@@ -1,3 +1,52 @@
+2016-03-30 Thomas Schwinge <thomas@codesourcery.com>
+ Julian Brown <julian@codesourcery.com>
+ Chung-Lin Tang <cltang@codesourcery.com>
+ Cesar Philippidis <cesar@codesourcery.com>
+ James Norris <jnorris@codesourcery.com>
+ Tom de Vries <tom@codesourcery.com>
+ Nathan Sidwell <nathan@codesourcery.com>
+
+ * c-c++-common/goacc/combined-directives.c: Clean up dg-*
+ directives.
+ * c-c++-common/goacc/loop-clauses.c: Likewise.
+ * g++.dg/goacc/template.C: Likewise.
+ * gfortran.dg/goacc/combined-directives.f90: Likewise.
+ * gfortran.dg/goacc/loop-1.f95: Likewise.
+ * gfortran.dg/goacc/loop-5.f95: Likewise.
+ * gfortran.dg/goacc/loop-6.f95: Likewise.
+ * gfortran.dg/goacc/loop-tree-1.f90: Likewise.
+ * c-c++-common/goacc-gomp/nesting-1.c: Update.
+ * c-c++-common/goacc-gomp/nesting-fail-1.c: Likewise.
+ * c-c++-common/goacc/clauses-fail.c: Likewise.
+ * c-c++-common/goacc/parallel-1.c: Likewise.
+ * c-c++-common/goacc/reduction-1.c: Likewise.
+ * c-c++-common/goacc/reduction-2.c: Likewise.
+ * c-c++-common/goacc/reduction-3.c: Likewise.
+ * c-c++-common/goacc/reduction-4.c: Likewise.
+ * c-c++-common/goacc/routine-3.c: Likewise.
+ * c-c++-common/goacc/routine-4.c: Likewise.
+ * c-c++-common/goacc/routine-5.c: Likewise.
+ * c-c++-common/goacc/tile.c: Likewise.
+ * g++.dg/goacc/template.C: Likewise.
+ * gfortran.dg/goacc/combined-directives.f90: Likewise.
+ * c-c++-common/goacc/nesting-1.c: Move dg-error test cases into...
+ * c-c++-common/goacc/nesting-fail-1.c: ... this file. Update.
+ * c-c++-common/goacc/kernels-1.c: Update. Incorporate...
+ * c-c++-common/goacc/kernels-empty.c: ... this file, and...
+ * c-c++-common/goacc/kernels-eternal.c: ... this file, and...
+ * c-c++-common/goacc/kernels-noreturn.c: ... this file.
+ * c-c++-common/goacc/host_data-1.c: New file. Incorporate...
+ * c-c++-common/goacc/use_device-1.c: ... this file.
+ * c-c++-common/goacc/host_data-2.c: New file. Incorporate...
+ * c-c++-common/goacc/host_data-5.c: ... this file, and...
+ * c-c++-common/goacc/host_data-6.c: ... this file.
+ * c-c++-common/goacc/loop-2-kernels.c: New file.
+ * c-c++-common/goacc/loop-2-parallel.c: Likewise.
+ * c-c++-common/goacc/loop-3.c: Likewise.
+ * g++.dg/goacc/reference.C: Likewise.
+ * g++.dg/goacc/routine-1.C: Likewise.
+ * g++.dg/goacc/routine-2.C: Likewise.
+
2016-03-30 Richard Biener <rguenther@suse.de>
PR middle-end/70450
@@ -1,3 +1,51 @@
+2016-04-04 Thomas Schwinge <thomas@codesourcery.com>
+
+ * c-c++-common/goacc/firstprivate.c: Remove file, moving its
+ content into...
+ * c-c++-common/goacc/clauses-fail.c: ... this file, and...
+ * c-c++-common/goacc/parallel-1.c: ... this file.
+ * c-c++-common/goacc/host_data-3.c: Remove file.
+ * c-c++-common/goacc/host_data-4.c: Remove file, moving its
+ content into...
+ * c-c++-common/goacc/host_data-2.c: ... this file.
+ * c-c++-common/goacc/kernels-loop-acc-loop-ptr-it.c: Remove file,
+ moving its content into...
+ * c-c++-common/goacc/kernels-1.c: ... this file.
+ * c-c++-common/goacc/loop-2.c: Rename to...
+ * c-c++-common/goacc/loop-2-parallel.c: ... this file.
+ * c-c++-common/goacc/loop-3.c: Remove (invalid) nesting tests
+ covered elsewhere.
+ * c-c++-common/goacc/loop-4.c: Rename to...
+ * c-c++-common/goacc/loop-2-kernels.c: ... this file.
+ * c-c++-common/goacc/loop-nest-1.c: Remove file.
+ * c-c++-common/goacc/loop-tile-k1.c: Remove file, moving its
+ content into...
+ * c-c++-common/goacc/tile.c: ... this file.
+ * c-c++-common/goacc/loop-tile-p1.c: Remove file, moving its
+ content into...
+ * c-c++-common/goacc/tile.c: ... this file.
+ * c-c++-common/goacc/non-routine.c: Remove file, moving its
+ content into...
+ * c-c++-common/goacc/nesting-fail-1.c: ... this file.
+ * c-c++-common/goacc/parallel-empty.c: Remove file, moving its
+ content into...
+ * c-c++-common/goacc/parallel-1.c: ... this file.
+ * c-c++-common/goacc/parallel-eternal.c: Remove file, moving its
+ content into...
+ * c-c++-common/goacc/parallel-1.c: ... this file.
+ * c-c++-common/goacc/parallel-noreturn.c: Remove file, moving its
+ content into...
+ * c-c++-common/goacc/parallel-1.c: ... this file.
+ * c-c++-common/goacc/routine-6.c: Remove file, moving its
+ content into...
+ * c-c++-common/goacc/routine-3.c: ... this file.
+ * c-c++-common/goacc/routine-7.c: Remove file, moving its
+ content into...
+ * c-c++-common/goacc/routine-4.c: ... this file.
+ * g++.dg/goacc/template-reduction.C: Remove file.
+ * gfortran.dg/goacc/parallel-tree.f95: Use dg-warning directives
+ instead of specifying the -w compiler option.
+
2016-03-11 Cesar Philippidis <cesar@codesourcery.com>
* c-c++-common/goacc/combined-directives-2.c: New test.
@@ -1,3 +1,5 @@
+/* Miscellaneous tests where clause parsing is expected to fail. */
+
void
f (void)
{
@@ -17,3 +19,13 @@ f (void)
for (i = 0; i < 2; ++i)
;
}
+
+
+void
+f2 (void)
+{
+ int a, b[100];
+
+#pragma acc parallel firstprivate (b[10:20]) /* { dg-error "expected ... before ... token" } */
+ ;
+}
@@ -1,7 +1,8 @@
-// { dg-do compile }
-// { dg-options "-fopenacc -fdump-tree-gimple" }
+// { dg-additional-options "-fdump-tree-gimple" }
// TODO
+// Remove the comments from the reduction test
+// after the FE learns that reduction variables may appear in data clauses too.
// Enable and update tree scanning for reduction clauses.
// Add device_type clauses and tree scanning.
deleted file mode 100644
@@ -1,9 +0,0 @@
-void
-foo (void)
-{
- int a, b[100];
-#pragma acc parallel firstprivate (a, b)
- ;
-#pragma acc parallel firstprivate (b[10:20]) /* { dg-error "expected" } */
- ;
-}
@@ -1,5 +1,4 @@
/* Test valid use of host_data directive. */
-/* { dg-do compile } */
int v1[3][3];
@@ -9,3 +8,17 @@ f (void)
#pragma acc host_data use_device(v1)
;
}
+
+
+void bar (float *, float *);
+
+void
+foo (float *x, float *y)
+{
+ int n = 1 << 10;
+#pragma acc data create(x[0:n]) copyout(y[0:n])
+ {
+#pragma acc host_data use_device(x,y)
+ bar (x, y);
+ }
+}
@@ -1,14 +1,14 @@
/* Test invalid use of host_data directive. */
-/* { dg-do compile } */
int v0;
-#pragma acc host_data use_device(v0) /* { dg-error "expected" } */
+#pragma acc host_data use_device(v0) /* { dg-error "expected declaration specifiers before" } */
+
void
f (void)
{
int v2 = 3;
-#pragma acc host_data copy(v2) /* { dg-error "not valid for" } */
+#pragma acc host_data copy(v2) /* { dg-error ".copy. is not valid for ..pragma acc host_data." } */
;
#pragma acc host_data use_device(v2)
@@ -21,3 +21,58 @@ f (void)
/* { dg-error ".use_device_ptr. variable is neither a pointer nor an array" "" { target c } 19 } */
/* { dg-error ".use_device_ptr. variable is neither a pointer, nor an arraynor reference to pointer or array" "" { target c++ } 19 } */
}
+
+
+void
+f2 (void)
+{
+ int x[100];
+
+#pragma acc enter data copyin (x)
+ /* Specifying an array index is not valid for host_data/use_device. */
+#pragma acc host_data use_device (x[4]) /* { dg-error "expected '\\\)' before '\\\[' token" } */
+ ;
+#pragma acc exit data delete (x)
+}
+
+
+void
+f3 (void)
+{
+ int x[100];
+
+#pragma acc data copyin (x[25:50])
+ {
+ int *xp;
+#pragma acc host_data use_device (x)
+ {
+ /* This use of the present clause is undefined behavior for OpenACC. */
+#pragma acc parallel present (x) copyout (xp) /* { dg-error "variable .x. declared in enclosing .host_data. region" } */
+ {
+ xp = x;
+ }
+ }
+ }
+}
+
+
+void
+f4 (void)
+{
+ int x[50];
+
+#pragma acc data copyin (x[10:30])
+ {
+ int *xp;
+#pragma acc host_data use_device (x)
+ {
+ /* Here 'x' being implicitly firstprivate for the parallel region
+ conflicts with it being declared as use_device in the enclosing
+ host_data region. */
+#pragma acc parallel copyout (xp)
+ {
+ xp = x; /* { dg-error "variable .x. declared in enclosing .host_data. region" } */
+ }
+ }
+ }
+}
deleted file mode 100644
@@ -1,18 +0,0 @@
-/* { dg-do compile } */
-
-int main (int argc, char* argv[])
-{
- int x = 5, y;
-
- #pragma acc enter data copyin (x)
- /* It's not clear what attempts to use non-pointer variables "directly"
- (rather than merely taking their address) should do in host_data regions.
