This crate is a Rust port of Google's high-performance SwissTable hash
map, adapted to make it a drop-in replacement for Rust's standard HashMap
and HashSet
types.
The original C++ version of SwissTable can be found here, and this CppCon talk gives an overview of how the algorithm works.
Since Rust 1.36, this is now the HashMap
implementation for the Rust standard
library. However you may still want to use this crate instead since it works
in environments without std
, such as embedded systems and kernels.
- Drop-in replacement for the standard library
HashMap
andHashSet
types. - Uses
AHash
as the default hasher, which is much faster than SipHash. - Around 2x faster than the previous standard library
HashMap
. - Lower memory usage: only 1 byte of overhead per entry instead of 8.
- Compatible with
#[no_std]
(but requires a global allocator with thealloc
crate). - Empty hash maps do not allocate any memory.
- SIMD lookups to scan multiple hash entries in parallel.
Compared to the previous implementation of std::collections::HashMap
(Rust 1.35).
With the hashbrown default AHash hasher (not HashDoS-resistant):
name oldstdhash ns/iter hashbrown ns/iter diff ns/iter diff % speedup
insert_ahash_highbits 20,846 7,397 -13,449 -64.52% x 2.82
insert_ahash_random 20,515 7,796 -12,719 -62.00% x 2.63
insert_ahash_serial 21,668 7,264 -14,404 -66.48% x 2.98
insert_erase_ahash_highbits 29,570 17,498 -12,072 -40.83% x 1.69
insert_erase_ahash_random 39,569 17,474 -22,095 -55.84% x 2.26
insert_erase_ahash_serial 32,073 17,332 -14,741 -45.96% x 1.85
iter_ahash_highbits 1,572 2,087 515 32.76% x 0.75
iter_ahash_random 1,609 2,074 465 28.90% x 0.78
iter_ahash_serial 2,293 2,120 -173 -7.54% x 1.08
lookup_ahash_highbits 3,460 4,403 943 27.25% x 0.79
lookup_ahash_random 6,377 3,911 -2,466 -38.67% x 1.63
lookup_ahash_serial 3,629 3,586 -43 -1.18% x 1.01
lookup_fail_ahash_highbits 5,286 3,411 -1,875 -35.47% x 1.55
lookup_fail_ahash_random 12,365 4,171 -8,194 -66.27% x 2.96
lookup_fail_ahash_serial 4,902 3,240 -1,662 -33.90% x 1.51
With the libstd default SipHash hasher (HashDoS-resistant):
name oldstdhash ns/iter hashbrown ns/iter diff ns/iter diff % speedup
insert_std_highbits 32,598 20,199 -12,399 -38.04% x 1.61
insert_std_random 29,824 20,760 -9,064 -30.39% x 1.44
insert_std_serial 33,151 17,256 -15,895 -47.95% x 1.92
insert_erase_std_highbits 74,731 48,735 -25,996 -34.79% x 1.53
insert_erase_std_random 73,828 47,649 -26,179 -35.46% x 1.55
insert_erase_std_serial 73,864 40,147 -33,717 -45.65% x 1.84
iter_std_highbits 1,518 2,264 746 49.14% x 0.67
iter_std_random 1,502 2,414 912 60.72% x 0.62
iter_std_serial 6,361 2,118 -4,243 -66.70% x 3.00
lookup_std_highbits 21,705 16,962 -4,743 -21.85% x 1.28
lookup_std_random 21,654 17,158 -4,496 -20.76% x 1.26
lookup_std_serial 18,726 14,509 -4,217 -22.52% x 1.29
lookup_fail_std_highbits 25,852 17,323 -8,529 -32.99% x 1.49
lookup_fail_std_random 25,913 17,760 -8,153 -31.46% x 1.46
lookup_fail_std_serial 22,648 14,839 -7,809 -34.48% x 1.53
Add this to your Cargo.toml
:
[dependencies]
hashbrown = "0.9"
Then:
use hashbrown::HashMap;
let mut map = HashMap::new();
map.insert(1, "one");
This crate has the following Cargo features:
nightly
: Enables nightly-only features:#[may_dangle]
.serde
: Enables serde serialization support.rayon
: Enables rayon parallel iterator support.raw
: Enables access to the experimental and unsafeRawTable
API.inline-more
: Adds inline hints to most functions, improving run-time performance at the cost of compilation time. (enabled by default)ahash
: Compiles with ahash as default hasher. (enabled by default)ahash-compile-time-rng
: Activates thecompile-time-rng
feature of ahash, to increase the DOS-resistance, but can result in issues forno_std
builds. More details in issue#124. (enabled by default)
Licensed under either of:
- Apache License, Version 2.0, (LICENSE-APACHE or http://www.apache.org/licenses/LICENSE-2.0)
- MIT license (LICENSE-MIT or http://opensource.org/licenses/MIT)
at your option.
Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in the work by you, as defined in the Apache-2.0 license, shall be dual licensed as above, without any additional terms or conditions.