Voyager-EUS2, a Microsoft Azure supercomputer, has made it onto the list of the world’s 10 fastest machines.
With a benchmark speed of 30 petaflops per second (Pflop/s), Microsoft’s supercomputer is still behind China’s Tianhe-2A and the US Department of Energy’s IBM-based Summit supercomputer. However, it is the only major cloud provider to have placed a supercomputer in the top 10 in the high-performance computing (HPC) Top500 category.
Voyager-EUS2 is the only new entry in the Top500’s top 10 list of the fastest supercomputers on the planet, led by the Japanese computer Fugaku with 7.63 million cores and a Linpack benchmark score of 442 Pflop/s.
‘Flops’ refers to floating point operations per second, which measure the performance of supercomputers used primarily in science and run mainly on programming languages such as Fortran.
Fugaku’s benchmark is three times ahead of the US Department of Energy-sponsored Summit Supercomputer, which is the fastest in the US with a Linpack score of 148.8 Pflop/s. It is made from IBM’s Power9 CPUs and has 4,356 nodes with 22 cores each supported by six Nvidia Tesla V100 GPUs.
Fourth on the list is Sierra, a supercomputer at the Lawrence Livermore National Laboratory at the University of California, which is also powered by Power9 CPUs and Nvidia Tesla V100 GPUs and achieves 94.6 Pflop/s.
At number ten, Microsoft’s Azure supercomputer is the only newcomer to the top 10 list and ranks at the top of all public cloud providers.
Microsoft’s Voyager-EUS2, which runs from the Azure East US 2 region, stands out for its various features: First, it runs a Linux distribution, namely the Ubuntu 18.04 long term servicing (LTS) edition, with 253,440 cores on AMD EPYC CPUs.
The tech giant boasted this week that its Azure cloud currently features five supercomputers in the top 500 list. Microsoft uses supercomputers for artificial intelligence and sells its Azure HPCs as a service.
The company also announced the general availability of the Azure virtual machine, called the “NDm A100 v4 Series” which features Nvidia A100 Tensor Core 80GB GPUs – twice as many as the Nvidia A100 Tensor Core GPUs.