- We choose to make it an error. */
- #pragma acc host_data use_device (x) /* TODO { dg-error "" } */
- {
- y = x;
- }
- #pragma acc exit data delete (x)
-
- return y - 5;
-}
deleted file mode 100644
@@ -1,14 +0,0 @@
-/* { dg-do compile } */
-
-int main (int argc, char* argv[])
-{
- int x[100];
-
- #pragma acc enter data copyin (x)
- /* Specifying an array index is not valid for host_data/use_device. */
- #pragma acc host_data use_device (x[4]) /* { dg-error "expected '\\\)' before '\\\[' token" } */
- ;
- #pragma acc exit data delete (x)
-
- return 0;
-}
deleted file mode 100644
@@ -1,23 +0,0 @@
-/* { dg-do compile } */
-
-#define N 1024
-
-int main (int argc, char* argv[])
-{
- int x[N];
-
-#pragma acc data copyin (x[0:N])
- {
- int *xp;
-#pragma acc host_data use_device (x)
- {
- /* This use of the present clause is undefined behavior for OpenACC. */
-#pragma acc parallel present (x) copyout (xp) /* { dg-error "variable 'x' declared in enclosing 'host_data' region" } */
- {
- xp = x;
- }
- }
- }
-
- return 0;
-}
deleted file mode 100644
@@ -1,25 +0,0 @@
-/* { dg-do compile } */
-
-#define N 1024
-
-int main (int argc, char* argv[])
-{
- int x[N];
-
-#pragma acc data copyin (x[0:N])
- {
- int *xp;
-#pragma acc host_data use_device (x)
- {
- /* Here 'x' being implicitly firstprivate for the parallel region
- conflicts with it being declared as use_device in the enclosing
- host_data region. */
-#pragma acc parallel copyout (xp)
- {
- xp = x; /* { dg-error "variable 'x' declared in enclosing 'host_data' region" } */
- }
- }
- }
-
- return 0;
-}
@@ -1,5 +1,3 @@
-/* { dg-additional-options "-Wall" } */
-
void
f (short c)
{
new file mode 100644
@@ -0,0 +1,45 @@
+int
+kernels_empty (void)
+{
+#pragma acc kernels
+ ;
+
+ return 0;
+}
+
+int
+kernels_eternal (void)
+{
+#pragma acc kernels
+ {
+ while (1)
+ ;
+ }
+
+ return 0;
+}
+
+int
+kernels_noreturn (void)
+{
+#pragma acc kernels
+ __builtin_abort ();
+
+ return 0;
+}
+
+
+float b[10][15][10];
+
+void
+kernels_loop_ptr_it (void)
+{
+ float *i;
+
+#pragma acc kernels
+ {
+#pragma acc loop
+ for (i = &b[0][0][0]; i < &b[0][0][10]; i++)
+ ;
+ }
+}
deleted file mode 100644
@@ -1,6 +0,0 @@
-void
-foo (void)
-{
-#pragma acc kernels
- ;
-}
deleted file mode 100644
@@ -1,11 +0,0 @@
-int
-main (void)
-{
-#pragma acc kernels
- {
- while (1)
- ;
- }
-
- return 0;
-}
deleted file mode 100644
@@ -1,14 +0,0 @@
-float b[10][15][10];
-
-void
-foo (void)
-{
- float *i;
-
-#pragma acc kernels
- {
-#pragma acc loop
- for (i = &b[0][0][0]; i < &b[0][0][10]; i++)
- ;
- }
-}
deleted file mode 100644
@@ -1,12 +0,0 @@
-int
-main (void)
-{
-
-#pragma acc kernels
- {
- __builtin_abort ();
- }
-
- return 0;
-}
-
similarity index 98%
rename from gcc/testsuite/c-c++-common/goacc/loop-4.c
rename to gcc/testsuite/c-c++-common/goacc/loop-2-kernels.c
@@ -1,7 +1,4 @@
-/* { dg-do compile } */
-
-int
-main ()
+void K(void)
{
int i, j;
@@ -120,7 +117,6 @@ main ()
{ }
}
-
#pragma acc kernels loop auto
for (i = 0; i < 10; i++)
{ }
@@ -139,7 +135,7 @@ main ()
#pragma acc kernels loop gang(static:*)
for (i = 0; i < 10; i++)
{ }
-
+
#pragma acc kernels loop worker
for (i = 0; i < 10; i++)
{ }
@@ -190,6 +186,4 @@ main ()
#pragma acc kernels loop vector auto // { dg-error "'auto' conflicts" "" { target c } }
for (i = 0; i < 10; i++) // { dg-error "'auto' conflicts" "" { target c++ } }
{ }
-
- return 0;
}
similarity index 98%
rename from gcc/testsuite/c-c++-common/goacc/loop-2.c
rename to gcc/testsuite/c-c++-common/goacc/loop-2-parallel.c
@@ -1,11 +1,7 @@
-/* { dg-do compile } */
-
-int
-main ()
+void P(void)
{
int i, j;
-
#pragma acc parallel
{
#pragma acc loop auto
@@ -163,8 +159,4 @@ main ()
#pragma acc parallel loop vector auto // { dg-error "'auto' conflicts" "" { target c } }
for (i = 0; i < 10; i++) // { dg-error "'auto' conflicts" "" { target c++ } }
{ }
-
-
- return 0;
}
-
@@ -1,7 +1,3 @@
-/* { dg-do compile } */
-/* { dg-additional-options "-fmax-errors=200" } */
-
-
void par1 (void)
{
int i, j;
@@ -35,44 +31,6 @@ void par1 (void)
}
}
-void k2 (void)
-{
- int i, j;
-
-#pragma acc kernels loop gang
- for (i = 0; i < 10; i++)
- {
-#pragma acc kernels loop gang // { dg-error "OpenACC construct inside of non-OpenACC region" }
- for (j = 1; j < 10; j++)
- { }
- }
-
-#pragma acc kernels loop worker
- for (i = 0; i < 10; i++)
- {
-#pragma acc kernels loop worker // { dg-error "OpenACC construct inside of non-OpenACC region" }
- for (j = 1; j < 10; j++)
- { }
-#pragma acc kernels loop gang // { dg-error "OpenACC construct inside of non-OpenACC region" }
- for (j = 1; j < 10; j++)
- { }
- }
-
-#pragma acc kernels loop vector
- for (i = 0; i < 10; i++)
- {
-#pragma acc kernels loop vector // { dg-error "OpenACC construct inside of non-OpenACC region" }
- for (j = 1; j < 10; j++)
- { }
-#pragma acc kernels loop worker // { dg-error "OpenACC construct inside of non-OpenACC region" }
- for (j = 1; j < 10; j++)
- { }
-#pragma acc kernels loop gang // { dg-error "OpenACC construct inside of non-OpenACC region" }
- for (j = 1; j < 10; j++)
- { }
- }
-}
-
void p2 (void)
{
int i, j;
@@ -84,30 +42,12 @@ void p2 (void)
for (i = 0; i < 10; i++) // { dg-error "argument not permitted" "" { target c++ } }
{ }
-#pragma acc parallel loop gang
- for (i = 0; i < 10; i++)
- {
-#pragma acc parallel loop gang // { dg-error "OpenACC construct inside of non-OpenACC region" }
- for (j = 1; j < 10; j++)
- { }
- }
-
#pragma acc parallel loop worker(5) // { dg-error "argument not permitted" "" { target c } }
for (i = 0; i < 10; i++) // { dg-error "argument not permitted" "" { target c++ } }
{ }
#pragma acc parallel loop worker(num:5) // { dg-error "argument not permitted" "" { target c } }
for (i = 0; i < 10; i++) // { dg-error "argument not permitted" "" { target c++ } }
{ }
-#pragma acc parallel loop worker
- for (i = 0; i < 10; i++)
- {
-#pragma acc parallel loop worker // { dg-error "OpenACC construct inside of non-OpenACC region" }
- for (j = 1; j < 10; j++)
- { }
-#pragma acc parallel loop gang // { dg-error "OpenACC construct inside of non-OpenACC region" }
- for (j = 1; j < 10; j++)
- { }
- }
#pragma acc parallel loop vector(5) // { dg-error "argument not permitted" "" { target c } }
for (i = 0; i < 10; i++) // { dg-error "argument not permitted" "" { target c++ } }
@@ -115,18 +55,4 @@ void p2 (void)
#pragma acc parallel loop vector(length:5) // { dg-error "argument not permitted" "" { target c } }
for (i = 0; i < 10; i++) // { dg-error "argument not permitted" "" { target c++ } }
{ }
-#pragma acc parallel loop vector
- for (i = 0; i < 10; i++)
- {
-#pragma acc parallel loop vector // { dg-error "OpenACC construct inside of non-OpenACC region" }
- for (j = 1; j < 10; j++)
- { }
-#pragma acc parallel loop worker // { dg-error "OpenACC construct inside of non-OpenACC region" }
- for (j = 1; j < 10; j++)
- { }
-#pragma acc parallel loop gang // { dg-error "OpenACC construct inside of non-OpenACC region" }
- for (j = 1; j < 10; j++)
- { }
- }
}
-
@@ -1,5 +1,3 @@
-/* { dg-do compile } */
-
int
main ()
{
deleted file mode 100644
@@ -1,16 +0,0 @@
-/* { dg-do compile } */
-
-int
-main ()
-{
- int i, j;
-#pragma acc kernels loop gang
- for (i = 0; i < 10; i++)
- {
-#pragma acc kernels loop gang // { dg-error "OpenACC construct inside of" "" }
- for (i = 0; i < 10; i++)
- { }
- }
-
- return 0;
-}
deleted file mode 100644
@@ -1,132 +0,0 @@
-/* { dg-do compile } */
-/* { dg-additional-options "-fmax-errors=200" } */
-
-void
-kern (void)
-{
- int i, j;
-
-#pragma acc kernels
- {
-#pragma acc loop tile // { dg-error "expected" }
- for (i = 0; i < 10; i++)
- { }
-#pragma acc loop tile() // { dg-error "expected" }
- for (i = 0; i < 10; i++)
- { }
-#pragma acc loop tile(1)
- for (i = 0; i < 10; i++)
- { }
-#pragma acc loop tile(2)
- for (i = 0; i < 10; i++)
- { }
-#pragma acc loop tile(6-2)
- for (i = 0; i < 10; i++)
- { }
-#pragma acc loop tile(6+2)
- for (i = 0; i < 10; i++)
- { }
-#pragma acc loop tile(*)
- for (i = 0; i < 10; i++)
- { }
-#pragma acc loop tile(*, 1)
- for (i = 0; i < 10; i++)
- {
- for (j = 0; j < 10; i++)
- { }
- }
-#pragma acc loop tile(-2) // { dg-warning "'tile' value must be positive" }
- for (i = 0; i < 10; i++)
- { }
-#pragma acc loop tile(i)
- for (i = 0; i < 10; i++)
- { }
-#pragma acc loop tile(2, 2, 1)
- for (i = 2; i < 4; i++)
- for (i = 4; i < 6; i++)
- { }
-#pragma acc loop tile(2, 2)
- for (i = 1; i < 5; i+=2)
- for (j = i+1; j < 7; i++)
- { }
-#pragma acc loop vector tile(*)
- for (i = 0; i < 10; i++)
- { }
-#pragma acc loop worker tile(*)
- for (i = 0; i < 10; i++)
- { }
-#pragma acc loop gang tile(*)
- for (i = 0; i < 10; i++)
- { }
-#pragma acc loop vector gang tile(*)
- for (i = 0; i < 10; i++)
- { }
-#pragma acc loop vector worker tile(*)
- for (i = 0; i < 10; i++)
- { }
-#pragma acc loop gang worker tile(*)
- for (i = 0; i < 10; i++)
- { }
- }
-}
-
-
-void k3 (void)
-{
- int i, j;
-
-#pragma acc kernels loop tile // { dg-error "expected" }
- for (i = 0; i < 10; i++)
- { }
-#pragma acc kernels loop tile() // { dg-error "expected" }
- for (i = 0; i < 10; i++)
- { }
-#pragma acc kernels loop tile(1)
- for (i = 0; i < 10; i++)
- { }
-#pragma acc kernels loop tile(*)
- for (i = 0; i < 10; i++)
- { }
-#pragma acc kernels loop tile(*, 1)
- for (i = 0; i < 10; i++)
- {
- for (j = 1; j < 10; j++)
- { }
- }
-#pragma acc kernels loop tile(-2) // { dg-warning "'tile' value must be positive" }
- for (i = 1; i < 10; i++)
- { }
-#pragma acc kernels loop tile(i)
- for (i = 1; i < 10; i++)
- { }
-#pragma acc kernels loop tile(2, 2, 1)
- for (i = 1; i < 3; i++)
- {
- for (j = 4; j < 6; j++)
- { }
- }
-#pragma acc kernels loop tile(2, 2)
- for (i = 1; i < 5; i++)
- {
- for (j = i + 1; j < 7; j += i)
- { }
- }
-#pragma acc kernels loop vector tile(*)
- for (i = 0; i < 10; i++)
- { }
-#pragma acc kernels loop worker tile(*)
- for (i = 0; i < 10; i++)
- { }
-#pragma acc kernels loop gang tile(*)
- for (i = 0; i < 10; i++)
- { }
