github_mirrors/vectorscan
3
0
Fork 1
mirror of https://github.com/VectorCamp/vectorscan.git synced 2026-01-02 22:54:43 +03:00
Code Issues Packages Projects Releases Wiki Activity

Speed up truffle with 256b TBL instructions

Browse Source 
256b wide SVE vectors allow some simplification of truffle.
Up to 40% speedup on graviton3. Going from 12500 MB/s to 17000 MB/s
onhe microbenchmark.
SVE2 also offer this capability for 128b vector with a speedup around
25% compared to normal SVE

Add unit tests and benchmark for this wide variant

Signed-off-by: Yoan Picchi <yoan.picchi@arm.com>
This commit is contained in:
Yoan Picchi
2024-04-23 12:04:40 +00:00
parent 154cb6333b
commit 7054378c93
23 changed files with 1125 additions and 78 deletions
Download Patch File Download Diff File Expand all files Collapse all files

11
src/nfa/accel.c
View File

@@ -142,9 +142,18 @@ const u8 *run_accel(const union AccelAux *accel, const u8 *c, const u8 *c_end) {
return c;
}
rv = truffleExec(accel->truffle.mask1, accel->truffle.mask2, c, c_end);
rv = truffleExec(accel->truffle.mask_lo, accel->truffle.mask_hi, c, c_end);
break;
#ifdef CAN_USE_WIDE_TRUFFLE
case ACCEL_TRUFFLE_WIDE:
DEBUG_PRINTF("accel Truffle Wide %p %p\n", c, c_end);
if (c + 15 >= c_end) {
return c;
}
rv = truffleExecWide(accel->truffle.mask, c, c_end);
break;
#endif
case ACCEL_DSHUFTI:
DEBUG_PRINTF("accel dshufti %p %p\n", c, c_end);
if (c + 15 + 1 >= c_end) {