aarch64: Adjust SVE vector costs for NVIDIA Olympus

After some more analysis, I'd like to adjust the SVE load
cost for NVIDIA Olympus to more accurately reflect the core SWOG.
This leads to more sensible Advanced SIMD vs SVE autovec decisions.

Bootstrapped and tested on aarch64-none-linux-gnu.

Signed-off-by: Kyrylo Tkachov <ktkachov@nvidia.com>
gcc/ChangeLog

	* config/aarch64/tuning_models/olympus.h (olympus_sve_vector_cost):
	Set align_load_cost to 6.

diff --git a/gcc/config/aarch64/tuning_models/olympus.h b/gcc/config/aarch64/tuning_models/olympus.h
index 404d79307dfa5b6f91b384029f0edddce9015675..29b87a40f68f548c9c0f0d523514992142b7d9bf 100644 (file)
--- a/gcc/config/aarch64/tuning_models/olympus.h
+++ b/gcc/config/aarch64/tuning_models/olympus.h
@@ -77,7 +77,7 @@ static sve_vec_cost olympus_sve_vector_cost =
     2, /* store_elt_extra_cost  */
     8, /* vec_to_scalar_cost  */
     4, /* scalar_to_vec_cost  */
-    4, /* align_load_cost  */
+    6, /* align_load_cost  */
     6, /* unalign_load_cost  */
     1, /* unalign_store_cost  */
     1  /* store_cost  */