Patchwork [AARCH64/Committed] Fix gcc.target/aarch64/vect-fmaxv-fminv-compile.c

login
register
mail settings
Submitter Andrew Pinski
Date Nov. 7, 2012, 12:18 a.m.
Message ID <CA+=Sn1=M6PBgbTB8cM_5Dev40=_TBe7XiM6hdG1z4sXyu66Ctw@mail.gmail.com>
Download mbox | patch
Permalink /patch/197604/
State New
Headers show

Comments

Andrew Pinski - Nov. 7, 2012, 12:18 a.m.
Hi,
  The problem here is the vector cost model causes some of the
functions not be vectorized even though we are scanning for the
explicit vectorization.  This fixes the testcase by turning off the
cost model.

Thanks,
Andrew Pinski

2012-11-06  Andrew Pinski  <apinski@cavium.com>

	* gcc.target/aarch64/vect-fmaxv-fminv-compile.c: Add -fno-vect-cost-model.

Patch

Index: gcc.target/aarch64/vect-fmaxv-fminv-compile.c
===================================================================
--- gcc.target/aarch64/vect-fmaxv-fminv-compile.c	(revision 193259)
+++ gcc.target/aarch64/vect-fmaxv-fminv-compile.c	(working copy)
@@ -1,6 +1,6 @@ 
 
 /* { dg-do compile } */
-/* { dg-options "-O3 -ffast-math" } */
+/* { dg-options "-O3 -ffast-math -fno-vect-cost-model" } */
 
 #include "vect-fmaxv-fminv.x"
 
Index: ChangeLog
===================================================================
--- ChangeLog	(revision 193272)
+++ ChangeLog	(working copy)
@@ -1,3 +1,7 @@ 
+2012-11-06  Andrew Pinski  <apinski@cavium.com>
+
+	* gcc.target/aarch64/vect-fmaxv-fminv-compile.c: Add -fno-vect-cost-model.
+
 2012-11-06  Aldy Hernandez  <aldyh@redhat.com>
 
 	* c-c++-common/tm/trxn-expr-3.c: Adjust for uninstrumented code path.