Zlib-ng 2.2.3 Rings In The New Year With ~17.8% Faster Inflate For AVX2

Zlib-ng 2.2.3 is out as the "next gen" Zlib replacement led by Hans Kristian Rosbach that retains a Zlib-compatible API while also offering a modernized API, modern C11 syntax, support for more CPU intrinsics, and other leading-edge features compared to upstream Zlib.

The Zlib-ng 2.2.3 project is out for the New Year as well as marking ten years of Zlib-ng being around for more aggressively shipping new performance optimizations atop Zlib.

With Zlib-ng 2.2.3 there is around 17.8% faster performance for inflate when running on x86_64 AMD/Intel processors supporting AVX2. The deflate performance is unchanged while the nearly 18% faster inflate performance is quite exciting and paired with a slightly smaller library size.


Zlib-ng 2.2.3 on AArch64 systems is also seeing around 2.3% faster inflate performance but again no change for deflate performance.

The Zlib-ng 2.2.3 release also has various build system fixes, improved pipelining for AVX-512 chunking, RISC-V fixes, and various other minor optimizations.

Downloads and more details on the Zlib-ng 2.2.3 release via GitHub .