RISC-V for HPC: An update of where we are and main action points

2025-06-18

AI Summary

This work systematically identifies key bottlenecks and ecosystem gaps hindering RISC-V’s adoption in high-performance computing (HPC), including insufficient floating-point performance, limited vectorization support, constrained memory bandwidth, and poor scalability for large-scale parallel workloads. To address these challenges, the study introduces the first cross-vendor, full-stack HPC suitability analysis framework—developed under the RISC-V HPC Special Interest Group—covering critical dimensions: ISA extensions (e.g., V, Zfa, Zfh), compiler optimizations (LLVM/GCC), MPI and OpenMP runtime compatibility, interconnect integration, and accelerator co-design. Based on this comprehensive assessment, the paper delivers the inaugural RISC-V HPC prioritized action roadmap and a concrete, implementable technical攻坚 (technical攻坚 translated as “technical攻坚”) list—spanning hardware enhancements, software toolchain improvements, and system-level integration strategies. The framework establishes an authoritative benchmark and practical guidance for coordinated hardware-software evolution toward production-ready RISC-V HPC systems.

Technology Category

• Machine Learning: Hardware-aware ML
• Humans and AI: Other Foundations of Human Computation & AI
• Computer Vision: Adversarial Attacks & Robustness

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Web performance, measurement, and characterization
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Responsible Web: Measurement, analysis, and circumvention of Web censorship

Abstract

This extended abstract is submitted on behalf of the RISC-V HPC SIG who have been undertaking an analysis to explore the current state and limitations of the RISC-V ecosystem for HPC. Whilst it is right to celebrate that there has been great progress made in recent years, we also highlight limitations and where effort should be focussed.

Problem

• Assessing RISC-V ecosystem state for HPC
• Identifying current RISC-V HPC limitations
• Prioritizing focus areas for RISC-V HPC

Innovation

• RISC-V ecosystem analysis for HPC
• Identifying current limitations and progress
• Focusing efforts on key action points