FEX-Emu/xxHash
1
0
Fork 0
mirror of https://github.com/FEX-Emu/xxHash.git synced 2024-11-23 14:39:40 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity
857 Commits 18 Branches 23 Tags 4.1 MiB
9eb91a3b53
Commit Graph

857 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
easyaspi314 (Devin)
9eb91a3b53 Let the Great Typo Hunt commence! 
Work in progress.

 - Fix many spelling/grammar issues, primarily in comments
 - Remove most spaces before punctuation
 - Update XXH3 comment
 - Wrap most comments to 80 columns
 - Unify most comments to use the same style
 - Use hexadecimal in the xxhash spec
 - Update help messages to better match POSIX/GNU conventions
 - Use HTML escapes in README.md to avoid UTF-8
 - Mark outdated benchmark/scores
 2020-03-02 15:20:49 -05:00
Yann Collet
22abeae0eb
Merge pull request #318 from Cyan4973/collisionTester 
Collision tester
 2020-03-01 20:40:57 -08:00
Yann Collet
645975d8ba fixed minor typos 2020-03-01 18:42:42 -08:00
Yann Collet
a603671c92 attempt to fix clang compilation of collision test 2020-03-01 17:51:48 -08:00
Yann Collet
3021a28972 attempt to disable MT on MINGW32 through build macro 2020-03-01 16:37:52 -08:00
Yann Collet
0d97e52912 test adding stdio 2020-03-01 15:56:09 -08:00
Yann Collet
8072b3798a forgot the include 2020-03-01 15:42:31 -08:00
Yann Collet
7d05df56cc win version must be enforced 2020-03-01 15:05:40 -08:00
Yann Collet
cec4da8c1a minor change for mingw32 compilation 2020-03-01 14:49:38 -08:00
Yann Collet
9e8468459f updated license 
test tools are GPLv2
 2020-03-01 14:27:29 -08:00
Yann Collet
646226f697 removed artefacts 2020-03-01 14:13:40 -08:00
Yann Collet
e091a41d8a
Merge pull request #317 from Cyan4973/restoreChecks 
restored checksum validation for xxh3 & xxh128
 2020-03-01 14:03:10 -08:00
Yann Collet
6ce6105837 updated doc 2020-03-01 14:02:34 -08:00
Yann Collet
1740ae0d05
Merge pull request #316 from Cyan4973/doubleSecret 
Combined secret for small inputs
 2020-03-01 11:21:56 -08:00
Yann Collet
6f82266eca restored checksum validation for xxh3 & xxh128 
triggered when starting `xxhsum` in benchmark mode.
 2020-02-29 19:38:06 -08:00
Yann Collet
dd5d4b5526 update the SLAB5 pattern generator 2020-02-28 19:35:16 -08:00
Yann Collet
18725e9251 completed doc with a few examples 2020-02-28 19:11:40 -08:00
Yann Collet
7e7d65260b make partial bitfield collision tester compatible with 32-bit 2020-02-28 17:41:33 -08:00
Yann Collet
e618ec4dcc simplify integration of other hash algorithms 2020-02-28 17:07:12 -08:00
Yann Collet
dadcbc4a0f first commit of the brute-force collision tester 2020-02-28 16:28:43 -08:00
Yann Collet
5dd837e8a4
Merge pull request #315 from easyaspi314/misc_fixups 
Comments + 32-bit path for 9to16_128b, improve arch detection
 2020-02-28 15:19:00 -08:00
easyaspi314 (Devin)
4bc56e466d Update comments for 9to16_128b, add 32-bit path 
Show how the mask was simplified out on 64-bit, and add the more literal
32-bit friendly path which avoids some dependencies on the low bits of
m128.high64 and removes an ADC.

I separated the comments because the arithmetic is no longer pure
128-bit arithmetic.
 2020-02-28 17:03:19 -05:00
Yann Collet
1141783b29 extended secret combination to xxh128 2020-02-28 12:14:47 -08:00
Yann Collet
d63a8b3e16 fix minor warning (unused function) 
will be used once checksum verification is re-enabled
 2020-02-28 16:02:19 -08:00
Yann Collet
4e88e37d21 disabled checksum tests 
since all values have changed.

