]> git.ipfire.org Git - thirdparty/glibc.git/commit
projects / thirdparty / glibc.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 20a2a75) | patch
AArch64: Optimise SVE scalar callbacks
authorJoe Ramsay <Joe.Ramsay@arm.com>
Thu, 6 Nov 2025 15:36:03 +0000 (15:36 +0000)
committerWilco Dijkstra <wilco.dijkstra@arm.com>
Thu, 6 Nov 2025 15:45:37 +0000 (15:45 +0000)
commit5b82fb18827e962af9f080fdf3c1a69802783f67
treeea1a7fbff42a91a7db817e304b1932631b250c7ftree | snapshot
parent20a2a756089eacd7e7f4c02e381e82b5d0e40a2ccommit | diff
AArch64: Optimise SVE scalar callbacks

Instead of using SVE instructions to marshall special results into the
correct lane, just write the entire vector (and the predicate) to
memory, then use cheaper scalar operations.

Geomean speedup of 16% in special intervals on Neoverse with GCC 14.

Reviewed-by: Wilco Dijkstra <Wilco.Dijkstra@arm.com>
sysdeps/aarch64/fpu/sv_math.h diff | blob | blame | history
A mirror of the official glibc repository