel® Xeon® E-2400 processors and professional GPUs, providing solutions to meet the performance and budget needs of modern users. About MiTAC Computing Technology Corporation MiTAC Computing Technology Corporation, a subsidiary of MiTAC Holdings Corp. (TSE:3706), specializes in cloud, AI/HPC and edge computing solutions and has over 30 years of design and manufacturing expertise. With a strong focus on large-scale data centers, MiTAC offers flexible and customized solutions for various systems and applications. Our product lineup includes TYAN servers, ORAN servers, high-performance AI servers, and other data center products. Originally, MiTAC operated the server channel business under the TYAN brand and began selling Intel DSG server products under the MiTAC brand in July 2023. As of October 2024, we have integrated the TYAN and MiTAC brands to streamline our operations and enhance brand management. MiTAC now serves as the sole brand name for all our products. MiTAC Computing Technology Corporation website:http://www.mitaccomputing.com/ Intel, the Intel logo, and other marks are trademarks of Intel Corporation or its subsidiaries. AMD, the AMD Arrow logo, AMD Instinct, EPYC, and combinations thereof are trademarks of Advanced Micro Devices, Inc. Other names are for informational purposes only and may be trademarks of their respective owners.

文章來源 : PR Newswire 美通社 發表時間 : 瀏覽次數 : 456 加入收藏 :
Supermicro推出適用於AI就緒資料中心的全新伺服器和GPU 加速系統,搭載AMD EPYC™ 9005系列CPU與AMD Instinct™ MI325X GPU

