RISC-V for HPC: Where we currently are and where we need to go

Talk by Nick Brown (from RISC-V H&ES project) on RISC-V for HPC at RISC-V Summit North America.
The powerhouse which unlocks the ability to simulate complex, real world problems, as well as powering AI and ML workloads, High Performance Computing (HPC) is a crucial part of the modern day world. Supercomputers most commonly leverage x86 CPUs and Nvidia or AMD GPUs, however, as the ever-increasing demand by users for more capability meets a growing focus on sustainability, alternative technologies such as RISC-V are important.
RISC-V can offer benefits in performance and energy efficiency to HPC through the potential for specialisation, however the HPC community is yet to embrace RISC-V. But with increased availability of commodity RISC-V high performance CPUs (e.g. the SG2042) and PCIe accelerator cards, RISC-V is becoming a more serious option.
In the RISC-V HPC SIG our role to help drive adoption, and in this talk I will describe where the RISC-V ecosystem currently lies for HPC, explore performance and energy efficiency of latest generation RISC-V hardware against that currently more commonplace in HPC, and highlight key areas that we as the RISC-V community should prioritise to drive RISC-V adoption in HPC. Ultimately acting as a call to action for the RISC-V community.