This post has been republished via RSS; it originally appeared at: Microsoft Tech Community - Latest Blogs - .
Microsoft announces new NDm A100 v4 Public AI Supercomputers and achieves Top10 Ranking in TOP500 list
Microsoft Azure is continually demonstrating leadership of HPC and AI in the cloud, making the most powerful AI Supercomputers available to the broadest range of users for real world applications.
5 years ago at Ignite we shared with you the world’s first cloud-based AI supercomputer.
At Microsoft Build in May 2020 we announced building a supercomputer in collaboration with OpenAI comparable to the world’s Top 5 supercomputers.
Since then, we have continued to push the boundaries. In the June 2021 TOP500 list Microsoft Azure took public cloud services to a new level, demonstrating publicly available AI supercomputers that took four consecutive spots from No. 26 to No. 29 on the TOP500 list. These clusters are part of the ND A100 v4 family, available on demand in four global regions today and these rankings were achieved on a portion of our overall cluster size. Each of the systems delivered 16.59 petaflops on the HPL benchmark used for the TOP500 rankings.
In the latest TOP500 list rankings, announced at Supercomputing 21, Microsoft Azure has joined the Top 10 club, with the highest new entry of the SC21 edition of the list at 30.05 Petaflops. Together with the original 4 clusters, now ranked at 32 – 35, that’s 5 supercomputers in the Top40. This is based on our newly announced NDm A100 80GB v4, available on demand.
“The convergence of HPC and AI is a revolution, bringing dramatic acceleration to every kind of simulation, and advancing fields across science and industry,.” said Ian Buck, Vice President and General Manager of Accelerated Computing at NVIDIA. “The Azure NDm A100 v4 instance combines the power of NVIDIA GPU acceleration and NVIDIA InfiniBand networking to enable researchers to make new discoveries faster and advance state-of-the-art science.”
Under the hood, eight NVIDIA A100 Tensor Core GPUs power each virtual instance of the Azure cluster. Each chip has its own HDR 200G InfiniBand link that can create fast connections to thousands of GPUs in Azure.
See our recent GA announcement for the new Azure NDm A100 v4 Virtual Machine now available in Azure.