全新Supermicro系統可幫助客戶實現資料中心升級,進而更順暢地運行AI工作負載 加州聖荷西2024年10月16日 /美通社/ -- Supermicro, Inc.(納斯達克股票代碼:SMCI)作為AI/ML、HPC、雲端、儲存和5G/邊緣領域的全方位IT解決方案提供企業,宣布推出搭載AMD EPYC™ 9005系列處理器和AMD Instinct™ MI325X GPU的全新伺服器、GPU加速系統與儲存伺服器系列。全新H14產品系列為業界內機種最廣泛的伺服器系列之一,此系列包含Supermicro的Hyper系統、Twin多節點伺服器和AI推論GPU系統,且這些機種皆可搭配氣冷或液冷散熱技術。新型「Zen5」處理器核心架構使用了具有完整資料路徑的AVX-512向量指令,支援基於CPU的AI推論,且每週期指令數(IPC)比先前第四代EPYC處理器提升17%,使每核心的效能更為優異。 100124-AMD Supermicro的H14系列採用最新第五代AMD EPYC處理器,每個CPU最多搭載192個核心,且散熱設計功率(TDP)最高可達500W。新型H14系列包含Hyper和FlexTwin™系統,可因應更高的散熱需求。該系列也包含三個適用於AI訓練於推論工作負載的系統,最多可支援10個GPU,並以AMD EPYC 9005系列CPU作為主要處理器。此外,H14系列也具有兩個支援AMD Instinct MI325X GPU的系統。 Supermicro總裁暨執行長梁見後表示:「相較於採用第二代EPYC 7002系列CPU的Supermicro H11系統,搭載EPYC 9005 64核心CPU的Supermicro H14伺服器在SPECrate®2017_fp_base的效能表現註1上提升了2.44倍。此大幅度的效能改善使客戶可將資料中心總佔用空間縮減至少三分之二註2,同時增加新的AI運算性能,進一步提升資料中心能源效率。憑藉Supermicro的液冷和氣冷技術、多元系統設計,以及經驗證的建構組件解決方案,H14伺服器系列能夠提供最高的效能、密度和能源效率。」 如需 Supermicro H14 產品系列的詳細資訊,請瀏覽:www.supermicro.com/aplus AMD執行副總裁暨資料中心解決方案事業群總經理Forrest Norrod表示:「Supermicro的『Building Block解決方案』使其能夠持續使基於AMD架構的解決方案快速上市,並涵蓋多種極具吸引力的系統設計。我們與Supermicro合作,運用其全球內部工程設計和製造能力,再結合他們的氣冷和液冷系統機架級整合技術,可在任何規模上使客戶透過AMD EPYC CPU和Instinct GPU快速實現價值。」 Supermicro的 H14伺服器陣容包含以下產品系列: Hyper - Supermicro的旗艦級企業伺服器,可支援兩個EPYC 9005 CPU,每個CPU最多包含192 個核心,並具有500W的功率,同時可透過24個DIMM插槽支援最高9TB記憶體,提供最高的效能。Hyper包括配備最高12個2.5吋NVMe/SATA磁碟機槽的1U機箱,或配備最高24個2.5吋NVMe/SATA磁碟機槽的2U機箱,採用先進的散熱設計,可容納最高效能的CPU,適用於需求嚴苛的AI推論、企業或雲端工作負載。 CloudDC - 此多功能系統經過最佳化設計,適用於雲端資料中心,並在1U規格的機殼內搭載單一EPYC 9005 CPU,具有最高12個2.5吋NVMe/SATA硬碟機槽。此系統採用OCP(開放運算平台)DC-MHS 規格(資料中心模組化硬體系統)技術,確保與OCP標準相容。 GrandTwin™ - GrandTwin為4節點運算平台,將單一EPYC 9005 CPU容納到高密度2U外形規格內。此系統經常用於物件儲存、虛擬化、高效能運算應用等多伺服器叢集應用。 FlexTwin™ - FlexTwin是一款2U 4節點高效能、高密度運算系統,其中每個節點具有雙EPYC 9005 CPU。此系統採用先進液冷技術,可實現高能源效率,且能使最高效能的 EPYC 9005 CPU順暢運行,適用於高效能運算、EDA和其他需求嚴苛的工作負載。 5U GPU 系統 - Supermicro搭載 EPYC CPU的5U PCIe GPU系統最多支援10個雙寬加速器,適合設計與視覺化應用。 4U GPU 系統(液冷) - 搭載EPYC CPU的高密度8路加速器平台,使用先進的液冷技術,在最緊湊的4U外形規格內支援OAM加速器,專為高效能AI和高效能運算應用所設計。 8U GPU 系統 - 此8路加速器系統使用AMD Instinct MI325X GPU及EPYC 9005 CPU,適用於大規模的LLM AI訓練。其8U機箱可被部署在任何氣冷資料中心內。 搭載AMD EPYC CPU的Supermicro H14產品現已可供客戶透過Supermicro JumpStart計畫進行測試。 Supermicro已於2024年10月10日在舊金山 Moscone中心舉行的AMD Advancing AI Day上展示了全新H14解決方案。 註1 Supermicro A+ Server 2023US-TR4搭配兩個AMD EPYC 7702 64核心CPU,在 SPECrate®2017_fp_base測試中得分為485(https://www.spec.org/cpu2017/results/res2020q1/cpu2017-20200204-20866.html,資料擷取日期為2024年10月2日),對照搭配兩個AMD EPYC 9555 64核心CPU的Supermicro AS-2126HS-TN,於SPECrate®2017_fp_base測試中得分為1670。搭載9555 CPU的AS-2126HS-TN結果已於2024年10月10日在www.spec.org/cpu2017/results上發布。 註2根據註(1)所述的SPECrate®2017_fp_base測試比較,使用搭載9555 CPU的AS-2126HS-TN與搭載7702 CPU的2023US-TR4相比,系統數量減少了70.9%。 關於Super Micro Computer, Inc. Supermicro(納斯達克股票代碼:SMCI)為應用最佳化全方位IT解決方案的全球領導者。Supermicro的成立據點及營運中心位於美國加州聖荷西,致力為企業、雲端、AI和5G電信/邊緣IT基礎架構提供領先市場的創新技術。我們是全方位IT解決方案製造商,提供伺服器、AI、儲存、物聯網、交換器系統、軟體及支援服務。Supermicro的主機板、電源和機箱設計專業進一步推動了我們的發展與產品生產,為全球客戶實現了從雲端到邊緣的下一代創新。我們的產品皆由企業內部團隊設計及製造(在美國、亞洲及荷蘭),透過全球化營運實現極佳的規模與效率,並藉營運最佳化降低總體擁有成本(TCO),以及經由綠色運算技術減少環境衝擊。屢獲殊榮的Server Building Block Solutions®產品組合,讓客戶可以自由選擇這些具高度彈性、可重複使用且極為多元的建構式組合系統,我們支援各種外形尺寸、處理器、記憶體、GPU、儲存、網路、電源和散熱解決方案(空調、自然氣冷或液冷),因此能為客戶的工作負載與應用提供最佳的效能。 Supermicro、Server Building Block Solutions和We Keep IT Green皆為Super Micro Computer, Inc.的商標和/或註冊商標。 所有其他品牌、名稱和商標皆為其各自所有者之財產。 AS-812GS AS-1116CS AS 2116GT AS-212 AS-5126GS AS-4125GS AS-2126FT    

文章來源 : PR Newswire 美通社 發表時間 : 瀏覽次數 : 741 加入收藏 :
2025 年 1 月 22 日 (星期三) 農曆十二月廿三日
首 頁 我的收藏 搜 尋 新聞發佈