HPC | 高效能運算

  • 高效能運算是什麼?
    高效能運算(High Performance Computing)簡稱HPC,係指能高速處理數據或執行指令的運算能力,特別形容每秒浮點運算次數(FLOPS)超越一兆次(teraFLOPS)的作業系統。全球領先的超級電腦每秒浮點運算次數已突破一千兆(petaFLOPS),下一個里程碑將是每秒一百萬兆浮點運算次數(exaFLOPS),即所謂的「百萬兆級運算」(exascale computing)。

    為達成此目標,須集結大量的CPU或GPU進行平行運算運算叢集是常見的高效能運算解決方案之一,數台電腦(或稱nodes「節點」)互相串聯成為群組,運算能力遠超過單一電腦主機,多數節點可協同解決一台電腦無法處理的問題。

  • 為何需要?
    無論透過運算叢集或頂尖大型電腦,為組織添加高效能運算能力都是必要任務。高效能運算可解決科學、工程學及商務上的許多問題,例如:

    科學:學術研究機構的科學團隊可利用高效能運算探討宇宙的誕生、探索粒子物理學、或模擬和預測天氣氣候變化。

    媒體娛樂業:例如算圖農場的高效能運算解決方案,可繪製動畫與特效、剪接影片、或將特殊活動直播給全球觀眾。

    人工智慧:機器學習是高效能運算的一種運用方法,目前備受矚目,可將人工智慧應用於各種不同領域,包括自駕車、臉部辨識軟體、語音辨識與翻譯、或無人機操作。

    石油與天然氣:高效能運算可處理衛星影像、海床聲納探查結果等數據,發現值得開採的原油或天然氣礦床。

    金融服務:高效能運算可追蹤股價或執行算法交易,也可分析交易紀錄以察覺非法行為。

    醫療健保:可透過高效能運算開發糖尿病等重大疾病的解方,或提升診斷速度與準確度,例如癌症篩檢技術。

  • 技嘉的特色
    技嘉H系列高密度伺服器G系列GPU協同運算伺服器,將大量運算能力集中在1U、2U或4U伺服器機殼內,特別適用於高效能運算解決方案。伺服器間透過Ethernet、Infiniband或Omni-Path等網路架構互相通聯,可組成運算叢集。高效能運算伺服器的範例包括技嘉H262系列,搭配AMD EPYC™ 7002系列處理器,可於2U伺服器機殼內達成512核心/1024執行緒數的運算能力(單一系統含四節點,各節點含兩顆64核心的AMD EPYC™處理器)。如果使用完整42U的位置(僅保留網路交換器的額外空間),可達成10240核心/20480執行緒數的超高效能,明顯提升整體運算能力。

  • WE RECOMMEND
    RELATED ARTICLES
    In the Quest for Higher Learning, High Density Servers Hold the Key
    A top technological university in Europe noticed rising demand for computing services across its various departments. It decided to build a next-generation data center with GIGABYTE's high density servers. With the right tools in place, scientists were able to accelerate their research, analyze massive amounts of information, and complete more data-intensive projects. Science advanced while the institute flourished.
    The European Organization for Nuclear Research Delves into Particle Physics with GIGABYTE Servers
    The European Organization for Nuclear Research (CERN) bought GIGABYTE's high-density GPU Servers outfitted with 2nd Gen AMD EPYC™ processors. Their purpose: to crunch the massive amount of data produced by subatomic particle experiments conducted with the Large Hadron Collider (LHC). The impressive processing power of the GPU Servers’ multi-core design has propelled the study of high energy physics to new heights.
    Decoding the Storm with GIGABYTE’s Computing Cluster
    Waseda University, the “Center for Disaster Prevention around the World”, has built a computing cluster with GIGABYTE’s GPU server and tower servers. They use it to study and prepare for natural disasters, such as tsunamis and storm surges. Efforts go into understanding the tropical cyclones of tomorrow, which are thought to become more dangerous due to climate change.
    Success Story: Achieving Naked-Eye 3D with Virtual Reality
    n'Space, a projector-based platform, is capable of implementing mixed reality without wearable devices. Users can immerse in a virtual environment and interact with 3D projections without the need to cover their faces with electronics. The real world just got an expansion pack in the infinite realm of the imagination. ArchiFiction, the company behind the invention, optimized the virtual experience with GIGABYTE's solution, which can process large amounts of data with high performance.