sha_256

Crates.iosha_256
lib.rssha_256
version1.0.1
sourcesrc
created_at2024-10-25 10:33:59.78671
updated_at2024-10-26 14:42:14.378613
descriptionOptimised sha-256
homepagehttps://github.com/goastler/sha_256
repositoryhttps://github.com/goastler/sha_256
max_upload_size
id1422323
size199,100
George Oastler (goastler)

documentation

README

Sha_256

A fast implementation of sha-256 in rust.

Features

  • Partially unrolled loops enhance efficiency by optimizing CPU cache utilization.

  • Bypasses Rust's safety checks to eliminate array index safety validations.

  • Utilizes only stack memory, avoiding dynamic memory allocation (malloc).

  • Reduces memory footprint through array reuse across multiple SHA-256 stages.

  • Prevents memory reallocation, allowing subsequent SHA-256 calls to reuse existing memory.

  • Optimized memory layout increases CPU cache hit rates.

  • Avoids unnecessary byte array conversions (e.g., u8a to u32a).

  • Written entirely in Rust, with no embedded assembly or specific CPU instructions.

  • No external dependencies.

  • Does not require the standard library (std).

Installation

In your project, run:

cargo add sha_256

Usage

Import the library

use sha_256::Sha256;

Create an instance of the sha256 struct.

let mut sha256: Sha256 = Sha256::new();

Create your message in bytes.

let bytes = &[0u8, 1u8, 2u8];

Run sha256 to create a digest/hash.

let hash: [u8; 32] = sha256.digest(bytes);

The general idea is "bytes in, bytes out". This is the most efficient input and output type to minimise conversions.

You will need to convert your input into bytes, e.g. string to bytes. See example project.

If you want the hash as a hex string you will need to convert it from bytes to hex afterwards. See example project.

Benchmark

How fast is this library? Up to 25% faster than the sha256 and sha. They contain use of Intel's SHA-NI cpu instructions (via a feature flag), whereas this library uses pure rust.

However, the above figures were obtained through some rough benchmarks on only my hardware. More thorough benchmarks are required, YMMV!

// TODO further benchmarks

Links

Commit count: 34

cargo fmt