At last year's SC19 conference, Microsoft Azure introduced HBv2 virtual machine clusters with the bold claim that they "rival the most advanced supercomputers on the planet." Just a year later, at the virtual Supercomputing 2020 (SC20) event, the software giant revealed that its public cloud computing service had joined the ranks of the world's most powerful data-intensive supercomputers by taking the position. 17 on the prestigious Graph500 list. According to Microsoft, this is the first time a public cloud has been placed on Graph500 and since the company's HBv2 VMs generate 1151 GTEPs (Giga-Traversed Edges Per Second), Azure's placement on the list ranks in the top six. percent of all time for published submissions. Microsoft also announced that it achieved a new scaling record for HPC based on the messaging interface (MPI) in the public cloud. By running Nanocale Molecular Dynamics (NAMD) on 86,400 processor cores, Azure has shown that researchers around the world can have pet-scale computing at their fingertips. The company also participated in the HPC COVID-19 consortium, and a team led by Azure's Dr. Jer-Ming Chia worked with researchers at the University of Illinois' Beckman Institute for Advanced Science and Technology to evaluate HBv2 VMS. to support future simulations of the SARS-CoV-2 virus. To the team's surprise, they found that the HBv2 clusters could not only meet the demands of the researchers, but their performance and scalability on Azure rivaled and even exceeded the capabilities of the Frontier supercomputer in some cases.
Graph500 vs TOP500
To compile its twice-yearly list of the Top 500 Supercomputers, TOP500 uses Jack Dongarra's Linpack benchmark, as it is widely used and performance figures are available for almost all affected systems. The Graph500 list, on the other hand, focuses on data-intensive workloads, which is why it uses its own benchmark. As government, business, and research organizations become more data-centric, Graph500 serves as a useful barometer for customers and partners trying to migrate complex data problems to the cloud. The BST (Breadth-first search) test is part of the Graph500 benchmark and emphasizes supercomputing and HPC environments in a number of ways, while emphasizing the ability to move data. The test uses the "popcount" CPU instruction which is particularly useful for client workloads in cryptography, molecular fingerprinting, and extremely dense data storage. In a blog post, Dr. William Chappell, vice president of Mission Systems at Microsoft, discussed how organizations can now use HBv2 enterprise clusters to solve complex data problems instead of configuring their own systems. he saying, “When mission-critical customers have a unique need, such as a difficult sparse graphics problem, they no longer need to configure their own system for world-class performance. As we compete with the results of the top ten machines in the world, this demo shows that anyone with a unique mission, including critical government users, can take advantage of our existing capabilities. Since this is done without the cost and burden of ownership, it changes the way mission users will access high-performance computing. I see this as a very important democratization of the impact of HPC. "