v2.6.0-alpha | Experimental Native Rust

[ronhombre]

OPTIMIZED/MASTER VERSION

Changelog:

• Built-in first-class Rust support with KeccakRust and neo-rust-gradle-plugin
• Add new isExtendable() API to KeccakParameter
• Add WASM support
• Add new KAT tests

Full Changelog: v2.5.1...v2.6.0-alpha

The attached bundle below has binaries for androidnativearm32, androidnativearm64, androidnativex64, js, jvm, linuxarm64, linuxx64, mingwx64, and wasi. No binaries available for iOS platform at the moment.

These are all signed so they should be plug-and-play for most people. Let me know if you encounter any issues or enjoyed the speed boosts. Don't be shy to participate in the discussion

---

This discussion was created from the release v2.6.0-alpha | Experimental Native Rust.

[ronhombre]

Throughput comparison JVM vs mingwx64 (Rust) vs mingwx64 (Kotlin)

128-Bit XOFs (168B Rate — Avg across RAWSHAKE, SHAKE, CSHAKE, KMACXOF)
JVM (Pure Kotlin)          | ███ 3,102.09 ms (330.1 MiB/s)
mingwx64 (KeccakRust)      | ███████ 7,219.66 ms (141.8 MiB/s)  [7.59x faster than Pure Native!]
mingwx64 (Pure Kotlin)     | ███████████████████████████████████████████████████████ 54,818.59 ms (18.7 MiB/s)

256-Bit XOFs (136B Rate — Avg across RAWSHAKE, SHAKE, CSHAKE, KMACXOF)
JVM (Pure Kotlin)          | ████ 3,685.16 ms (277.9 MiB/s)
mingwx64 (KeccakRust)      | ████████ 8,490.58 ms (120.6 MiB/s)  [7.99x faster than Pure Native!]
mingwx64 (Pure Kotlin)     | ████████████████████████████████████████████████████████████████████ 67,863.34 ms (15.1 MiB/s)
                             0s        10s       20s       30s       40s       50s       60s       70s

I tried getting more performance boosts but this is the best it can get according to my experiments. The FFI boundary layer started affecting performance, and as it turned out, my Kotlin implementation was already very efficient so trying to conform everything to allow more data computed in Rust simply causes more slow downs for the other targets.