# Performance

> How fast zstd is: measured numbers, what profiling found, and its benchmark history.

[magmalake](https://magmalake.dev/) [zstd](https://magmalake.dev/s/zstd)

###  Primitives

| Operation | magmalake | Reference |
| --- | --- | --- |
| zstd decompress FFI — this measures libzstd, not Mojo | 11.1–32.9 GB/s | the range spans compression levels and compressible versus random input |

scroll the table sideways →

Apple M4 — ten cores, four of them performance — macOS, stable Mojo 1.0.0, with nothing else building on the machine. Measured on 5 September 2026; the Iceberg rows on 15 September, the boundary rows on 19 September. [How these were measured](https://magmalake.dev/performance#method).

### Found by profiling

**A dlopen on every single decompress call.** About 450 µs per call, spent opening a library that was already open.

Every merge also runs its benchmarks on a GitHub runner: [the benchmark history](https://magmalake.github.io/zstd.mojo/benchmarks/). Read it for trends; a shared runner is slower and noisier than the table.

[A magmalake project](https://magmalake.dev/)

Apache 2.0 · © 2026 Marius Seritan