1.8.0 - Add no_std and WASM support with full backward compatibilit

What's Changed

• Add no_std and WASM support with full backward compatibility by @loadingalias
• Use feature_detection instead of func calls for fusion by @onethumb
• Improve test and release coverage for x86 by @onethumb

Full Changelog: 1.7.1...1.8.0

1.7.1 - Ensure Miri passes on x86_64 and x86

What's Changed

• Ensure Miri passes on x86_64 and x86 by @onethumb

Full Changelog: 1.7.0...1.7.1

1.7.0 - Support Wasm targets, publish immutable GH releases, improve key generation docs

What's Changed

• Fix/no std feature (currently wasm compatible; groundwork for no_std) by @loadingalias
• Support and publish immutable releases on GitHub by @onethumb
• Improve key generation docs by @onethumb

Full Changelog: 1.6.0...1.7.0

1.6.0 - Improve feature detection, performance, and publish binary packages

• Improve runtime feature detection (and performance) by @onethumb
• remove libc by @usamoi
• Enable generating and publishing binary packages by @onethumb

Full changes

1.5.0 - Add static library, improve state management

• Improve state handling by @onethumb
• Add support for building a static library by @onethumb

Full changes

1.4.1 - Fix SIGILL on non-AVX512 systems

Fixes a SIGILL on systems which don't support AVX512.

• change unconditional x86-64-v4 reliance to the former x86-64-v2 reliance by @ultrabear

Full changes

1.4.0 - Doubles throughput on modern x86_64 systems to >100GiB/s

This release leverages the VPCLMULQDQ AVX512 support released in Rust 1.89.0, delivering a 2X throughput increase on modern x86_64 systems (Intel Ice Lake and later, AMD Zen4 and later).

• Enable VPCLMULQDQ support on Rust 1.89+ by @onethumb
• Support custom CRC parameters by @onethumb
• Add checksum command-line utility by @onethumb
• Remove bindgen by @yoshi-taka

Full changes

1.3.0 - Now in 100% pure Rust (and much faster!)

This release has replaced the C bindings with a 100% pure Rust implementation, making it easier to integrate with existing workflows. It's also significantly more performant on most architectures, especially on smaller input sizes.

• Replace C bindings for CRC32 fusion calculation by @onethumb
• Improve VPCLMULQDQ to use 512-bit wide registers by @onethumb
• Implement hardware accelerated XOR3 support by @onethumb

Full changes

1.2.2 - Remove println! from software fallback

• Remove println! from software fallback by @landonxjames

Full changes

1.2.1 - Limit FFI to supported architectures

• Limit FFI to supported architectures by @onethumb

Full changes