-#pragma acc kernels loop vector gang tile(*)
- for (i = 0; i < 10; i++)
- { }
-#pragma acc kernels loop vector worker tile(*)
- for (i = 0; i < 10; i++)
- { }
-#pragma acc kernels loop gang worker tile(*)
- for (i = 0; i < 10; i++)
- { }
-}
deleted file mode 100644
@@ -1,128 +0,0 @@
-/* { dg-do compile } */
-/* { dg-additional-options "-fmax-errors=200" } */
-
-
-void par (void)
-{
- int i, j;
-
-#pragma acc parallel
- {
-#pragma acc loop tile // { dg-error "expected" }
- for (i = 0; i < 10; i++)
- { }
-#pragma acc loop tile() // { dg-error "expected" }
- for (i = 0; i < 10; i++)
- { }
-#pragma acc loop tile(1)
- for (i = 0; i < 10; i++)
- { }
-#pragma acc loop tile(*)
- for (i = 0; i < 10; i++)
- { }
-#pragma acc loop tile(2)
- for (i = 0; i < 10; i++)
- {
- for (j = 1; j < 10; j++)
- { }
- }
-#pragma acc loop tile(-2) // { dg-warning "'tile' value must be positive" }
- for (i = 1; i < 10; i++)
- { }
-#pragma acc loop tile(i)
- for (i = 1; i < 10; i++)
- { }
-#pragma acc loop tile(2, 2, 1)
- for (i = 1; i < 3; i++)
- {
- for (j = 4; j < 6; j++)
- { }
- }
-#pragma acc loop tile(2, 2)
- for (i = 1; i < 5; i+=2)
- {
- for (j = i + 1; j < 7; j+=i)
- { }
- }
-#pragma acc loop vector tile(*)
- for (i = 0; i < 10; i++)
- { }
-#pragma acc loop worker tile(*)
- for (i = 0; i < 10; i++)
- { }
-#pragma acc loop gang tile(*)
- for (i = 0; i < 10; i++)
- { }
-#pragma acc loop vector gang tile(*)
- for (i = 0; i < 10; i++)
- { }
-#pragma acc loop vector worker tile(*)
- for (i = 0; i < 10; i++)
- { }
-#pragma acc loop gang worker tile(*)
- for (i = 0; i < 10; i++)
- { }
- }
-}
-void p3 (void)
-{
- int i, j;
-
-
-#pragma acc parallel loop tile // { dg-error "expected" }
- for (i = 0; i < 10; i++)
- { }
-#pragma acc parallel loop tile() // { dg-error "expected" }
- for (i = 0; i < 10; i++)
- { }
-#pragma acc parallel loop tile(1)
- for (i = 0; i < 10; i++)
- { }
-#pragma acc parallel loop tile(*)
- for (i = 0; i < 10; i++)
- { }
-#pragma acc parallel loop tile(*, 1)
- for (i = 0; i < 10; i++)
- {
- for (j = 1; j < 10; j++)
- { }
- }
-#pragma acc parallel loop tile(-2) // { dg-warning "'tile' value must be positive" }
- for (i = 1; i < 10; i++)
- { }
-#pragma acc parallel loop tile(i)
- for (i = 1; i < 10; i++)
- { }
-#pragma acc parallel loop tile(2, 2, 1)
- for (i = 1; i < 3; i++)
- {
- for (j = 4; j < 6; j++)
- { }
- }
-#pragma acc parallel loop tile(2, 2)
- for (i = 1; i < 5; i+=2)
- {
- for (j = i + 1; j < 7; j++)
- { }
- }
-#pragma acc parallel loop vector tile(*)
- for (i = 0; i < 10; i++)
- { }
-#pragma acc parallel loop worker tile(*)
- for (i = 0; i < 10; i++)
- { }
-#pragma acc parallel loop gang tile(*)
- for (i = 0; i < 10; i++)
- { }
-#pragma acc parallel loop vector gang tile(*)
- for (i = 0; i < 10; i++)
- { }
-#pragma acc parallel loop vector worker tile(*)
- for (i = 0; i < 10; i++)
- { }
-#pragma acc parallel loop gang worker tile(*)
- for (i = 0; i < 10; i++)
- { }
-
-}
-
@@ -7,9 +7,9 @@ f_acc_parallel (void)
{
#pragma acc parallel
{
-#pragma acc parallel /* { dg-error ".parallel. construct inside of .parallel. region" } */
+#pragma acc parallel /* { dg-bogus ".parallel. construct inside of .parallel. region" "not implemented" { xfail *-*-* } } */
;
-#pragma acc kernels /* { dg-error ".kernels. construct inside of .parallel. region" } */
+#pragma acc kernels /* { dg-bogus ".kernels. construct inside of .parallel. region" "not implemented" { xfail *-*-* } } */
;
#pragma acc data /* { dg-error ".data. construct inside of .parallel. region" } */
;
@@ -26,9 +26,9 @@ f_acc_kernels (void)
{
#pragma acc kernels
{
-#pragma acc parallel /* { dg-error ".parallel. construct inside of .kernels. region" } */
+#pragma acc parallel /* { dg-bogus ".parallel. construct inside of .kernels. region" "not implemented" { xfail *-*-* } } */
;
-#pragma acc kernels /* { dg-error ".kernels. construct inside of .kernels. region" } */
+#pragma acc kernels /* { dg-bogus ".kernels. construct inside of .kernels. region" "not implemented" { xfail *-*-* } } */
;
#pragma acc data /* { dg-error ".data. construct inside of .kernels. region" } */
;
@@ -64,3 +64,13 @@ f_acc_routine (void)
#pragma acc parallel /* { dg-error "OpenACC region inside of OpenACC routine, nested parallelism not supported yet" } */
;
}
+
+void
+f (void)
+{
+ int i, v = 0;
+
+#pragma acc loop gang reduction (+:v) /* { dg-error "loop directive must be associated with an OpenACC compute region" } */
+ for (i = 0; i < 10; i++)
+ v++;
+}
deleted file mode 100644
@@ -1,16 +0,0 @@
-/* This program validates the behavior of acc loops which are
- not associated with a parallel or kernles region or routine. */
-
-/* { dg-do compile } */
-
-int
-main ()
-{
- int i, v = 0;
-
-#pragma acc loop gang reduction (+:v) /* { dg-error "loop directive must be associated with an OpenACC compute region" } */
- for (i = 0; i < 10; i++)
- v++;
-
- return v;
-}
new file mode 100644
@@ -0,0 +1,38 @@
+int
+parallel_empty (void)
+{
+#pragma acc parallel
+ ;
+
+ return 0;
+}
+
+int
+parallel_eternal (void)
+{
+#pragma acc parallel
+ {
+ while (1)
+ ;
+ }
+
+ return 0;
+}
+
+int
+parallel_noreturn (void)
+{
+#pragma acc parallel
+ __builtin_abort ();
+
+ return 0;
+}
+
+int
+parallel_clauses (void)
+{
+ int a, b[100];
+
+#pragma acc parallel firstprivate (a, b)
+ ;
+}
deleted file mode 100644
@@ -1,6 +0,0 @@
-void
-foo (void)
-{
-#pragma acc parallel
- ;
-}
deleted file mode 100644
@@ -1,11 +0,0 @@
-int
-main (void)
-{
-#pragma acc parallel
- {
- while (1)
- ;
- }
-
- return 0;
-}
deleted file mode 100644
@@ -1,12 +0,0 @@
-int
-main (void)
-{
-
-#pragma acc parallel
- {
- __builtin_abort ();
- }
-
- return 0;
-}
-
@@ -1,6 +1,5 @@
/* Integer reductions. */
-#define vl 32
#define n 1000
int
@@ -11,56 +10,56 @@ main(void)
int lresult;
/* '+' reductions. */
-#pragma acc parallel vector_length (vl)
-#pragma acc loop vector reduction (+:result)
+#pragma acc parallel
+#pragma acc loop gang worker vector reduction (+:result)
for (i = 0; i < n; i++)
result += array[i];
/* '*' reductions. */
-#pragma acc parallel vector_length (vl)
-#pragma acc loop vector reduction (*:result)
+#pragma acc parallel
+#pragma acc loop gang worker vector reduction (*:result)
for (i = 0; i < n; i++)
result *= array[i];
/* 'max' reductions. */
-#pragma acc parallel vector_length (vl)
-#pragma acc loop vector reduction (max:result)
+#pragma acc parallel
+#pragma acc loop gang worker vector reduction (max:result)
for (i = 0; i < n; i++)
result = result > array[i] ? result : array[i];
/* 'min' reductions. */
-#pragma acc parallel vector_length (vl)
-#pragma acc loop vector reduction (min:result)
+#pragma acc parallel
+#pragma acc loop gang worker vector reduction (min:result)
for (i = 0; i < n; i++)
result = result < array[i] ? result : array[i];
/* '&' reductions. */
-#pragma acc parallel vector_length (vl)
-#pragma acc loop vector reduction (&:result)
+#pragma acc parallel
+#pragma acc loop gang worker vector reduction (&:result)
for (i = 0; i < n; i++)
result &= array[i];
/* '|' reductions. */
-#pragma acc parallel vector_length (vl)
-#pragma acc loop vector reduction (|:result)
+#pragma acc parallel
+#pragma acc loop gang worker vector reduction (|:result)
for (i = 0; i < n; i++)
result |= array[i];
/* '^' reductions. */
-#pragma acc parallel vector_length (vl)
-#pragma acc loop vector reduction (^:result)
+#pragma acc parallel
+#pragma acc loop gang worker vector reduction (^:result)
for (i = 0; i < n; i++)
result ^= array[i];
/* '&&' reductions. */
-#pragma acc parallel vector_length (vl)
-#pragma acc loop vector reduction (&&:lresult)
+#pragma acc parallel
+#pragma acc loop gang worker vector reduction (&&:lresult)
for (i = 0; i < n; i++)
lresult = lresult && (result > array[i]);
/* '||' reductions. */
-#pragma acc parallel vector_length (vl)
-#pragma acc loop vector reduction (||:lresult)
+#pragma acc parallel
+#pragma acc loop gang worker vector reduction (||:lresult)
for (i = 0; i < n; i++)
lresult = lresult || (result > array[i]);
@@ -1,6 +1,5 @@
/* float reductions. */
-#define vl 32
#define n 1000
int
@@ -11,38 +10,38 @@ main(void)
int lresult;
/* '+' reductions. */
-#pragma acc parallel vector_length (vl)
-#pragma acc loop vector reduction (+:result)
+#pragma acc parallel
+#pragma acc loop gang worker vector reduction (+:result)
for (i = 0; i < n; i++)
result += array[i];
/* '*' reductions. */
-#pragma acc parallel vector_length (vl)
-#pragma acc loop vector reduction (*:result)
+#pragma acc parallel
+#pragma acc loop gang worker vector reduction (*:result)
for (i = 0; i < n; i++)
result *= array[i];
/* 'max' reductions. */
-#pragma acc parallel vector_length (vl)
-#pragma acc loop vector reduction (max:result)
+#pragma acc parallel
+#pragma acc loop gang worker vector reduction (max:result)
for (i = 0; i < n; i++)
result = result > array[i] ? result : array[i];
/* 'min' reductions. */
-#pragma acc parallel vector_length (vl)
-#pragma acc loop vector reduction (min:result)
+#pragma acc parallel
+#pragma acc loop gang worker vector reduction (min:result)
for (i = 0; i < n; i++)
result = result < array[i] ? result : array[i];
/* '&&' reductions. */
-#pragma acc parallel vector_length (vl)
-#pragma acc loop vector reduction (&&:lresult)
+#pragma acc parallel
+#pragma acc loop gang worker vector reduction (&&:lresult)
for (i = 0; i < n; i++)
lresult = lresult && (result > array[i]);
/* '||' reductions. */
-#pragma acc parallel vector_length (vl)
-#pragma acc loop vector reduction (||:lresult)
+#pragma acc parallel
+#pragma acc loop gang worker vector reduction (||:lresult)
for (i = 0; i < n; i++)
lresult = lresult || (result > array[i]);
@@ -1,6 +1,5 @@
/* double reductions. */
-#define vl 32
#define n 1000
int
@@ -11,38 +10,38 @@ main(void)
int lresult;
/* '+' reductions. */
-#pragma acc parallel vector_length (vl)
-#pragma acc loop vector reduction (+:result)
+#pragma acc parallel
+#pragma acc loop gang worker vector reduction (+:result)
for (i = 0; i < n; i++)
