projects / gcc.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: c286fb4) | patch
[AArch64] PR79262: Adjust vector cost
authorWilco Dijkstra <wdijkstr@arm.com>
Tue, 19 Nov 2019 15:57:54 +0000 (15:57 +0000)
committerWilco Dijkstra <wilco@gcc.gnu.org>
Tue, 19 Nov 2019 15:57:54 +0000 (15:57 +0000)
commit4bf29d15f2e01348a45a1f4e1a135962f123fdd6
tree86fc900066f70f2a6fb97c69023f1f611a1db523tree
parentc286fb4ed5706cb4de40da0eb798e31bb16eb05bcommit | diff
[AArch64] PR79262: Adjust vector cost

PR79262 has been fixed for almost all AArch64 cpus, however the example is still
vectorized in a few cases, resulting in lower performance.  Adjust the vector
cost slightly so that so that -mcpu=cortex-a53 now has identical performance as
-mcpu=cortex-a57 on libquantum.

    gcc/
PR target/79262
* config/aarch64/aarch64.c (generic_vector_cost): Adjust
vec_to_scalar_cost.

From-SVN: r278452
gcc/ChangeLog diff | blob | history
gcc/config/aarch64/aarch64.c diff | blob | history