Zstd 1.5.4 Released With Many Performance Improvements


After being in development for nearly one year , Zstd 1.5.4 has been released with many «significant performance improvements across multiple scenarios» for this increasingly-used data compression algorithm.

Zstandard 1.5.4 features a lot of performance optimizations/tuning that «cumulatively benefit a good portion of installed base in one way or another.» There are ARM optimizations and more NEON work, optimization work around large files, I/O performance optimizations, and a range of other performance work that together adds up quite nicely for making Zstd 1.5.4 an exciting release.

Zstd 1.5.4 performance changes
A lot of performance optimization work can be found in Zstd 1.5.4.

Zstd 1.5.4 also adds support for externally-defined sequence producers. This API allows users to provide their own custom sequence producer that the Zstd library invokes to process each blocks. The release notes mention that this will end up allowing hardware-accelerated match-finders such as for being able to support Intel QuickAssist with Xeon Scalable «Sapphire Rapids» with Zstd. This will be quite fun and look forward to learning more / seeing what comes of it…

Downloads and more details on the Zstandard 1.5.4 feature update via GitHub. Stay tuned for some fresh Zstd benchmarks off this new version with various hardware.