result += array[i];
/* '*' reductions. */
-#pragma acc parallel vector_length (vl)
-#pragma acc loop vector reduction (*:result)
+#pragma acc parallel
+#pragma acc loop gang worker vector reduction (*:result)
for (i = 0; i < n; i++)
result *= array[i];
/* 'max' reductions. */
-#pragma acc parallel vector_length (vl)
-#pragma acc loop vector reduction (max:result)
+#pragma acc parallel
+#pragma acc loop gang worker vector reduction (max:result)
for (i = 0; i < n; i++)
result = result > array[i] ? result : array[i];
/* 'min' reductions. */
-#pragma acc parallel vector_length (vl)
-#pragma acc loop vector reduction (min:result)
+#pragma acc parallel
+#pragma acc loop gang worker vector reduction (min:result)
for (i = 0; i < n; i++)
result = result < array[i] ? result : array[i];
/* '&&' reductions. */
-#pragma acc parallel vector_length (vl)
-#pragma acc loop vector reduction (&&:lresult)
+#pragma acc parallel
+#pragma acc loop gang worker vector reduction (&&:lresult)
for (i = 0; i < n; i++)
lresult = lresult && (result > array[i]);
/* '||' reductions. */
-#pragma acc parallel vector_length (vl)
-#pragma acc loop vector reduction (||:lresult)
+#pragma acc parallel
+#pragma acc loop gang worker vector reduction (||:lresult)
for (i = 0; i < n; i++)
lresult = lresult || (result > array[i]);
@@ -1,6 +1,5 @@
/* complex reductions. */
-#define vl 32
#define n 1000
int
@@ -11,44 +10,26 @@ main(void)
int lresult;
/* '+' reductions. */
-#pragma acc parallel vector_length (vl)
-#pragma acc loop vector reduction (+:result)
+#pragma acc parallel
+#pragma acc loop gang worker vector reduction (+:result)
for (i = 0; i < n; i++)
result += array[i];
/* '*' reductions. */
-#pragma acc parallel vector_length (vl)
-#pragma acc loop vector reduction (*:result)
+#pragma acc parallel
+#pragma acc loop gang worker vector reduction (*:result)
for (i = 0; i < n; i++)
result *= array[i];
- /* 'max' reductions. */
-#if 0
- // error: 'result' has invalid type for 'reduction(max)'
-#pragma acc parallel vector_length (vl)
-#pragma acc loop vector reduction (max:result)
- for (i = 0; i < n; i++)
- result = result > array[i] ? result : array[i];
-#endif
-
- /* 'min' reductions. */
-#if 0
- // error: 'result' has invalid type for 'reduction(min)'
-#pragma acc parallel vector_length (vl)
-#pragma acc loop vector reduction (min:result)
- for (i = 0; i < n; i++)
- result = result < array[i] ? result : array[i];
-#endif
-
/* '&&' reductions. */
-#pragma acc parallel vector_length (vl)
-#pragma acc loop vector reduction (&&:lresult)
+#pragma acc parallel
+#pragma acc loop gang worker vector reduction (&&:lresult)
for (i = 0; i < n; i++)
lresult = lresult && (__real__(result) > __real__(array[i]));
/* '||' reductions. */
-#pragma acc parallel vector_length (vl)
-#pragma acc loop vector reduction (||:lresult)
+#pragma acc parallel
+#pragma acc loop gang worker vector reduction (||:lresult)
for (i = 0; i < n; i++)
lresult = lresult || (__real__(result) > __real__(array[i]));
@@ -1,64 +1,118 @@
+/* Test invalid calls to routines. */
+
#pragma acc routine gang
-void gang (void) /* { dg-message "declared here" 3 } */
+int
+gang () /* { dg-message "declared here" 3 } */
{
#pragma acc loop gang worker vector
for (int i = 0; i < 10; i++)
{
}
+
+ return 1;
}
#pragma acc routine worker
-void worker (void) /* { dg-message "declared here" 2 } */
+int
+worker () /* { dg-message "declared here" 2 } */
{
#pragma acc loop worker vector
for (int i = 0; i < 10; i++)
{
}
+
+ return 1;
}
#pragma acc routine vector
-void vector (void) /* { dg-message "declared here" 1 } */
+int
+vector () /* { dg-message "declared here" } */
{
#pragma acc loop vector
for (int i = 0; i < 10; i++)
{
}
+
+ return 1;
}
#pragma acc routine seq
-void seq (void)
+int
+seq ()
{
+ return 1;
}
-int main ()
+int
+main ()
{
-
-#pragma acc parallel num_gangs (32) num_workers (32) vector_length (32)
+ int red = 0;
+#pragma acc parallel copy (red)
{
- #pragma acc loop gang /* { dg-message "loop here" 1 } */
+ /* Independent/seq loop tests. */
+#pragma acc loop reduction (+:red) // { dg-warning "insufficient partitioning" }
+ for (int i = 0; i < 10; i++)
+ red += gang ();
+
+#pragma acc loop reduction (+:red)
+ for (int i = 0; i < 10; i++)
+ red += worker ();
+
+#pragma acc loop reduction (+:red)
+ for (int i = 0; i < 10; i++)
+ red += vector ();
+
+ /* Gang routine tests. */
+#pragma acc loop gang reduction (+:red) /* { dg-message "containing loop" } */
+ for (int i = 0; i < 10; i++)
+ red += gang (); // { dg-error "routine call uses same" }
+
+#pragma acc loop worker reduction (+:red) /* { dg-message "containing loop" } */
+ for (int i = 0; i < 10; i++)
+ red += gang (); // { dg-error "routine call uses same" }
+
+#pragma acc loop vector reduction (+:red) /* { dg-message "containing loop" } */
for (int i = 0; i < 10; i++)
- {
- gang (); /* { dg-error "routine call uses same" } */
- worker ();
- vector ();
- seq ();
- }
- #pragma acc loop worker /* { dg-message "loop here" 2 } */
+ red += gang (); // { dg-error "routine call uses same" }
+
+ /* Worker routine tests. */
+#pragma acc loop gang reduction (+:red)
for (int i = 0; i < 10; i++)
- {
- gang (); /* { dg-error "routine call uses same" } */
- worker (); /* { dg-error "routine call uses same" } */
- vector ();
- seq ();
- }
- #pragma acc loop vector /* { dg-message "loop here" 3 } */
+ red += worker ();
+
+#pragma acc loop worker reduction (+:red) /* { dg-message "containing loop" } */
+ for (int i = 0; i < 10; i++)
+ red += worker (); // { dg-error "routine call uses same" }
+
+#pragma acc loop vector reduction (+:red) /* { dg-message "containing loop" } */
+ for (int i = 0; i < 10; i++)
+ red += worker (); // { dg-error "routine call uses same" }
+
+ /* Vector routine tests. */
+#pragma acc loop gang reduction (+:red)
+ for (int i = 0; i < 10; i++)
+ red += vector ();
+
+#pragma acc loop worker reduction (+:red)
+ for (int i = 0; i < 10; i++)
+ red += vector ();
+
+#pragma acc loop vector reduction (+:red) /* { dg-message "containing loop" } */
+ for (int i = 0; i < 10; i++)
+ red += vector (); // { dg-error "routine call uses same" }
+
+ /* Seq routine tests. */
+#pragma acc loop gang reduction (+:red)
+ for (int i = 0; i < 10; i++)
+ red += seq ();
+
+#pragma acc loop worker reduction (+:red)
+ for (int i = 0; i < 10; i++)
+ red += seq ();
+
+#pragma acc loop vector reduction (+:red)
for (int i = 0; i < 10; i++)
- {
- gang (); /* { dg-error "routine call uses same" } */
- worker (); /* { dg-error "routine call uses same" } */
- vector (); /* { dg-error "routine call uses same" } */
- seq ();
- }
+ red += seq ();
}
return 0;
@@ -1,3 +1,4 @@
+/* Test invalid intra-routine parallelism. */
void gang (void);
void worker (void);
@@ -14,6 +15,24 @@ void seq (void)
worker (); /* { dg-error "routine call uses" } */
vector (); /* { dg-error "routine call uses" } */
seq ();
+
+ int red;
+
+#pragma acc loop reduction (+:red) // { dg-warning "insufficient partitioning" }
+ for (int i = 0; i < 10; i++)
+ red ++;
+
+#pragma acc loop gang reduction (+:red) // { dg-error "disallowed by containing routine" }
+ for (int i = 0; i < 10; i++)
+ red ++;
+
+#pragma acc loop worker reduction (+:red) // { dg-error "disallowed by containing routine" }
+ for (int i = 0; i < 10; i++)
+ red ++;
+
+#pragma acc loop vector reduction (+:red) // { dg-error "disallowed by containing routine" }
+ for (int i = 0; i < 10; i++)
+ red ++;
}
void vector (void) /* { dg-message "declared here" 1 } */
@@ -22,6 +41,24 @@ void vector (void) /* { dg-message "declared here" 1 } */
worker (); /* { dg-error "routine call uses" } */
vector ();
seq ();
+
+ int red;
+
+#pragma acc loop reduction (+:red)
+ for (int i = 0; i < 10; i++)
+ red ++;
+
+#pragma acc loop gang reduction (+:red) // { dg-error "disallowed by containing routine" }
+ for (int i = 0; i < 10; i++)
+ red ++;
+
+#pragma acc loop worker reduction (+:red) // { dg-error "disallowed by containing routine" }
+ for (int i = 0; i < 10; i++)
+ red ++;
+
+#pragma acc loop vector reduction (+:red)
+ for (int i = 0; i < 10; i++)
+ red ++;
}
void worker (void) /* { dg-message "declared here" 2 } */
@@ -30,6 +67,24 @@ void worker (void) /* { dg-message "declared here" 2 } */
worker ();
vector ();
seq ();
+
+ int red;
+
+#pragma acc loop reduction (+:red)
+ for (int i = 0; i < 10; i++)
+ red ++;
+
+#pragma acc loop gang reduction (+:red) // { dg-error "disallowed by containing routine" }
+ for (int i = 0; i < 10; i++)
+ red ++;
+
+#pragma acc loop worker reduction (+:red)
+ for (int i = 0; i < 10; i++)
+ red ++;
+
+#pragma acc loop vector reduction (+:red)
+ for (int i = 0; i < 10; i++)
+ red ++;
}
void gang (void) /* { dg-message "declared here" 3 } */
@@ -38,4 +93,22 @@ void gang (void) /* { dg-message "declared here" 3 } */
worker ();
vector ();
seq ();
+
+ int red;
+
+#pragma acc loop reduction (+:red)
+ for (int i = 0; i < 10; i++)
+ red ++;
+
+#pragma acc loop gang reduction (+:red)
+ for (int i = 0; i < 10; i++)
+ red ++;
+
+#pragma acc loop worker reduction (+:red)
+ for (int i = 0; i < 10; i++)
+ red ++;
+
+#pragma acc loop vector reduction (+:red)
+ for (int i = 0; i < 10; i++)
+ red ++;
}
deleted file mode 100644
@@ -1,120 +0,0 @@
-/* Test invalid calls to routines. */
-/* { dg-do compile } */
-
-#pragma acc routine gang
-int
-gang () /* { dg-message "declared here" 3 } */
-{
- #pragma acc loop gang worker vector
- for (int i = 0; i < 10; i++)
- {
- }
-
- return 1;
-}
-
-#pragma acc routine worker
-int
-worker () /* { dg-message "declared here" 2 } */
-{
- #pragma acc loop worker vector
- for (int i = 0; i < 10; i++)
- {
- }
-
- return 1;
-}
-
-#pragma acc routine vector
-int
-vector () /* { dg-message "declared here" } */
-{
- #pragma acc loop vector
- for (int i = 0; i < 10; i++)
- {
- }
-
- return 1;
-}
-
-#pragma acc routine seq
-int
-seq ()
-{
- return 1;
-}
-
-int
-main ()
-{
- int red = 0;
-#pragma acc parallel copy (red)
- {
