Rust port of N-R-K/ChibiHash. See the article ChibiHash: A small, fast 64-bit hash function for more information.
All credit for the algorithm goes to N-R-K.
Since crate version 0.4.0
the crate offers two versions of the algorithm:
v1
is the original implementation and the default one.v2
is a new implementation with better performance and passes all tests in smhasher3.
If you import the crate without any version specifier, the v1
version is used.
The v1
version can also be explicitly selected by importing chibihash::v1::*
instead.
If you want the latest and greatest version, you can import chibihash::v2::*
.
The v2
version will be the default in the next major version.
- 64-bit hash function
- Deterministic
- Fast
- No dependencies
no-std
compatible- Multiple ways to use ChibiHash:
- Direct Hashing: One-shot hashing using
chibi_hash64()
- Simple Hasher: Basic implementation using
ChibiHasher
(implementsstd::hash::Hasher
) - Streaming Hasher: Memory-efficient streaming with
StreamingChibiHasher
(implementsstd::hash::Hasher
) - BuildHasher:
ChibiHasher
implementsBuildHasher
. This allows using ChibiHash as the default hasher forstd::collections::HashMap
andstd::collections::HashSet
. UseChibiHashMap
andChibiHashSet
types.
- Direct Hashing: One-shot hashing using
use chibihash::{chibi_hash64, ChibiHasher, StreamingChibiHasher, ChibiHashMap, ChibiHashSet};
use std::hash::Hasher;
fn main() {
// Method 1: Direct hashing
let hash = chibi_hash64(b"yellow world", 42);
println!("Direct hash: {:016x}", hash);
// Method 2: Using Hasher trait
let mut hasher = ChibiHasher::new(42);
hasher.write(b"yellow world");
println!("Hasher trait: {:016x}", hasher.finish());
// Method 3: Streaming hashing
let mut hasher = StreamingChibiHasher::new(0);
hasher.update(b"yellow ");
hasher.update(b"world");
println!("Streaming: {:016x}", hasher.finalize());
// Method 4: BuildHasher for HashMap
let mut map: ChibiHashMap<String, i32> = ChibiHashMap::default();
map.insert("hello".to_string(), 42);
println!("BuildHasher HashMap: {:?}", map.get("hello"));
// Method 5: BuildHasher for HashSet
let mut set: ChibiHashSet<String> = ChibiHashSet::default();
set.insert("hello".to_string());
println!("BuildHasher HashSet: {}", set.contains("hello"));
}
Run cargo test
to see the tests.
Run cargo bench
to see the benchmarks. See target/criterion/report/index.html
for the HTML report.
The repository also contains a benchmark comparing the Rust implementation to the C implementation. Run cargo bench --features ffi
to see the benchmark. The C version can be found from the csrc
directory. The benchmark utilises FFI to call the C version.
Based on limited testing, the pure Rust implementation is faster than the C version when the input sizes are small (below 1024 bytes). With larger input sizes they are equal. Possibly due to the overhead of the FFI interface itself.
Copy-paste from the original repository. Same applies here.
Here are some reasons to avoid using this:
- For cryptographic purposes.
- For protecting against collision attacks (SipHash is the recommended one for this purpose).
- When you need very strong probability against collisions: ChibiHash does very minimal amount of mixing compared to other hashes (e.g xxhash64). And so chances of collision should in theory be higher.
MIT. The original C version is under the Unlicense.