tag | 561aa2fb6a1556174a12c4024aa06e326db684a4 | |
---|---|---|
tagger | The Android Open Source Project <initial-contribution@android.com> | Wed Apr 12 14:25:06 2023 -0700 |
object | 9944fad6a715f9abbe69e97a00e3e2dda3ae7196 |
Android U Beta 1 GPL
commit | 9944fad6a715f9abbe69e97a00e3e2dda3ae7196 | [log] [tgz] |
---|---|---|
author | Jeff Vander Stoep <jeffv@google.com> | Tue Jan 31 18:50:10 2023 +0000 |
committer | Automerger Merge Worker <android-build-automerger-merge-worker@system.gserviceaccount.com> | Tue Jan 31 18:50:10 2023 +0000 |
tree | 7e50cf6536dab87cd690e673add78339732d1e11 | |
parent | c88f6fc5d85d651497ee7bb0105067e347566265 [diff] | |
parent | 96133be2492e133dae51ca98df7916369b4ce14c [diff] |
Update TEST_MAPPING am: 96133be249 Original change: https://android-review.googlesource.com/c/platform/external/rust/crates/fxhash/+/2411798 Change-Id: I3c9b065918db218c60ae145e6485f80ca2983ce8 Signed-off-by: Automerger Merge Worker <android-build-automerger-merge-worker@system.gserviceaccount.com>
This hashing algorithm was extracted from the Rustc compiler. This is the same hashing algoirthm used for some internal operations in FireFox. The strength of this algorithm is in hashing 8 bytes at a time on 64-bit platforms, where the FNV algorithm works on one byte at a time.
It is not a cryptographically secure hash, so it is strongly recommended that you do not use this hash for cryptographic purproses. Furthermore, this hashing algorithm was not designed to prevent any attacks for determining collisions which could be used to potentially cause quadratic behavior in HashMap
s. So it is not recommended to expose this hash in places where collissions or DDOS attacks may be a concern.
Building an Fx backed hashmap.
extern crate fxhash; use fxhash::FxHashMap; let mut hashmap = FxHashMap::new(); hashmap.insert("black", 0); hashmap.insert("white", 255);
Building an Fx backed hashset.
extern crate fxhash; use fxhash::FxHashSet; let mut hashmap = FxHashSet::new(); hashmap.insert("black"); hashmap.insert("white");
Generally fxhash
is than fnv
on u32
, u64
, or any byte sequence with length >= 5. However, keep in mind that hashing speed is not the only characteristic worth considering. That being said, Rustc had an observable increase in speed when switching from fnv
backed hashmaps to fx
based hashmaps.
bench_fnv_003 ... bench: 3 ns/iter (+/- 0) bench_fnv_004 ... bench: 2 ns/iter (+/- 0) bench_fnv_011 ... bench: 6 ns/iter (+/- 1) bench_fnv_012 ... bench: 5 ns/iter (+/- 1) bench_fnv_023 ... bench: 14 ns/iter (+/- 3) bench_fnv_024 ... bench: 14 ns/iter (+/- 4) bench_fnv_068 ... bench: 57 ns/iter (+/- 11) bench_fnv_132 ... bench: 145 ns/iter (+/- 30) bench_fx_003 ... bench: 4 ns/iter (+/- 0) bench_fx_004 ... bench: 3 ns/iter (+/- 1) bench_fx_011 ... bench: 5 ns/iter (+/- 2) bench_fx_012 ... bench: 4 ns/iter (+/- 1) bench_fx_023 ... bench: 7 ns/iter (+/- 3) bench_fx_024 ... bench: 4 ns/iter (+/- 1) bench_fx_068 ... bench: 10 ns/iter (+/- 3) bench_fx_132 ... bench: 19 ns/iter (+/- 5) bench_seahash_003 ... bench: 30 ns/iter (+/- 12) bench_seahash_004 ... bench: 32 ns/iter (+/- 22) bench_seahash_011 ... bench: 30 ns/iter (+/- 4) bench_seahash_012 ... bench: 31 ns/iter (+/- 1) bench_seahash_023 ... bench: 32 ns/iter (+/- 6) bench_seahash_024 ... bench: 31 ns/iter (+/- 5) bench_seahash_068 ... bench: 40 ns/iter (+/- 9) bench_seahash_132 ... bench: 50 ns/iter (+/- 12)