Open Source Compiler Tool Chains for RISC V Past, Present and Future
summary
RISC‑V compiler work is usable but uneven: GCC is more mature (out‑of‑tree since 2011) while LLVM reached mainline recently and gained RV64 support in 2018–2019. Key gaps: musl/C library work, compiler‑rt build pain, and sparse Fortran/Flang support. Prototype RV32 bit‑manipulation lowers code size dramatically on some workloads — a 16% size win on AES/SHA‑256 — and iterative tuning with OpenTuner yields ~15% gains; sustained progress needs paid compiler engineering.
AI-written summary · about
chapters (8)
- 0:00Intro
- 0:48Tool Chain Components (GNU)
- 2:17Tool Chain Components (Clang/LLVM)
- 4:32History
- 5:49Future Developments
- 10:32Today's Size Optimization Techniques
- 13:08Today's Speed Optimization Techniques
- 15:52Future Size Optimization Techniques
related talks
24mRISC V LLVM: Towards A Production-Ready LLVM-Based Toolchain
16mImprovements to RISC-V Vector code generation in LLVM
17mRISC-V LLVM State of the Union
28mLLVM for RISCV
17mMaking LLVM RISC-V Compiler More Performant for Everyone
18mPreview RISC-V GCC Features and Optimization Improvements for GCC-14
17mRecent Developments in Optimizing Compilers for RISC-V
08mPorting and Optimization V8 for RISC-V