- /* Independent/seq loop tests. */
-#pragma acc loop reduction (+:red) // { dg-warning "insufficient partitioning" }
- for (int i = 0; i < 10; i++)
- red += gang ();
-
-#pragma acc loop reduction (+:red)
- for (int i = 0; i < 10; i++)
- red += worker ();
-
-#pragma acc loop reduction (+:red)
- for (int i = 0; i < 10; i++)
- red += vector ();
-
- /* Gang routine tests. */
-#pragma acc loop gang reduction (+:red) /* { dg-message "containing loop" } */
- for (int i = 0; i < 10; i++)
- red += gang (); // { dg-error "routine call uses same" }
-
-#pragma acc loop worker reduction (+:red) /* { dg-message "containing loop" } */
- for (int i = 0; i < 10; i++)
- red += gang (); // { dg-error "routine call uses same" }
-
-#pragma acc loop vector reduction (+:red) /* { dg-message "containing loop" } */
- for (int i = 0; i < 10; i++)
- red += gang (); // { dg-error "routine call uses same" }
-
- /* Worker routine tests. */
-#pragma acc loop gang reduction (+:red)
- for (int i = 0; i < 10; i++)
- red += worker ();
-
-#pragma acc loop worker reduction (+:red) /* { dg-message "containing loop" } */
- for (int i = 0; i < 10; i++)
- red += worker (); // { dg-error "routine call uses same" }
-
-#pragma acc loop vector reduction (+:red) /* { dg-message "containing loop" } */
- for (int i = 0; i < 10; i++)
- red += worker (); // { dg-error "routine call uses same" }
-
- /* Vector routine tests. */
-#pragma acc loop gang reduction (+:red)
- for (int i = 0; i < 10; i++)
- red += vector ();
-
-#pragma acc loop worker reduction (+:red)
- for (int i = 0; i < 10; i++)
- red += vector ();
-
-#pragma acc loop vector reduction (+:red) /* { dg-message "containing loop" } */
- for (int i = 0; i < 10; i++)
- red += vector (); // { dg-error "routine call uses same" }
-
- /* Seq routine tests. */
-#pragma acc loop gang reduction (+:red)
- for (int i = 0; i < 10; i++)
- red += seq ();
-
-#pragma acc loop worker reduction (+:red)
- for (int i = 0; i < 10; i++)
- red += seq ();
-
-#pragma acc loop vector reduction (+:red)
- for (int i = 0; i < 10; i++)
- red += seq ();
- }
-
- return 0;
-}
deleted file mode 100644
@@ -1,94 +0,0 @@
-/* Test invalid intra-routine parallelism. */
-/* { dg-do compile } */
-
-#pragma acc routine gang
-int
-gang (int red)
-{
-#pragma acc loop reduction (+:red)
- for (int i = 0; i < 10; i++)
- red ++;
-
-#pragma acc loop gang reduction (+:red)
- for (int i = 0; i < 10; i++)
- red ++;
-
-#pragma acc loop worker reduction (+:red)
- for (int i = 0; i < 10; i++)
- red ++;
-
-#pragma acc loop vector reduction (+:red)
- for (int i = 0; i < 10; i++)
- red ++;
-
- return 1;
-}
-
-#pragma acc routine worker
-int
-worker (int red)
-{
-#pragma acc loop reduction (+:red)
- for (int i = 0; i < 10; i++)
- red ++;
-
-#pragma acc loop gang reduction (+:red) // { dg-error "disallowed by containing routine" }
- for (int i = 0; i < 10; i++)
- red ++;
-
-#pragma acc loop worker reduction (+:red)
- for (int i = 0; i < 10; i++)
- red ++;
-
-#pragma acc loop vector reduction (+:red)
- for (int i = 0; i < 10; i++)
- red ++;
-
- return 1;
-}
-
-#pragma acc routine vector
-int
-vector (int red)
-{
-#pragma acc loop reduction (+:red)
- for (int i = 0; i < 10; i++)
- red ++;
-
-#pragma acc loop gang reduction (+:red) // { dg-error "disallowed by containing routine" }
- for (int i = 0; i < 10; i++)
- red ++;
-
-#pragma acc loop worker reduction (+:red) // { dg-error "disallowed by containing routine" }
- for (int i = 0; i < 10; i++)
- red ++;
-
-#pragma acc loop vector reduction (+:red)
- for (int i = 0; i < 10; i++)
- red ++;
-
- return 1;
-}
-
-#pragma acc routine seq
-int
-seq (int red)
-{
-#pragma acc loop reduction (+:red) // { dg-warning "insufficient partitioning" }
- for (int i = 0; i < 10; i++)
- red ++;
-
-#pragma acc loop gang reduction (+:red) // { dg-error "disallowed by containing routine" }
- for (int i = 0; i < 10; i++)
- red ++;
-
-#pragma acc loop worker reduction (+:red) // { dg-error "disallowed by containing routine" }
- for (int i = 0; i < 10; i++)
- red ++;
-
-#pragma acc loop vector reduction (+:red) // { dg-error "disallowed by containing routine" }
- for (int i = 0; i < 10; i++)
- red ++;
-
- return 1;
-}
@@ -1,5 +1,3 @@
-/* { dg-do compile } */
-
int
main ()
{
@@ -71,3 +69,259 @@ main ()
return 0;
}
+
+
+void par (void)
+{
+ int i, j;
+
+#pragma acc parallel
+ {
+#pragma acc loop tile // { dg-error "expected" }
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc loop tile() // { dg-error "expected" }
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc loop tile(1)
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc loop tile(*)
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc loop tile(2)
+ for (i = 0; i < 10; i++)
+ {
+ for (j = 1; j < 10; j++)
+ { }
+ }
+#pragma acc loop tile(-2) // { dg-warning "'tile' value must be positive" }
+ for (i = 1; i < 10; i++)
+ { }
+#pragma acc loop tile(i)
+ for (i = 1; i < 10; i++)
+ { }
+#pragma acc loop tile(2, 2, 1)
+ for (i = 1; i < 3; i++)
+ {
+ for (j = 4; j < 6; j++)
+ { }
+ }
+#pragma acc loop tile(2, 2)
+ for (i = 1; i < 5; i+=2)
+ {
+ for (j = i + 1; j < 7; j+=i)
+ { }
+ }
+#pragma acc loop vector tile(*)
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc loop worker tile(*)
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc loop gang tile(*)
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc loop vector gang tile(*)
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc loop vector worker tile(*)
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc loop gang worker tile(*)
+ for (i = 0; i < 10; i++)
+ { }
+ }
+}
+void p3 (void)
+{
+ int i, j;
+
+
+#pragma acc parallel loop tile // { dg-error "expected" }
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc parallel loop tile() // { dg-error "expected" }
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc parallel loop tile(1)
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc parallel loop tile(*)
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc parallel loop tile(*, 1)
+ for (i = 0; i < 10; i++)
+ {
+ for (j = 1; j < 10; j++)
+ { }
+ }
+#pragma acc parallel loop tile(-2) // { dg-warning "'tile' value must be positive" }
+ for (i = 1; i < 10; i++)
+ { }
+#pragma acc parallel loop tile(i)
+ for (i = 1; i < 10; i++)
+ { }
+#pragma acc parallel loop tile(2, 2, 1)
+ for (i = 1; i < 3; i++)
+ {
+ for (j = 4; j < 6; j++)
+ { }
+ }
+#pragma acc parallel loop tile(2, 2)
+ for (i = 1; i < 5; i+=2)
+ {
+ for (j = i + 1; j < 7; j++)
+ { }
+ }
+#pragma acc parallel loop vector tile(*)
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc parallel loop worker tile(*)
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc parallel loop gang tile(*)
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc parallel loop vector gang tile(*)
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc parallel loop vector worker tile(*)
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc parallel loop gang worker tile(*)
+ for (i = 0; i < 10; i++)
+ { }
+
+}
+
+
+void
+kern (void)
+{
+ int i, j;
+
+#pragma acc kernels
+ {
+#pragma acc loop tile // { dg-error "expected" }
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc loop tile() // { dg-error "expected" }
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc loop tile(1)
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc loop tile(2)
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc loop tile(6-2)
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc loop tile(6+2)
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc loop tile(*)
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc loop tile(*, 1)
+ for (i = 0; i < 10; i++)
+ {
+ for (j = 0; j < 10; i++)
+ { }
+ }
+#pragma acc loop tile(-2) // { dg-warning "'tile' value must be positive" }
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc loop tile(i)
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc loop tile(2, 2, 1)
+ for (i = 2; i < 4; i++)
+ for (i = 4; i < 6; i++)
+ { }
+#pragma acc loop tile(2, 2)
+ for (i = 1; i < 5; i+=2)
+ for (j = i+1; j < 7; i++)
+ { }
+#pragma acc loop vector tile(*)
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc loop worker tile(*)
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc loop gang tile(*)
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc loop vector gang tile(*)
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc loop vector worker tile(*)
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc loop gang worker tile(*)
+ for (i = 0; i < 10; i++)
+ { }
+ }
+}
+
+
+void k3 (void)
+{
+ int i, j;
+
+#pragma acc kernels loop tile // { dg-error "expected" }
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc kernels loop tile() // { dg-error "expected" }
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc kernels loop tile(1)
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc kernels loop tile(*)
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc kernels loop tile(*, 1)
+ for (i = 0; i < 10; i++)
+ {
+ for (j = 1; j < 10; j++)
+ { }
+ }
+#pragma acc kernels loop tile(-2) // { dg-warning "'tile' value must be positive" }
+ for (i = 1; i < 10; i++)
+ { }
+#pragma acc kernels loop tile(i)
+ for (i = 1; i < 10; i++)
+ { }
+#pragma acc kernels loop tile(2, 2, 1)
+ for (i = 1; i < 3; i++)
+ {
+ for (j = 4; j < 6; j++)
+ { }
+ }
+#pragma acc kernels loop tile(2, 2)
+ for (i = 1; i < 5; i++)
+ {
+ for (j = i + 1; j < 7; j += i)
+ { }
+ }
+#pragma acc kernels loop vector tile(*)
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc kernels loop worker tile(*)
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc kernels loop gang tile(*)
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc kernels loop vector gang tile(*)
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc kernels loop vector worker tile(*)
+ for (i = 0; i < 10; i++)
+ { }
+#pragma acc kernels loop gang worker tile(*)
+ for (i = 0; i < 10; i++)
+ { }
+}
deleted file mode 100644
