mold

mold is a high-performance drop-in replacement for existing Unix linkers, designed to speed up builds. In benchmarks from August 2026, it links 4.9x faster than LLVM lld and 1.9x faster than wild at the median. It is written by the original developer of LLVM lld and has been in production use since 2021, serving as the default linker for many large open-source projects and companies. mold supports x86-64, i386, ARM 32/64, RISC-V 32/64, PowerPC 32/64, s390x, LoongArch 32/64, SPARC64, m68k, and SH-4. Its speed comes from pervasive parallelism and efficient data structures and algorithms, as detailed in the paper "mold: A Massively Parallel Linker" (ASPLOS 2027). Installation options include binary packages for various systems, prebuilt binaries attached to GitHub releases, and building from source with Cargo. Usage is straightforward: pass `-fuse-ld=mold` to Clang or recent GCC, use `-B` for older GCC, or use the `mold -run` feature to intercept linker invocations via LD_PRELOAD. It also integrates with Rust, Nim, Conan, and GitHub Actions. mold has been developed openly since 2020 with over 140 contributors. Its test suite covers every linker feature and runs in CI for all supported architectures, natively or under QEMU, and under ASAN and TSAN. Before each release, it builds all of Gentoo Linux's ~19,000 packages as a regression check. The project is funded by sponsors, including corporate and individual supporters.