According to the latest reports, the US Department of Energy’s Argonne National Laboratory has tapped Nvidia’s A100 GPUs paired with AMD’s CPUs for its new Polaris supercomputer.
This is because Intel has reportedly made too many delays in delivering the Intel-powered Aurora supercomputer. The main reason for the delay from Intel is the difficulties that the blue team faced with the production of their Sapphire Rapids server chips.
Coming to the supercomputer, Polaris, it will employ four Nvidia A100s and two AMD CPUs per node. This will be a total of 2,240 Nvidia GPUs spread across 560 nodes helping Polaris to deliver up to 44 PetaFLOPS of FP64 performance.
This will be a much less powerful machine than the planned Aurora exascale supercomputer that was supposed to deliver up to one ExaFLOP of sustained performance when it arrives in 2022 – 2023.
however, the Polaris will still reach up to 1.4 “AI ExaFLOPS” of performance, which isn’t measured with the standard FP64 workload used to quantify supercomputer performance. And the supercomputer’s 44 PetaFLOPs of performance qualifies for a spot in the top ten of the Top 500 list of the fastest supercomputers in the world.
“The system will accelerate transformative scientific exploration, such as advancing cancer treatments, exploring clean energy, and propelling particle collision research to discover new approaches to physics. And it will transport the ALCF into the era of exascale AI by enabling researchers to update their scientific workloads for Aurora, Argonne’s forthcoming exascale system.”
as for the CPU which will power the system, it will be AMD’s 32-core Epyc Rome 7532 CPUs and it will then be changed in March 2022 to newer 32-core Epyc Milan 7543 chips. The system will then be updated later from a Slingshot 10 to the Slingshot 11 fabric to match Aurora.