diff options
| author | Noah Goldstein <goldstein.w.n@gmail.com> | 2024-05-24 12:38:51 -0500 |
|---|---|---|
| committer | Noah Goldstein <goldstein.w.n@gmail.com> | 2024-05-30 12:36:09 -0500 |
| commit | 46b5e98ef6f1b9f4b53851f152ecb8209064b26c (patch) | |
| tree | f3008f5bbe51de418ade2aeb13363408c7ba2536 /sysdeps/x86/dl-tunables.list | |
| parent | 5bf0ab80573d66e4ae5d94b094659094336da90f (diff) | |
| download | glibc-46b5e98ef6f1b9f4b53851f152ecb8209064b26c.tar.gz glibc-46b5e98ef6f1b9f4b53851f152ecb8209064b26c.tar.xz glibc-46b5e98ef6f1b9f4b53851f152ecb8209064b26c.zip | |
x86: Add seperate non-temporal tunable for memset
The tuning for non-temporal stores for memset vs memcpy is not always the same. This includes both the exact value and whether non-temporal stores are profitable at all for a given arch. This patch add `x86_memset_non_temporal_threshold`. Currently we disable non-temporal stores for non Intel vendors as the only benchmarks showing its benefit have been on Intel hardware. Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
Diffstat (limited to 'sysdeps/x86/dl-tunables.list')
| -rw-r--r-- | sysdeps/x86/dl-tunables.list | 3 |
1 files changed, 3 insertions, 0 deletions
diff --git a/sysdeps/x86/dl-tunables.list b/sysdeps/x86/dl-tunables.list index 7d82da0dec..a0a1299592 100644 --- a/sysdeps/x86/dl-tunables.list +++ b/sysdeps/x86/dl-tunables.list @@ -30,6 +30,9 @@ glibc { x86_non_temporal_threshold { type: SIZE_T } + x86_memset_non_temporal_threshold { + type: SIZE_T + } x86_rep_movsb_threshold { type: SIZE_T # Since there is overhead to set up REP MOVSB operation, REP |