| Message ID | 78fa9345-97af-d0e0-954f-6ed9925884cf@linux.ibm.com |
|---|---|
| State | New |
| Headers | show |
| Series | [testsuite] Enable vect_usad_char effective target for PowerPC and fix up SAD_EXPR testcases | expand |
On Tue, 19 Feb 2019, Pat Haugen wrote: > Power9 added support for V16QImode SAD operations. While making the check_effective_target change I noticed that the tests will also pass on Power7/Power8 even though they don't have the optab support. The reason is the tests are only checking that the source pattern is recognized, not that a SAD_EXPR was actually generated. So I've updated the tests also. > > Ok for trunk? OK. Richard. > -Pat > > > testsuite/ChangeLog: > 2019-02-19 Pat Haugen <pthaugen@us.ibm.com> > > * lib/target-supports.exp (check_effective_target_vect_usad_char): > Add PowerPC support. > * gcc.dg/vect/slp-reduc-sad.c: Update scan string. > * gcc.dg/vect/vect-reduc-sad.c: Likewise. > > > Index: gcc/testsuite/lib/target-supports.exp > =================================================================== > --- gcc/testsuite/lib/target-supports.exp (revision 268784) > +++ gcc/testsuite/lib/target-supports.exp (working copy) > @@ -5982,7 +5982,9 @@ proc check_effective_target_vect_usad_ch > expr { [istarget i?86-*-*] > || [istarget x86_64-*-*] > || ([istarget aarch64*-*-*] > - && ![check_effective_target_aarch64_sve])}}] > + && ![check_effective_target_aarch64_sve]) > + || ([istarget powerpc*-*-*] > + && [check_p9vector_hw_available])}}] > } > > # Return 1 if the target plus current options supports both signed > Index: gcc/testsuite/gcc.dg/vect/slp-reduc-sad.c > =================================================================== > --- gcc/testsuite/gcc.dg/vect/slp-reduc-sad.c (revision 268784) > +++ gcc/testsuite/gcc.dg/vect/slp-reduc-sad.c (working copy) > @@ -58,6 +58,6 @@ main () > return 0; > } > > -/* { dg-final { scan-tree-dump "vect_recog_sad_pattern: detected" "vect" } } */ > +/* { dg-final { scan-tree-dump "sad pattern recognized" "vect" } } */ > /* { dg-final { scan-tree-dump "vectorizing stmts using SLP" "vect" } } */ > /* { dg-final { scan-tree-dump-times "vectorized 1 loops" 1 "vect" } } */ > Index: gcc/testsuite/gcc.dg/vect/vect-reduc-sad.c > =================================================================== > --- gcc/testsuite/gcc.dg/vect/vect-reduc-sad.c (revision 268784) > +++ gcc/testsuite/gcc.dg/vect/vect-reduc-sad.c (working copy) > @@ -49,6 +49,6 @@ main (void) > return 0; > } > > -/* { dg-final { scan-tree-dump-times "vect_recog_sad_pattern: detected" 1 "vect" } } */ > +/* { dg-final { scan-tree-dump-times "sad pattern recognized" 1 "vect" } } */ > /* { dg-final { scan-tree-dump-times "vectorized 1 loops" 1 "vect" } } */ > >
On 2/19/19 2:07 PM, Pat Haugen wrote: > Power9 added support for V16QImode SAD operations. While making the check_effective_target change I noticed that the tests will also pass on Power7/Power8 even though they don't have the optab support. The reason is the tests are only checking that the source pattern is recognized, not that a SAD_EXPR was actually generated. So I've updated the tests also. > > Ok for trunk? > > -Pat > > > testsuite/ChangeLog: > 2019-02-19 Pat Haugen <pthaugen@us.ibm.com> > > * lib/target-supports.exp (check_effective_target_vect_usad_char): > Add PowerPC support. > * gcc.dg/vect/slp-reduc-sad.c: Update scan string. > * gcc.dg/vect/vect-reduc-sad.c: Likewise. OK jeff
Index: gcc/testsuite/lib/target-supports.exp =================================================================== --- gcc/testsuite/lib/target-supports.exp (revision 268784) +++ gcc/testsuite/lib/target-supports.exp (working copy) @@ -5982,7 +5982,9 @@ proc check_effective_target_vect_usad_ch expr { [istarget i?86-*-*] || [istarget x86_64-*-*] || ([istarget aarch64*-*-*] - && ![check_effective_target_aarch64_sve])}}] + && ![check_effective_target_aarch64_sve]) + || ([istarget powerpc*-*-*] + && [check_p9vector_hw_available])}}] } # Return 1 if the target plus current options supports both signed Index: gcc/testsuite/gcc.dg/vect/slp-reduc-sad.c =================================================================== --- gcc/testsuite/gcc.dg/vect/slp-reduc-sad.c (revision 268784) +++ gcc/testsuite/gcc.dg/vect/slp-reduc-sad.c (working copy) @@ -58,6 +58,6 @@ main () return 0; } -/* { dg-final { scan-tree-dump "vect_recog_sad_pattern: detected" "vect" } } */ +/* { dg-final { scan-tree-dump "sad pattern recognized" "vect" } } */ /* { dg-final { scan-tree-dump "vectorizing stmts using SLP" "vect" } } */ /* { dg-final { scan-tree-dump-times "vectorized 1 loops" 1 "vect" } } */ Index: gcc/testsuite/gcc.dg/vect/vect-reduc-sad.c =================================================================== --- gcc/testsuite/gcc.dg/vect/vect-reduc-sad.c (revision 268784) +++ gcc/testsuite/gcc.dg/vect/vect-reduc-sad.c (working copy) @@ -49,6 +49,6 @@ main (void) return 0; } -/* { dg-final { scan-tree-dump-times "vect_recog_sad_pattern: detected" 1 "vect" } } */ +/* { dg-final { scan-tree-dump-times "sad pattern recognized" 1 "vect" } } */ /* { dg-final { scan-tree-dump-times "vectorized 1 loops" 1 "vect" } } */