@@ -1,14 +0,0 @@
-/* { dg-do compile } */
-
-void bar (float *, float *);
-
-void
-foo (float *x, float *y)
-{
- int n = 1 << 10;
-#pragma acc data create(x[0:n]) copyout(y[0:n])
- {
-#pragma acc host_data use_device(x,y)
- bar (x, y);
- }
-}
deleted file mode 100644
@@ -1,100 +0,0 @@
-extern void abort ();
-
-const int n = 100;
-
-// Check explicit template copy map
-
-template<typename T> T
-sum (T array[])
-{
- T s = 0;
-
-#pragma acc parallel loop num_gangs (10) gang reduction (+:s) copy (s, array[0:n])
- for (int i = 0; i < n; i++)
- s += array[i];
-
- return s;
-}
-
-// Check implicit template copy map
-
-template<typename T> T
-sum ()
-{
- T s = 0;
- T array[n];
-
- for (int i = 0; i < n; i++)
- array[i] = i+1;
-
-#pragma acc parallel loop num_gangs (10) gang reduction (+:s) copy (s)
- for (int i = 0; i < n; i++)
- s += array[i];
-
- return s;
-}
-
-// Check present and async
-
-template<typename T> T
-async_sum (T array[])
-{
- T s = 0;
-
-#pragma acc parallel loop num_gangs (10) gang async (1) present (array[0:n])
- for (int i = 0; i < n; i++)
- array[i] = i+1;
-
-#pragma acc parallel loop num_gangs (10) gang reduction (+:s) present (array[0:n]) copy (s) async wait (1)
- for (int i = 0; i < n; i++)
- s += array[i];
-
-#pragma acc wait
-
- return s;
-}
-
-// Check present and async and an explicit firstprivate
-
-template<typename T> T
-async_sum (int c)
-{
- T s = 0;
-
-#pragma acc parallel loop num_gangs (10) gang reduction (+:s) copy(s) firstprivate (c) async wait (1)
- for (int i = 0; i < n; i++)
- s += i+c;
-
-#pragma acc wait
-
- return s;
-}
-
-int
-main()
-{
- int a[n];
- int result = 0;
-
- for (int i = 0; i < n; i++)
- {
- a[i] = i+1;
- result += i+1;
- }
-
- if (sum (a) != result)
- abort ();
-
- if (sum<int> () != result)
- abort ();
-
-#pragma acc enter data copyin (a)
- if (async_sum (a) != result)
- abort ();
-
- if (async_sum<int> (1) != result)
- abort ();
-#pragma acc exit data delete (a)
-
- return 0;
-}
@@ -1,5 +1,3 @@
-/* { dg-additional-options "-w" } */
-
#pragma acc routine
template <typename T> T
accDouble(int val)
@@ -17,60 +15,62 @@ oacc_parallel_copy (T a)
double z = 4;
#pragma acc parallel num_gangs (a) num_workers (a) vector_length (a) default (none) copyout (b) copyin (a)
- {
+#pragma acc loop gang worker vector
+ for (int i = 0; i < 1; i++)
b = a;
- }
#pragma acc parallel num_gangs (a) copy (w, x, y, z)
- {
- w = accDouble<char>(w);
- x = accDouble<int>(x);
- y = accDouble<float>(y);
- z = accDouble<double>(z);
- }
+#pragma acc loop
+ for (int i = 0; i < 1; i++)
+ {
+ w = accDouble<char>(w);
+ x = accDouble<int>(x);
+ y = accDouble<float>(y);
+ z = accDouble<double>(z);
+ }
#pragma acc parallel num_gangs (a) if (1)
{
#pragma acc loop independent collapse (2) device_type (nvidia) gang
- for (int i = 0; i < a; i++)
- for (int j = 0; j < 5; j++)
- b = a;
+ for (int i = 0; i < a; i++)
+ for (int j = 0; j < 5; j++)
+ b = a;
#pragma acc loop auto tile (a, 3)
- for (int i = 0; i < a; i++)
- for (int j = 0; j < 5; j++)
- b = a;
+ for (int i = 0; i < a; i++)
+ for (int j = 0; j < 5; j++)
+ b = a;
#pragma acc loop seq
- for (int i = 0; i < a; i++)
- b = a;
+ for (int i = 0; i < a; i++)
+ b = a;
}
T c;
#pragma acc parallel num_workers (10)
- {
+#pragma acc loop worker
+ for (int i = 0; i < 1; i++)
+ {
#pragma acc atomic capture
- c = b++;
+ c = b++;
#pragma atomic update
- c++;
+ c++;
#pragma acc atomic read
- b = a;
+ b = a;
#pragma acc atomic write
- b = a;
- }
+ b = a;
+ }
#pragma acc parallel reduction (+:c)
- {
- c = 1;
- }
+ c = 1;
#pragma acc data if (1) copy (b)
{
- #pragma acc parallel
+#pragma acc parallel
{
b = a;
}
@@ -78,9 +78,9 @@ oacc_parallel_copy (T a)
#pragma acc enter data copyin (b)
#pragma acc parallel present (b)
- {
- b = a;
- }
+ {
+ b = a;
+ }
#pragma acc update host (b)
#pragma acc update self (b)
@@ -113,9 +113,7 @@ oacc_kernels_copy (T a)
#pragma acc kernels loop reduction (+:c)
for (int i = 0; i < 10; i++)
- {
- c = 1;
- }
+ c = 1;
#pragma acc data if (1) copy (b)
{
@@ -127,9 +125,10 @@ oacc_kernels_copy (T a)
#pragma acc enter data copyin (b)
#pragma acc kernels present (b)
- {
- b = a;
- }
+ {
+ b = a;
+ }
+
return b;
}
@@ -11,7 +11,7 @@
subroutine test
implicit none
- integer a(100), i, j, z
+ integer a(100), i, j, y, z
! PARALLEL
@@ -79,10 +79,10 @@ subroutine test
end do
!$acc end parallel loop
-! !$acc parallel loop reduction (+:z) copy (z)
-! do i = 1, 100
-! end do
-! !$acc end parallel loop
+ !$acc parallel loop reduction (+:y) copy (y)
+ do i = 1, 100
+ end do
+ !$acc end parallel loop
! KERNELS
@@ -150,10 +150,10 @@ subroutine test
end do
!$acc end kernels loop
-! !$acc kernels loop reduction (+:z) copy (z)
-! do i = 1, 100
-! end do
-! !$acc end kernels loop
+ !$acc kernels loop reduction (+:y) copy (y)
+ do i = 1, 100
+ end do
+ !$acc end kernels loop
end subroutine test
! { dg-final { scan-tree-dump-times "acc loop private.i. private.j. collapse.2." 2 "gimple" { xfail *-*-* } } }
@@ -165,3 +165,5 @@ end subroutine test
! { dg-final { scan-tree-dump-times "acc loop private.i. private.j. tile.2, 3" 2 "gimple" { xfail *-*-* } } }
! { dg-final { scan-tree-dump-times "acc loop private.i. independent" 2 "gimple" { xfail *-*-* } } }
! { dg-final { scan-tree-dump-times "private.z" 2 "gimple" } }
+! { dg-final { scan-tree-dump-times "omp target oacc_\[^ \]+ map.force_tofrom:y" 2 "gimple" { xfail *-*-* } } }
+! { dg-final { scan-tree-dump-times "acc loop private.i. reduction..:y." 2 "gimple" { xfail *-*-* } } }
@@ -1,5 +1,3 @@
-! { dg-do compile }
-! { dg-additional-options "-fmax-errors=100" }
module test
implicit none
contains
@@ -29,14 +27,18 @@ subroutine test1
i = i + 1
end do
!$acc loop
- do 300 d = 1, 30, 6 ! { dg-error "integer" }
+ do 300 d = 1, 30, 6
i = d
300 a(i) = 1
+ ! { dg-warning "Deleted feature: Loop variable at .1. must be integer" "" { target *-*-* } 30 }
+ ! { dg-error "ACC LOOP iteration variable must be of type integer" "" { target *-*-* } 30 }
!$acc loop
- do d = 1, 30, 5 ! { dg-error "integer" }
+ do d = 1, 30, 5
i = d
a(i) = 2
end do
+ ! { dg-warning "Deleted feature: Loop variable at .1. must be integer" "" { target *-*-* } 36 }
+ ! { dg-error "ACC LOOP iteration variable must be of type integer" "" { target *-*-* } 36 }
!$acc loop
do i = 1, 30
if (i .eq. 16) exit ! { dg-error "EXIT statement" }
@@ -144,8 +146,10 @@ subroutine test1
end do
!$acc parallel loop collapse(2)
do i = 1, 3
- do r = 4, 6 ! { dg-error "integer" }
+ do r = 4, 6
end do
+ ! { dg-warning "Deleted feature: Loop variable at .1. must be integer" "" { target *-*-* } 149 }
+ ! { dg-error "ACC LOOP iteration variable must be of type integer" "" { target *-*-* } 149 }
end do
! Both seq and independent are not allowed
@@ -167,4 +171,3 @@ subroutine test1
end subroutine test1
end module test
-! { dg-prune-output "Deleted" }
@@ -1,6 +1,3 @@
-! { dg-do compile }
-! { dg-additional-options "-fmax-errors=100" }
-
program test
implicit none
integer :: i, j
@@ -1,11 +1,3 @@
-! { dg-do compile }
-! { dg-additional-options "-fmax-errors=100" }
-
-! This error is temporary. Remove when support is added for these clauses
-! in the middle end.
-! { dg-prune-output "sorry, unimplemented" }
-! { dg-prune-output "Error: work-sharing region" }
-
program test
implicit none
integer :: i, j
@@ -1,4 +1,3 @@
-! { dg-do compile }
! { dg-additional-options "-fdump-tree-original -std=f2008" }
! test for tree-dump-original and spaces-commas
@@ -1,5 +1,4 @@
-! { dg-do compile }
-! { dg-additional-options "-fdump-tree-original -w" }
+! { dg-additional-options "-fdump-tree-original" }
! test for tree-dump-original and spaces-commas
@@ -12,9 +11,13 @@ program test
!$acc reduction(max:q), copy(i), copyin(j), copyout(k), create(m) &
!$acc present(o), pcopy(p), pcopyin(r), pcopyout(s), pcreate(t) &
!$acc deviceptr(u), private(v), firstprivate(w)
+ ! { dg-warning "region is gang partitioned but does not contain gang partitioned code" "" { target *-*-* } 13 }
+ ! { dg-warning "region is worker partitioned but does not contain worker partitioned code" "" { target *-*-* } 13 }
+ ! { dg-warning "region is vector partitioned but does not contain vector partitioned code" "" { target *-*-* } 13 }
!$acc end parallel
end program test
+
! { dg-final { scan-tree-dump-times "pragma acc parallel" 1 "original" } }
! { dg-final { scan-tree-dump-times "if" 1 "original" } }
@@ -1,3 +1,71 @@
+2016-03-30 Thomas Schwinge <thomas@codesourcery.com>
+ James Norris <jnorris@codesourcery.com>
+ Nathan Sidwell <nathan@codesourcery.com>
+ Julian Brown <julian@codesourcery.com>
+ Cesar Philippidis <cesar@codesourcery.com>
+ Chung-Lin Tang <cltang@codesourcery.com>
+ Tom de Vries <tom@codesourcery.com>
+
+ * testsuite/libgomp.oacc-c-c++-common/clauses-1.c: Update.
+ * testsuite/libgomp.oacc-c-c++-common/deviceptr-1.c: Likewise.
+ * testsuite/libgomp.oacc-c-c++-common/if-1.c: Likewise.
+ * testsuite/libgomp.oacc-c-c++-common/vector-loop.c: Likewise.
+ * testsuite/libgomp.oacc-fortran/asyncwait-1.f90: Likewise.
+ * testsuite/libgomp.oacc-fortran/asyncwait-2.f90: Likewise.
+ * testsuite/libgomp.oacc-fortran/asyncwait-3.f90: Likewise.
+ * testsuite/libgomp.oacc-fortran/declare-1.f90: Likewise.
+ * testsuite/libgomp.oacc-c-c++-common/asyncwait-1.c: Likewise.
+ XFAIL.
+ * testsuite/libgomp.oacc-c-c++-common/firstprivate-1.c: Update.
+ Incorporate...
+ * testsuite/libgomp.oacc-c-c++-common/firstprivate-2.c: ... this
+ file.
+ * testsuite/libgomp.oacc-c++/template-reduction.C: New file.
+ * testsuite/libgomp.oacc-c-c++-common/gang-static-1.c: Likewise.
+ * testsuite/libgomp.oacc-c-c++-common/gang-static-2.c: Likewise.
+ * testsuite/libgomp.oacc-c-c++-common/kernels-loop-clauses.c:
+ Likewise.
+ * testsuite/libgomp.oacc-c-c++-common/private-variables.c:
+ Likewise.
+ * testsuite/libgomp.oacc-c-c++-common/reduction-7.c: Likewise.
+ * testsuite/libgomp.oacc-c-c++-common/routine-1.c: Likewise.
+ * testsuite/libgomp.oacc-c-c++-common/routine-4.c: Likewise.
