AI-Driven Software Porting to RISC-V (riscv.org)

Google-backed RISE and RISC-V International launched a call for university-led proposals that use AI — especially generative models — to accelerate porting and hardening of open-source software for RISC‑V. The program targets research and tooling that can automatically rewrite build systems, adapt tests and code to RISC‑V’s memory model, generate RISC‑V vector code, inline assembly or intrinsics, and create or update critical infrastructure like simulators, compilers, architectural tests and formal models. Suggested areas include model/tuning innovations, prompting or agentic workflows for porting, test-suite/fuzzer generators, performance models, and concrete forward progress on server-grade RISC‑V stacks. Awards are compute-credit grants (up to $50k USD) delivered as Google Cloud credits to run Google’s Gemini models; credits cannot be converted to cash. Eligible applicants are active researchers at degree-granting institutions advising students; proposals (max three pages) are due Oct 2, 2025, with decisions by Oct 21, 2025. Reviewers from RISE and RISC‑V International will favor open, upstream-impacting work with clear methodology, timelines, and disclosure plans (code/data/publication). For AI/ML researchers this is a practical, funded pathway to build end-to-end toolchains and evaluation pipelines that can measurably speed RISC‑V adoption and produce reusable infrastructure and benchmarks for the broader ecosystem.