To be re-enabled later.
 2020-02-28 15:45:55 -08:00
Yann Collet
5f9256b302 extended strategy to len==0 2020-02-28 15:43:39 -08:00
Yann Collet
24a5b0add4 fix overlap 2020-02-28 15:36:03 -08:00
Yann Collet
c131e94ca7 fix ptr arithmetic error 2020-02-28 15:33:34 -08:00
Yann Collet
79813c9e1e Merge branch 'dev' into doubleSecret 2020-02-28 15:29:48 -08:00
Yann Collet
cd078d265d xxh3: use combined secrets for small inputs 
to make it more difficult to retrieve information about the secret.

Note : should cost nothing for usage with "default" `kSecret`
 2020-02-28 15:28:17 -08:00
Yann Collet
152d50900b xxh3 len[9-16] : minor quality improvement 
in O1 popcount distribution
 2020-02-28 11:37:13 -08:00
easyaspi314 (Devin)
3b12ce3102 Improve architecture detection 
- Properly detect MSVC x86
 - Make ARM more detailed - there are so many variants that change
   xxHash's performance that "arm" or "arm + NEON" is not specific
   enough.
 - Make sure x86_64 and aarch64 show SSE2 and NEON respectively.
 2020-02-28 13:39:27 -05:00
Yann Collet
002fcb2219
Merge pull request #314 from Cyan4973/xxh128_len9to16 
xxh128 len[9-16] : small simplification
 2020-02-28 10:12:28 -08:00
Yann Collet
cb929781f2 fix 12-bytes input quality issues 2020-02-27 17:25:52 -08:00
Yann Collet
2b8da00df1
Merge pull request #313 from easyaspi314/documentation 
Improve documentation of internals
 2020-02-27 15:54:01 -08:00
Yann Collet
c33a16dc43 selected final variant for len9to16 2020-02-27 15:51:52 -08:00
easyaspi314 (Devin)
e5883c4a33 Improve wording on XXH3_accumulate_512 comment 
Adds more details of the changes compared to original UMAC and reduces
run-ons.
 2020-02-27 18:48:16 -05:00
Yann Collet
c9b1b2320c simplified a bit more 
a still "good" MC2 distribution
 2020-02-27 15:26:43 -08:00
easyaspi314 (Devin)
f2a78adb2a Be consistent with wording 2020-02-27 18:10:34 -05:00
easyaspi314 (Devin)
9363917d08 Swap good and bad in 64-bit subset comment, expand 
Flows better.
 2020-02-27 18:07:48 -05:00
Yann Collet
91d43dbc7b simplified len9to16 
only insert seed from input_hi
also : slightly reduced multiplication budget

Overall mix is nonetheless better from an MC2 perspective.
 2020-02-27 15:04:23 -08:00
easyaspi314 (Devin)
2685b58257 Add a quick comment about the 64-bit arithmetic subset 
Added in the place for the Thumb sanity check because it made the most
sense.

Also noted that the requirements were not much more than XXH32.
 2020-02-27 17:53:26 -05:00
easyaspi314 (Devin)
ce652cc503 Note the primary reused subroutines 2020-02-27 17:21:55 -05:00
easyaspi314 (Devin)
1acf797a85 Move 17to128 and 129to240 into a more logical location 
Now, it is sorted by length from short to long.

Also, mention how mix32B is slower but better resists multiply by zero.
 2020-02-27 15:43:18 -05:00
easyaspi314 (Devin)
fb999b1960 Fix yet another typo 
Thanks, @aras-p
 2020-02-27 15:36:35 -05:00
easyaspi314 (Devin)
b655d2db93 Fix copy-paste issue 2020-02-27 14:43:30 -05:00
easyaspi314 (Devin)
39feadea2e Fix minor typi 2020-02-27 14:40:07 -05:00
easyaspi314 (Devin)
a1ca4ff0e2 Comment on FARSH keys and Mum variant, indent fix 2020-02-27 14:22:45 -05:00
easyaspi314 (Devin)
0ffbf28843 Add some extra details, fix typo 2020-02-27 14:13:57 -05:00
easyaspi314 (Devin)
bba53920a5 Mention the seed-dependent collisions in mix16B 
We know it exists, don't hide it.

It is highly unlikely to occur with proper seeding and random inputs,
and it doesn't occur on the 128-bit version, so make sure people are
aware of it.
 2020-02-27 14:00:54 -05:00