+ * testsuite/libgomp.oacc-c-c++-common/routine-wv-2.c: Likewise.
+ * testsuite/libgomp.oacc-fortran/clauses-1.f90: Likewise.
+ * testsuite/libgomp.oacc-fortran/default-1.f90: Likewise.
+ * testsuite/libgomp.oacc-fortran/firstprivate-1.f90: Likewise.
+ * testsuite/libgomp.oacc-fortran/gang-static-1.f90: Likewise.
+ * testsuite/libgomp.oacc-fortran/if-1.f90: Likewise.
+ * testsuite/libgomp.oacc-fortran/implicit-firstprivate-ref.f90:
+ Likewise.
+ * testsuite/libgomp.oacc-fortran/pr68813.f90: Likewise.
+ * testsuite/libgomp.oacc-fortran/private-variables.f90: Likewise.
+ * testsuite/libgomp.oacc-c-c++-common/kernels-1.c: Merge this
+ file...
+ * testsuite/libgomp.oacc-c-c++-common/parallel-1.c: ..., and this
+ file into...
+ * testsuite/libgomp.oacc-c-c++-common/data-clauses.h: ... this new
+ file. Update.
+ * testsuite/libgomp.oacc-c-c++-common/data-clauses-kernels.c: New
+ file.
+ * testsuite/libgomp.oacc-c-c++-common/data-clauses-parallel.c:
+ Likewise.
+ * testsuite/libgomp.oacc-c-c++-common/kernels-2.c: Rename to...
+ * testsuite/libgomp.oacc-c-c++-common/data-clauses-kernels-ipa-pta.c:
+ ... this new file. Update.
+ * testsuite/libgomp.oacc-c-c++-common/parallel-2.c: Rename to...
+ * testsuite/libgomp.oacc-c-c++-common/data-clauses-parallel-ipa-pta.c:
+ ... this new file. Update.
+ * testsuite/libgomp.oacc-c-c++-common/mode-transitions.c: New
+ file. Incorporate...
+ * testsuite/libgomp.oacc-c-c++-common/worker-single-1a.c: ... this
+ file, and...
+ * testsuite/libgomp.oacc-c-c++-common/worker-single-4.c: ... this
+ file, and...
+ * testsuite/libgomp.oacc-c-c++-common/worker-single-6.c: ... this
+ file.
+ * testsuite/libgomp.oacc-c-c++-common/update-1-2.c: Remove file.
+
2016-03-29 Thomas Schwinge <thomas@codesourcery.com>
* testsuite/libgomp.oacc-c++/c++.exp [!lang_test_file_found]: Call
@@ -1,3 +1,212 @@
+2016-04-04 Thomas Schwinge <thomas@codesourcery.com>
+
+ * testsuite/libgomp.oacc-c-c++-common/asyncwait-1.c: Don't XFAIL.
+ * testsuite/libgomp.oacc-c-c++-common/routine-1.c: Extend testing
+ to cover more parallelism levels, and asynchronous kernel
+ launches.
+ * testsuite/libgomp.oacc-c-c++-common/data-3.c: Remove OpenACC
+ present directives.
+ * testsuite/libgomp.oacc-fortran/collapse-5.f90: Remove OpenACC
+ copy directives.
+ * testsuite/libgomp.oacc-fortran/collapse-6.f90: Likewise.
+ * testsuite/libgomp.oacc-fortran/collapse-7.f90: Likewise.
+ * testsuite/libgomp.oacc-fortran/collapse-8.f90: Likewise.
+ * testsuite/libgomp.oacc-c-c++-common/firstprivate-3.c: Remove
+ file.
+ * testsuite/libgomp.oacc-c-c++-common/firstprivate-4.c: Remove
+ file, moving its content into...
+ * testsuite/libgomp.oacc-c-c++-common/firstprivate-1.c: ... this
+ file.
+ * testsuite/libgomp.oacc-c-c++-common/private-vars-par-gang-1.c:
+ Remove file, moving its content into...
+ * testsuite/libgomp.oacc-c-c++-common/firstprivate-1.c: ... this
+ file.
+ * testsuite/libgomp.oacc-c-c++-common/kernels-loop-4.c: Rename
+ file to...
+ * testsuite/libgomp.oacc-c-c++-common/kernels-loop-clauses.c:
+ ... this file. Clean up dg-* directives.
+ * testsuite/libgomp.oacc-c-c++-common/loop-auto-1.c: Use
+ dg-warning directives instead of specifying the -w compiler
+ option.
+ * testsuite/libgomp.oacc-c-c++-common/loop-w-1.c: Likewise.
+ * testsuite/libgomp.oacc-c-c++-common/routine-g-1.c: Likewise.
+ * testsuite/libgomp.oacc-c-c++-common/routine-w-1.c: Likewise.
+ * testsuite/libgomp.oacc-c-c++-common/routine-4.c: Change
+ parallelism used instead of specifying the -w compiler option.
+ * testsuite/libgomp.oacc-fortran/routine-7.f90: Likewise.
+ * testsuite/libgomp.oacc-c-c++-common/loop-reduction-gang-np-1.c:
+ Merge this file, and...
+ * testsuite/libgomp.oacc-c-c++-common/loop-reduction-gv-np-1.c:
+ ... this file, and...
+ * testsuite/libgomp.oacc-c-c++-common/loop-reduction-gw-np-1.c:
+ ... this file, and...
+ * testsuite/libgomp.oacc-c-c++-common/loop-reduction-gwv-np-1.c:
+ ... this file, and...
+ * testsuite/libgomp.oacc-c-c++-common/loop-reduction-gwv-np-2.c:
+ ... this file, and...
+ * testsuite/libgomp.oacc-c-c++-common/loop-reduction-gwv-np-3.c:
+ ... this file, and...
+ * testsuite/libgomp.oacc-c-c++-common/loop-reduction-gwv-np-4.c:
+ ... this file, and...
+ * testsuite/libgomp.oacc-c-c++-common/loop-reduction-vector-p-1.c:
+ ... this file, and...
+ * testsuite/libgomp.oacc-c-c++-common/loop-reduction-vector-p-2.c:
+ ... this file, and...
+ * testsuite/libgomp.oacc-c-c++-common/loop-reduction-worker-p-1.c:
+ ... this file, and...
+ * testsuite/libgomp.oacc-c-c++-common/loop-reduction-wv-p-1.c:
+ ... this file, and...
+ * testsuite/libgomp.oacc-c-c++-common/loop-reduction-wv-p-2.c:
+ ... this file, and...
+ * testsuite/libgomp.oacc-c-c++-common/loop-reduction-wv-p-3.c:
+ ... this file into...
+ * testsuite/libgomp.oacc-c-c++-common/reduction-7.c: ... this new
+ file. Use dg-warning directives instead of specifying the -w
+ compiler option.
+ * testsuite/libgomp.oacc-c-c++-common/vec-partn-1.c: Merge this
+ file, and...
+ * testsuite/libgomp.oacc-c-c++-common/vec-partn-2.c: ... this
+ file, and...
+ * testsuite/libgomp.oacc-c-c++-common/vec-partn-3.c: ... this
+ file, and...
+ * testsuite/libgomp.oacc-c-c++-common/vec-partn-4.c: ... this
+ file, and...
+ * testsuite/libgomp.oacc-c-c++-common/vec-partn-5.c: ... this
+ file, and...
+ * testsuite/libgomp.oacc-c-c++-common/vec-partn-6.c: ... this
+ file, and...
+ * testsuite/libgomp.oacc-c-c++-common/vec-single-1.c: ... this
+ file, and...
+ * testsuite/libgomp.oacc-c-c++-common/vec-single-2.c: ... this
+ file, and...
+ * testsuite/libgomp.oacc-c-c++-common/vec-single-3.c: ... this
+ file, and...
+ * testsuite/libgomp.oacc-c-c++-common/vec-single-4.c: ... this
+ file, and...
+ * testsuite/libgomp.oacc-c-c++-common/vec-single-5.c: ... this
+ file, and...
+ * testsuite/libgomp.oacc-c-c++-common/vec-single-6.c: ... this
+ file, and...
+ * testsuite/libgomp.oacc-c-c++-common/vector-broadcast.c: ... this
+ file, and...
+ * testsuite/libgomp.oacc-c-c++-common/worker-partn-1.c: ... this
+ file, and...
+ * testsuite/libgomp.oacc-c-c++-common/worker-partn-2.c: ... this
+ file, and...
+ * testsuite/libgomp.oacc-c-c++-common/worker-partn-3.c: ... this
+ file, and...
+ * testsuite/libgomp.oacc-c-c++-common/worker-partn-4.c: ... this
+ file, and...
+ * testsuite/libgomp.oacc-c-c++-common/worker-partn-5.c: ... this
+ file, and...
+ * testsuite/libgomp.oacc-c-c++-common/worker-partn-6.c: ... this
+ file, and...
+ * testsuite/libgomp.oacc-c-c++-common/worker-partn-7.c: ... this
+ file, and...
+ * testsuite/libgomp.oacc-c-c++-common/worker-partn-8.c: ... this
+ file, and...
+ * testsuite/libgomp.oacc-c-c++-common/worker-single-1.c: ... this
+ file, and...
+ * testsuite/libgomp.oacc-c-c++-common/worker-single-1a.c: ... this
+ file, and...
+ * testsuite/libgomp.oacc-c-c++-common/worker-single-2.c: ... this
+ file, and...
+ * testsuite/libgomp.oacc-c-c++-common/worker-single-3.c: ... this
+ file, and...
+ * testsuite/libgomp.oacc-c-c++-common/worker-single-4.c: ... this
+ file, and...
+ * testsuite/libgomp.oacc-c-c++-common/worker-single-5.c: ... this
+ file, and...
+ * testsuite/libgomp.oacc-c-c++-common/worker-single-6.c: ... this
+ file, and...
+ * testsuite/libgomp.oacc-c-c++-common/mode-transitions.c: ... this
+ new file. Use dg-warning directives instead of specifying the -w
+ compiler option.
+ * testsuite/libgomp.oacc-c-c++-common/private-vars-local-gang-1.c:
+ Merge this file, and...
+ * testsuite/libgomp.oacc-c-c++-common/private-vars-local-worker-1.c:
+ ... this file, and...
+ * testsuite/libgomp.oacc-c-c++-common/private-vars-local-worker-2.c:
+ ... this file, and...
+ * testsuite/libgomp.oacc-c-c++-common/private-vars-local-worker-3.c:
+ ... this file, and...
+ * testsuite/libgomp.oacc-c-c++-common/private-vars-local-worker-4.c:
+ ... this file, and...
+ * testsuite/libgomp.oacc-c-c++-common/private-vars-local-worker-5.c:
+ ... this file, and...
+ * testsuite/libgomp.oacc-c-c++-common/private-vars-loop-gang-1.c:
+ ... this file, and...
+ * testsuite/libgomp.oacc-c-c++-common/private-vars-loop-gang-2.c:
+ ... this file, and...
+ * testsuite/libgomp.oacc-c-c++-common/private-vars-loop-gang-3.c:
+ ... this file, and...
+ * testsuite/libgomp.oacc-c-c++-common/private-vars-loop-gang-4.c:
+ ... this file, and...
+ * testsuite/libgomp.oacc-c-c++-common/private-vars-loop-gang-5.c:
+ ... this file, and...
+ * testsuite/libgomp.oacc-c-c++-common/private-vars-loop-gang-6.c:
+ ... this file, and...
+ * testsuite/libgomp.oacc-c-c++-common/private-vars-loop-vector-1.c:
+ ... this file, and...
+ * testsuite/libgomp.oacc-c-c++-common/private-vars-loop-vector-2.c:
+ ... this file, and...
+ * testsuite/libgomp.oacc-c-c++-common/private-vars-loop-worker-1.c:
+ ... this file, and...
+ * testsuite/libgomp.oacc-c-c++-common/private-vars-loop-worker-2.c:
+ ... this file, and...
+ * testsuite/libgomp.oacc-c-c++-common/private-vars-loop-worker-3.c:
+ ... this file, and...
