public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
* [PATCH/AARCH64] Improve ThunderX2 T99 address cost model
@ 2017-01-23 23:39 Andrew Pinski
  2017-01-23 23:59 ` Richard Earnshaw
  0 siblings, 1 reply; 2+ messages in thread
From: Andrew Pinski @ 2017-01-23 23:39 UTC (permalink / raw)
  To: GCC Patches

[-- Attachment #1: Type: text/plain, Size: 507 bytes --]

Hi,
  On ThunderX 2, the logical shifts that are part of the address cause
a 1 cycle extra and this is not modeled correctly.  For induction
variables we don't want to do the shift.  When I change the cost for
the shift of the addresses, I get a 12% improvement on HMMER; all over
benchmarks in SPEC CPU 2006 were neutral.

OK? Bootstrapped and tested on aarch64-linux-gnu with no regressions.

Thanks,
Andrew Pinski

ChangeLog:
* config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve cost table.

[-- Attachment #2: improve-t2addresscost.diff.txt --]
[-- Type: text/plain, Size: 516 bytes --]

Index: gcc/config/aarch64/aarch64.c
===================================================================
--- gcc/config/aarch64/aarch64.c	(revision 244839)
+++ gcc/config/aarch64/aarch64.c	(working copy)
@@ -273,9 +273,9 @@ static const struct cpu_addrcost_table q
 static const struct cpu_addrcost_table thunderx2t99_addrcost_table =
 {
     {
-      0, /* hi  */
-      0, /* si  */
-      0, /* di  */
+      1, /* hi  */
+      1, /* si  */
+      1, /* di  */
       2, /* ti  */
     },
   0, /* pre_modify  */

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2017-01-23 23:59 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2017-01-23 23:39 [PATCH/AARCH64] Improve ThunderX2 T99 address cost model Andrew Pinski
2017-01-23 23:59 ` Richard Earnshaw

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).