[email protected]

lld/ELF with my parallel symbol interning patches, vs mold (2026-10-21, Rust), vs wild (2026-11-04), linking ELF clang-release, on Apple M4 Max.

lld 24.x optimization preview: parallel symbol interning. Linking clang-release at --threads=8: 181 -> 151 ms

lld linking Chromium (CREL, --icf=all): 3.46 -> 2.08 s. In chromium-debug, Moore's partition refinement algorithm performs poorly.

edgls is a C/C++ language server built on the EDG C/C++ front end. It is a fork of ccls with the clang-based parts replaced: the indexer, diagnostics, compile command handling, code completion, signature help, and formatting. github.com/MaskRay/edgls Verdict: Slower than clang

I made a typo earlier and wrote 'useful' instead of 'useless' regarding the 'testsuite must pass' requirement. The author agreed without question. Their other replies are almost direct copy pastes.

Extractive patches from a new contributor to llvm binary tuilities. "But it's Codex output the author doesn't really own, so reviewers end up supplying the understanding." llvm-project really should really adopt a ghostty-style vouch system.

Linking chrome-release (--gc-sections --export-dynamic, no ICF) with lld/mold/wild

Linking clang 21 with lld/mold/wild at 1-16 threads on Intel Core i7-14700K (8P+12E cores). Built around Sep 12 mold & wild lose ~30-40% when 2 threads share one P-core's SMT pair vs 2 distinct cores - real parallel linkers. ld.lld shows zero SMT sensitivity.

When code is this cheap to produce, saying no is the scarce resource. I don't think maintainers can absorb the volume that's coming....

The bottleneck moved from writing patches to converging on designs

github.com/llvm/llvm-project… ([Support] Add cgroup-aware thread pool strategies- #218478) 😿😿😿

有人用 nvcc --relocatable-device-code=true or AMD HIP -fgpu-rdc吗？whole-program compilation or separate compilation (LTO or non-LTO)?