Microsoft has developed the Azure HPC & AI Collaboration Centers program in partnership with its partners, which supports partnerships with well-known organizations in an attempt to establish and share best practices with both the HPC and AI groups.
With Duke University, Purdue University, Red Oak Consulting, Numerical Algorithms Group (NAG), and RedLine Results, Microsoft is delivering this software in collaboration with AMD, Intel, and Nvidia. So far, five Azure Collaboration Centers have been developed with Duke University, Purdue University, Red Oak Consulting, Numerical Algorithms Group (NAG), and RedLine Performance.
Microsoft Azure is able to provide differentiated, purpose-built cloud capabilities for the full range of HPC & AI workloads, including the most challenging use cases. However, HPC & AI clusters that use HPC-class InfiniBand interconnect with genuine low latency (~1us) and high bandwidth are at the heart of the infrastructure, usually using the latest processor technologies from its program partners.
Azure is able to demonstrate the scaling of supercomputer workloads to 12 times more CPU cores than other clouds as a result of these performance benefits and to provide cost-to-solution leadership over other solutions. Simultaneously, Microsoft has grown HPC and AI expertise across the board, from engineering teams to customer engagements to after-sales support.
Collaboration Centers for the HPC and AI
The ultimate aim of the new initiative by Microsoft and its partners is to explain and share best practices on how to use Azure HPC & AI capabilities to unlock technologies and competitiveness for customers.
In conjunction with Nvidia, the Duke University Azure HPC Collaboration Center aims to use Azure GPU technologies to develop customized cloud hemodynamic simulation capabilities to enhance our understanding and treatment of coronary bifurcation lesions.
Purdue University’s Azure HPC & AI Collaboration Center will show best practices for using the Azure HPC cloud to complement on-premises HPC systems like Purdue’s AMD-based Anvil supercomputer, in collaboration with AMD.
In collaboration with Intel, the Read Oak Consulting Azure HPC & AI Collaboration Center will create and demonstrate best practices for creating HPC clusters on Azure using the chipmaker’s technology.