Nvidia unveils Ampere-based RTX A6000,A40 GPU,Suitable for workstations and data centers

       Nvidia on Monday announced RTX A6000 and A40 "Ampere" GPUs for high-end workstations and data centers,And these brands are not well known。before this,The company recently decided to replace the Tesla V100 with the Nvidia A100,and may stop offering the "Titan" brand for its high-end consumer cards。many people think,This move could be the end of the road for its 'Quadro' brand of professional graphics cards。

       New GPUs bring heavy computing power,The company said it would,Designers and Engineers Accelerate Rendering,AI,VR / AR and computational workload。They are built on the NVIDIA Ampere architecture,Has a new RT core,Tensor cores and CUDA cores。

RTX A6000

       根據Nvidia的說法第二代RT內核的輸送量是前一代的兩倍再加上併發的光線跟蹤著色和計算。Simultaneously,第三代Tensor內核的輸送量高達前一代的5倍稀疏性則高達10倍並支持新的TF32和BF16數據格式至於新的CUDA內核它們可提供高達上一代FP32輸送量2倍的效能從而顯著提高了圖形和計算能力

       在硬數位方面兩款卡均採用Nvidia的Ampere架構並包含10572個CUDA內核以及48GB的GDDR6 RAM其他規格也相對相似包括336張量內核,384比特記憶體匯流排寬度和300W TDP

       but,儘管前者用於標準工作站而後者則用於商業資料中心和服務器場。same,雖然兩者都帶有GDDR6 RAM但A6000的時鐘頻率為16Gbps而A40的記憶體時鐘為14.5Gbps這些卡將僅使用DisplayPort進行視頻輸出並將通過PCIe Gen 4連接到電腦

       Nvidia尚未宣佈其新GPU的確切發佈日期或價格但表示RTX A6000將從包括PNYLeadtekIngram MicroRyoyo和nvidia.com在內的許多通路合作夥伴處購買十二月明年戴爾,HP,聯想和Boxx等領先的系統製造商也將提供它們

       至於A40從明年年初開始包括CiscoDellFujitsuHewlett Packard Enterprise和Lenovo在內的OEM工作站和服務器廠商也將提供A40

Leave a Reply

Your email address will not be published. Required fields are marked *