[ was : Re: [openacc, testsuite, committed] Fix
libgomp.oacc-c-c++-common/loop-g-{1,2}.c for non-nvidia devices ]
On 09/28/2017 08:47 AM, Tom de Vries wrote:
Hi,
this patch makes the test-cases libgomp.oacc-c-c++-common/loop-g-{1,2}.c
work for non-nvidia devices.
For nvidia devices, a vector_length of 32 is required for the test to pass.
For devices with a non-32 forced vector_length, this test-case will fail
the test for excess errors due to:
...
warning: using vector_length (x), ignoring 32
...
Fixed by removing the explicit vector_length setting. For nvidia
devices, 32 is required, but that's also the forced default, so there's
no need to be explicit about it.
Committed as obvious.
Committed this similar patch.
Thanks,
- Tom
Fix libgomp.oacc-c-c++-common/{loop-red-g-1,routine-g-1}.c for non-nvidia devices
2017-10-05 Tom de Vries <t...@codesourcery.com>
* testsuite/libgomp.oacc-c-c++-common/loop-red-g-1.c (main): Remove
vector_length(32) clause from acc parallel directive.
* testsuite/libgomp.oacc-c-c++-common/routine-g-1.c (main): Same.
---
libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-g-1.c | 2 +-
libgomp/testsuite/libgomp.oacc-c-c++-common/routine-g-1.c | 2 +-
2 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-g-1.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-g-1.c
index d241d41..929e01c 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-g-1.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-g-1.c
@@ -11,7 +11,7 @@ int main ()
int ondev = 0;
int t = 0, h = 0;
-#pragma acc parallel num_gangs(32) vector_length(32) copy(ondev)
+#pragma acc parallel num_gangs(32) copy(ondev)
{
#pragma acc loop gang reduction (+:t)
for (unsigned ix = 0; ix < N; ix++)
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-g-1.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-g-1.c
index 9d14c3b..b6ab713 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-g-1.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-g-1.c
@@ -36,7 +36,7 @@ int main ()
for (ix = 0; ix < N;ix++)
ary[ix] = -1;
-#pragma acc parallel num_gangs(32) vector_length(32) copy(ary) copy(ondev)
+#pragma acc parallel num_gangs(32) copy(ary) copy(ondev)
{
ondev = __builtin_acc_on_device (5);
gang (ary);