Closed jalil-salame closed 8 months ago
AHash is about 8-9% faster than SipHash for our usecase (SipHash is the default hashing algorithm used by std::collections::HashMap).
std::collections::HashMap
In the serialization bench, it is about 20% faster.
Also added clippy lints to prevent introducing std::collections::HashMap back into the codebase.
This is in contrast to #13 which uses hashbrown::HashSet/Map. hashbrown is strictly faster on my Intel CPU.
hashbrown::HashSet/Map
hashbrown
This PR is for you to compare it yourself against #13 . IMO #13 is better.
Closed in favor of #13
AHash is about 8-9% faster than SipHash for our usecase (SipHash is the default hashing algorithm used by
std::collections::HashMap
).In the serialization bench, it is about 20% faster.
Also added clippy lints to prevent introducing
std::collections::HashMap
back into the codebase.This is in contrast to #13 which uses
hashbrown::HashSet/Map
.hashbrown
is strictly faster on my Intel CPU.This PR is for you to compare it yourself against #13 . IMO #13 is better.