rust-lang/hashbrown

hashbrown is a Rust port of Google's SwissTable hash map that serves as a drop-in replacement for the standard library's HashMap and HashSet types.

The project addresses the need for high-performance hash map implementations in Rust, particularly in environments where the standard library is unavailable. It implements Google's SwissTable algorithm, which uses a compact metadata layout with only one byte of overhead per entry instead of eight, and employs SIMD lookups to scan multiple hash entries in parallel. The implementation uses foldhash as its default hasher, which is significantly faster than SipHash, though it provides lower HashDoS resistance. The tool works in no_std environments when paired with the alloc crate and does not allocate memory for empty hash maps.

Developers should choose this crate if they need hash map performance in embedded systems, kernels, or other no_std contexts where the standard library is unavailable. It is also suitable for applications where the performance gains from SwissTable's design justify the trade-off of reduced HashDoS resistance compared to SipHash. The standard library has incorporated this implementation since Rust 1.36, so the algorithm's correctness and maturity are well-established; however, this crate remains relevant for no_std use cases and for those who want to customize the hasher or take advantage of features like rayon parallel iterator support.

Development activity shows consistent engagement with the codebase. The project maintains a changelog documenting updates. Multiple optional features are available, including nightly-only optimizations, serialization support via serde, parallel iteration through rayon, and allocator customization. The crate offers inline hints through an optional feature to improve runtime performance, and supports the equivalent trait for customized comparisons. A deprecated RawEntry API is available behind a feature flag for backward compatibility.