mold

# mold: A Modern Linker

mold is a high-performance drop-in replacement for existing Unix linkers,
designed to speed up builds. In our August 2026 benchmarks, it links 4.9x
faster than [LLVM lld](https://lld.llvm.org) and 1.9x faster than
[wild](https://github.com/wild-linker/wild) at the median; see
[Benchmark](#benchmark) for the full results.

mold is written by the original developer of LLVM lld, the linker that
Android, Chrome, FreeBSD, PlayStation, Nintendo Switch, and other production
systems are built with. mold started as an effort to build an even faster
linker from scratch, free of the architectural limits its author had run into
while optimizing lld. It has been in production use since 2021, and today it
is the default linker of many large open-source projects and is used
internally by many 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.

## Why does linking speed matter?

If you are using a compiled language such as C, C++, or Rust, a build consists
of two phases. In the first phase, a compiler compiles source files into
object files (`.o` files). In the second phase, a linker takes all object
files and combines them into a single executable or shared library file.

The second phase can be time-consuming if your build output is large. mold can
speed up this process, saving you time and preventing distractions while
waiting for a lengthy build to finish. The difference is most noticeable
during rapid debug-edit-rebuild cycles.

## Benchmark

Here is a performance comparison of lld, wild, and mold when linking nine
large programs on two machines:

- AMD Ryzen Threadripper 7980X (64 cores) running Ubuntu 24.04
- Apple M1 Ultra (16 performance cores and 4 efficiency cores) running Fedora
  Asahi Remix 42; the benchmark is restricted to the performance cores

The Threadripper represents many-core workstation and server processors, and
the M1 Ultra represents high-performance desktop processors.

All three linkers were built from source in release configuration as of
2026-08-28 and ran with their default options. Times are wall-clock times as
seen by the caller, the median of three runs after a warm-up. The ARM64 rows
are the ARM64 builds of the same programs. The benchmark suite, including all
linker inputs, is available on [Zenodo](https://zenodo.org/records/21882261).

**AMD Ryzen Threadripper 7980X, debug builds**

| Program (output size)             | lld    | wild  | mold  | wild/mold | lld/mold
|-----------------------------------|--------|-------|-------|-----------|---------
| Blender 5.2 (2.46 GiB)            | 4.72s  | 1.98s | 0.86s | 2.3x      | 5.5x
| Chromium 145 (4.51 GiB)           | 16.64s | 3.98s | 1.65s | 2.4x      | 10.1x
| Chromium 145 ARM64 (4.76 GiB)     | 20.91s | N/A   | 1.83s | N/A       | 11.4x
| Clang 21 (4.19 GiB)               | 6.19s  | 3.71s | 1.34s | 2.8x      | 4.6x
| ClickHouse 26.1 (5.58 GiB)        | 6.62s  | 4.40s | 0.98s