github_mirrors/vectorscan
3
0
Fork 1
mirror of https://github.com/VectorCamp/vectorscan.git synced 2025-06-28 16:41:01 +03:00
Code Issues Packages Projects Releases Wiki Activity

noodle: use SSE palignr

Browse Source
This commit is contained in:
Matthew Barr 2016-04-21 15:39:47 +10:00
parent 75195f5f2e
commit fe8ffc5544
2 changed files with 5 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
src/hwlm/noodle_engine.c
View File

@ -1,5 +1,5 @@
/* /*
* Copyright (c) 2015, Intel Corporation * Copyright (c) 2015-2016, Intel Corporation
* *
* Redistribution and use in source and binary forms, with or without * Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions are met: * modification, are permitted provided that the following conditions are met:
@ -37,6 +37,7 @@
#include "util/compare.h" #include "util/compare.h"
#include "util/masked_move.h" #include "util/masked_move.h"
#include "util/simd_utils.h" #include "util/simd_utils.h"
#include "util/simd_utils_ssse3.h"
#include <ctype.h> #include <ctype.h>
#include <stdbool.h> #include <stdbool.h>

6
src/hwlm/noodle_engine_sse.c
View File

@ -1,5 +1,5 @@
/* /*
* Copyright (c) 2015, Intel Corporation * Copyright (c) 2015-2016, Intel Corporation
* *
* Redistribution and use in source and binary forms, with or without * Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions are met: * modification, are permitted provided that the following conditions are met:
@ -190,8 +190,8 @@ hwlm_error_t scanDoubleFast(const u8 *buf, size_t len, const u8 *key,
m128 v = noCase ? and128(load128(d), caseMask) : load128(d); m128 v = noCase ? and128(load128(d), caseMask) : load128(d);
m128 z1 = eq128(mask1, v); m128 z1 = eq128(mask1, v);
m128 z2 = eq128(mask2, v); m128 z2 = eq128(mask2, v);
u32 z = movemask128(and128(or128(lastz1, shiftLeft8Bits(z1)), z2)); u32 z = movemask128(and128(palignr(z1, lastz1, 15), z2));
lastz1 = _mm_srli_si128(z1, 15); lastz1 = z1;
// On large packet buffers, this prefetch appears to get us about 2%. // On large packet buffers, this prefetch appears to get us about 2%.
__builtin_prefetch(d + 128); __builtin_prefetch(d + 128);