Accelerating Arm for Zettascale

Arm-based HPC has made a strong impact on the HPC community as evidenced by several Arm-based projects including the Japanese Fugaku supercomputer, Sandia’s Astra system, the UK’s GW4/EPSRC efforts and Europe’s “Mont-Blanc” project. Arm is now a major player in the HPC field, and the past two years have demonstrated that Arm HPC will be an option for future large-scale cluster deployments. At the same time, momentum for Arm HPC has shifted in the last year to cloud-based HPC as typified by Amazon’s Graviton 2 processor and Arm’s announcement of the Neoverse 2 platform. This leads us to consider two questions: 1) Has traditional Arm HPC growth been supplanted by the rapid growth of cloud-based HPC? and 2) What trends and challenges do we see in terms of growing Arm HPC for both traditional and cloud HPC environments?

In this BoF we will discuss these topics with an expert panel composed of leading government and industry practitioners, including participants from government and academic labs and vendors working on developing the next generation of Arm HPC hardware. With the coming deployment of the first exascale machines, we plan to look even farther out to the next boundary of “zettascale” computing. Specifically, panelists will discuss whether there are any potential software or tool roadblocks in supporting a diverse Arm HPC hardware ecosystem and how we can best collaborate to facilitate the growth and interoperability of future Arm-based clusters.

---

Contributors:

• Jeffrey Young (Georgia Institute of Technology)
• Michèle Weiland (Edinburgh Parallel Computing Centre (EPCC))
• Mitsuhisa Sato (RIKEN)
• Steve Poole (Los Alamos National Laboratory)
• Roxana Rusitoru (Arm)
• Satoshi Matsuoka (RIKEN Center for Computational Science)
• Brendan Bouffler (Amazon)
• John Linford (NVIDIA)