| commit | 4af6844aa5d3577e327f15dd877a38a043cb236a | |
| author | Noah Goldstein <goldstein.w.n@gmail.com> | |
| Wed, 19 Oct 2022 00:44:06 +0000 (18 17:44 -0700) | ||
| committer | Noah Goldstein <goldstein.w.n@gmail.com> | |
| Thu, 20 Oct 2022 00:31:03 +0000 (19 17:31 -0700) | ||
| tree | 8032676690c0b37aee3c26e5dde1bb600289ef4e | treesnapshot (tar.gz zip) |
| parent | b79f8ff26aa6151d2d2167afcddcd1ec46cfbc81 | commitdiff |
x86: Optimize memrchr-evex.S
Optimizations are:
1. Use the fact that lzcnt(0) -> VEC_SIZE for memchr to save a branch
in short string case.
2. Save several instructions in len = [VEC_SIZE, 4 * VEC_SIZE] case.
3. Use more code-size efficient instructions.
- tzcnt ... -> bsf ...
- vpcmpb $0 ... -> vpcmpeq ...
Code Size Changes:
memrchr-evex.S : -29 bytes
Net perf changes:
Reported as geometric mean of all improvements / regressions from N=10
runs of the benchtests. Value as New Time / Old Time so < 1.0 is
improvement and 1.0 is regression.
memrchr-evex.S : 0.949 (Mostly from improvements in small strings)
Full results attached in email.
Full check passes on x86-64.
Optimizations are:
1. Use the fact that lzcnt(0) -> VEC_SIZE for memchr to save a branch
in short string case.
2. Save several instructions in len = [VEC_SIZE, 4 * VEC_SIZE] case.
3. Use more code-size efficient instructions.
- tzcnt ... -> bsf ...
- vpcmpb $0 ... -> vpcmpeq ...
Code Size Changes:
memrchr-evex.S : -29 bytes
Net perf changes:
Reported as geometric mean of all improvements / regressions from N=10
runs of the benchtests. Value as New Time / Old Time so < 1.0 is
improvement and 1.0 is regression.
memrchr-evex.S : 0.949 (Mostly from improvements in small strings)
Full results attached in email.
Full check passes on x86-64.
| sysdeps/x86_64/multiarch/memrchr-evex.S | diffblobblamehistory |