projects / platform / upstream / openblas.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 83de62c) | patch
POWER10: Improve axpy performance
authorRajalakshmi Srinivasaraghavan <rajis@linux.ibm.com>
Thu, 10 Dec 2020 17:51:42 +0000 (11:51 -0600)
committerRajalakshmi Srinivasaraghavan <rajis@linux.ibm.com>
Thu, 10 Dec 2020 17:51:42 +0000 (11:51 -0600)
commit346e30a46a4758eb4d9b8e5783c0b9c3c6b3ce6f
tree011abc9bf6af7f9eb4c0f88b9c386c2021eb121atree | snapshot
parent83de62c20db4c9fe9d70f49f138d2f6035855da0commit | diff
POWER10: Improve axpy performance

This patch aligns the stores to 32 byte boundary for saxpy and daxpy
before entering into vector pair loop. Fox caxpy, changed the store
instructions to stxv to improve performance of unaligned cases.
kernel/power/caxpy_microk_power10.c diff | blob | history
kernel/power/daxpy_power10.c diff | blob | history
kernel/power/saxpy_power10.c diff | blob | history
Domain: Machine Learning / ML Framework;