powerpc: Optimize memrchr for power8

System Internals / glibc - Rajalakshmi Srinivasaraghavan [linux.vnet.ibm.com] - 2 October 2017 08:01 EDT

Vectorized loops are used for sizes greater than 32B to improve performance over power7 optimization. This shows as an average of 25% improvement depending on the position of search character. The performance is same for shorter strings.

59ba2d2 powerpc: Optimize memrchr for power8
ChangeLog | 12 +
sysdeps/powerpc/powerpc64/multiarch/Makefile | 5 +-
.../powerpc/powerpc64/multiarch/ifunc-impl-list.c | 3 +
.../powerpc/powerpc64/multiarch/memrchr-power8.S | 28 ++
.../powerpc/powerpc64/multiarch/memrchr-ppc64.c | 14 +-
sysdeps/powerpc/powerpc64/power8/memrchr.S | 321 +++++++++++++++++++++
6 files changed, 380 insertions(+), 3 deletions(-)

Upstream: sourceware.org


  • Share