+ * testsuite/libgomp.oacc-c-c++-common/private-vars-loop-worker-4.c:
+ ... this file, and...
+ * testsuite/libgomp.oacc-c-c++-common/private-vars-loop-worker-5.c:
+ ... this file, and...
+ * testsuite/libgomp.oacc-c-c++-common/private-vars-loop-worker-6.c:
+ ... this file, and...
+ * testsuite/libgomp.oacc-c-c++-common/private-vars-loop-worker-7.c:
+ ... this file, and...
+ * testsuite/libgomp.oacc-c-c++-common/private-vars-par-gang-2.c:
+ ... this file, and...
+ * testsuite/libgomp.oacc-c-c++-common/private-vars-par-gang-3.c:
+ ... this file into...
+ * testsuite/libgomp.oacc-c-c++-common/private-variables.c:
+ ... this new file. Use dg-warning directives instead of
+ specifying the -w compiler option.
+ * testsuite/libgomp.oacc-fortran/private-vars-loop-gang-1.f90:
+ Merge this file, and...
+ * testsuite/libgomp.oacc-fortran/private-vars-loop-gang-2.f90:
+ ... this file, and...
+ * testsuite/libgomp.oacc-fortran/private-vars-loop-gang-3.f90:
+ ... this file, and...
+ * testsuite/libgomp.oacc-fortran/private-vars-loop-gang-6.f90:
+ ... this file, and...
+ * testsuite/libgomp.oacc-fortran/private-vars-loop-vector-1.f90:
+ ... this file, and...
+ * testsuite/libgomp.oacc-fortran/private-vars-loop-vector-2.f90:
+ ... this file, and...
+ * testsuite/libgomp.oacc-fortran/private-vars-loop-worker-1.f90:
+ ... this file, and...
+ * testsuite/libgomp.oacc-fortran/private-vars-loop-worker-2.f90:
+ ... this file, and...
+ * testsuite/libgomp.oacc-fortran/private-vars-loop-worker-3.f90:
+ ... this file, and...
+ * testsuite/libgomp.oacc-fortran/private-vars-loop-worker-4.f90:
+ ... this file, and...
+ * testsuite/libgomp.oacc-fortran/private-vars-loop-worker-5.f90:
+ ... this file, and...
+ * testsuite/libgomp.oacc-fortran/private-vars-loop-worker-6.f90:
+ ... this file, and...
+ * testsuite/libgomp.oacc-fortran/private-vars-loop-worker-7.f90:
+ ... this file, and...
+ * testsuite/libgomp.oacc-fortran/private-vars-par-gang-2.f90:
+ ... this file into...
+ * testsuite/libgomp.oacc-fortran/private-variables.f90: ... this
+ new file. Use dg-warning directives instead of specifying the -w
+ compiler option.
+ * testsuite/libgomp.oacc-c-c++-common/routine-2.c: Remove file.
+ * testsuite/libgomp.oacc-c-c++-common/routine-vec-1.c: Likewise.
+ * testsuite/libgomp.oacc-c-c++-common/routine-work-1.c: Likewise.
+ * testsuite/libgomp.oacc-fortran/update-1-2.f90: Likewise.
+
2016-03-24 Thomas Schwinge <thomas@codesourcery.com>
* testsuite/libgomp.oacc-c-c++-common/par-loop-comb-reduction-2.c:
@@ -1,7 +1,3 @@
-/* { dg-do run } */
-
-#include <cstdlib>
-
const int n = 100;
// Check explicit template copy map
@@ -85,17 +81,17 @@ main()
}
if (sum (a) != result)
- abort ();
+ __builtin_abort ();
if (sum<int> () != result)
- abort ();
+ __builtin_abort ();
#pragma acc enter data copyin (a)
if (async_sum (a) != result)
- abort ();
+ __builtin_abort ();
if (async_sum<int> (1) != result)
- abort ();
+ __builtin_abort ();
#pragma acc exit data delete (a)
return 0;
@@ -1,6 +1,4 @@
/* { dg-do run { target openacc_nvidia_accel_selected } } */
-/* <http://news.gmane.org/find-root.php?message_id=%3C87pp0aaksc.fsf%40kepler.schwinge.homeip.net%3E>.
- { dg-xfail-run-if "TODO" { *-*-* } } */
/* { dg-additional-options "-lcuda" } */
#include <openacc.h>
@@ -1,5 +1,7 @@
/* This test exercises combined directives. */
+/* { dg-do run } */
+
#include <stdlib.h>
int
@@ -25,7 +25,7 @@ main (int argc, char **argv)
}
#pragma acc enter data copyin (a[0:N]) copyin (b[0:N]) copyin (N) async
-#pragma acc parallel async wait present (a[0:N]) present (b[0:N]) present (N)
+#pragma acc parallel async wait
#pragma acc loop
for (i = 0; i < N; i++)
b[i] = a[i];
@@ -49,7 +49,7 @@ main (int argc, char **argv)
}
#pragma acc update device (a[0:N], b[0:N]) async (1)
-#pragma acc parallel async (1) present (a[0:N]) present (b[0:N]) present (N)
+#pragma acc parallel async (1)
#pragma acc loop
for (i = 0; i < N; i++)
b[i] = a[i];
@@ -78,17 +78,17 @@ main (int argc, char **argv)
#pragma acc update device (b[0:N]) async (2)
#pragma acc enter data copyin (c[0:N], d[0:N]) async (3)
-#pragma acc parallel async (1) wait (1,2) present (a[0:N]) present (b[0:N]) present (c[0:N]) present (d[0:N]) present (N)
+#pragma acc parallel async (1) wait (1,2)
#pragma acc loop
for (i = 0; i < N; i++)
b[i] = (a[i] * a[i] * a[i]) / a[i];
-#pragma acc parallel async (2) wait (1,3) present (a[0:N]) present (b[0:N]) present (c[0:N]) present (d[0:N]) present (N)
+#pragma acc parallel async (2) wait (1,3)
#pragma acc loop
for (i = 0; i < N; i++)
c[i] = (a[i] + a[i] + a[i] + a[i]) / a[i];
-#pragma acc parallel async (3) wait (1,3) present (a[0:N]) present (b[0:N]) present (c[0:N]) present (d[0:N]) present (N)
+#pragma acc parallel async (3) wait (1,3)
#pragma acc loop
for (i = 0; i < N; i++)
d[i] = ((a[i] * a[i] + a[i]) / a[i]) - a[i];
@@ -123,19 +123,19 @@ main (int argc, char **argv)
#pragma acc update device (a[0:N], b[0:N], c[0:N], d[0:N]) async (1)
#pragma acc enter data copyin (e[0:N]) async (5)
-#pragma acc parallel async (1) wait (1) present (a[0:N]) present (b[0:N]) present (c[0:N]) present (d[0:N]) present (e[0:N]) present (N)
+#pragma acc parallel async (1) wait (1)
for (int ii = 0; ii < N; ii++)
b[ii] = (a[ii] * a[ii] * a[ii]) / a[ii];
-#pragma acc parallel async (2) wait (1) present (a[0:N]) present (b[0:N]) present (c[0:N]) present (d[0:N]) present (e[0:N]) present (N)
+#pragma acc parallel async (2) wait (1)
for (int ii = 0; ii < N; ii++)
c[ii] = (a[ii] + a[ii] + a[ii] + a[ii]) / a[ii];
-#pragma acc parallel async (3) wait (1) present (a[0:N]) present (b[0:N]) present (c[0:N]) present (d[0:N]) present (e[0:N]) present (N)
+#pragma acc parallel async (3) wait (1)
for (int ii = 0; ii < N; ii++)
d[ii] = ((a[ii] * a[ii] + a[ii]) / a[ii]) - a[ii];
-#pragma acc parallel wait (1,5) async (4) present (a[0:N]) present (b[0:N]) present (c[0:N]) present (d[0:N]) present (e[0:N]) present (N)
+#pragma acc parallel wait (1,5) async (4)
for (int ii = 0; ii < N; ii++)
e[ii] = a[ii] + b[ii] + c[ii] + d[ii];
similarity index 86%
rename from libgomp/testsuite/libgomp.oacc-c-c++-common/kernels-2.c
rename to libgomp/testsuite/libgomp.oacc-c-c++-common/data-clauses-kernels-ipa-pta.c
@@ -3,4 +3,4 @@
/* Override the compiler's "avoid offloading" decision.
{ dg-additional-options "-foffload-force" } */
-#include "kernels-1.c"
+#include "data-clauses-kernels.c"
similarity index 71%
rename from libgomp/testsuite/libgomp.oacc-c-c++-common/kernels-1.c
rename to libgomp/testsuite/libgomp.oacc-c-c++-common/data-clauses-kernels.c
@@ -1,7 +1,5 @@
/* Override the compiler's "avoid offloading" decision.
{ dg-additional-options "-foffload-force" } */
-#include <stdlib.h>
-
-#define EXEC_DIRECTIVE kernels
+#define CONSTRUCT kernels
#include "data-clauses.h"
similarity index 75%
rename from libgomp/testsuite/libgomp.oacc-c-c++-common/parallel-2.c
rename to libgomp/testsuite/libgomp.oacc-c-c++-common/data-clauses-parallel-ipa-pta.c
@@ -1,4 +1,4 @@
/* { dg-do run { target lto } } */
/* { dg-additional-options "-fipa-pta -flto -flto-partition=max" } */
-#include "parallel-1.c"
+#include "data-clauses-parallel.c"
new file mode 100644
@@ -0,0 +1,2 @@
+#define CONSTRUCT parallel
+#include "data-clauses.h"
@@ -7,145 +7,145 @@ int main(void)
i = -1;
j = -2;
v = 0;
-#pragma acc EXEC_DIRECTIVE /* copyout */ present_or_copyout (v) copyin (i, j)
+#pragma acc CONSTRUCT /* copyout */ present_or_copyout (v) copyin (i, j)
{
if (i != -1 || j != -2)
- abort ();
+ __builtin_abort ();
i = 2;
j = 1;
if (i != 2 || j != 1)
- abort ();
+ __builtin_abort ();
v = 1;
}
#if ACC_MEM_SHARED
if (v != 1 || i != 2 || j != 1)
- abort ();
+ __builtin_abort ();
#else
if (v != 1 || i != -1 || j != -2)
- abort ();
+ __builtin_abort ();
#endif
i = -1;
j = -2;
v = 0;
-#pragma acc EXEC_DIRECTIVE /* copyout */ present_or_copyout (v) copyout (i, j)
+#pragma acc CONSTRUCT /* copyout */ present_or_copyout (v) copyout (i, j)
{
i = 2;
j = 1;
if (i != 2 || j != 1)
- abort ();
+ __builtin_abort ();
v = 1;
}
if (v != 1 || i != 2 || j != 1)
- abort ();
+ __builtin_abort ();
i = -1;
j = -2;
v = 0;
-#pragma acc EXEC_DIRECTIVE /* copyout */ present_or_copyout (v) copy (i, j)
+#pragma acc CONSTRUCT /* copyout */ present_or_copyout (v) copy (i, j)
{
if (i != -1 || j != -2)
- abort ();
+ __builtin_abort ();
i = 2;
j = 1;
if (i != 2 || j != 1)
- abort ();
+ __builtin_abort ();
v = 1;
}
if (v != 1 || i != 2 || j != 1)
- abort ();
+ __builtin_abort ();
i = -1;
j = -2;
v = 0;
-#pragma acc EXEC_DIRECTIVE /* copyout */ present_or_copyout (v) create (i, j)
+#pragma acc CONSTRUCT /* copyout */ present_or_copyout (v) create (i, j)
{
i = 2;
j = 1;
if (i != 2 || j != 1)
- abort ();
+ __builtin_abort ();
v = 1;
}
#if ACC_MEM_SHARED
if (v != 1 || i != 2 || j != 1)
- abort ();
+ __builtin_abort ();
#else
if (v != 1 || i != -1 || j != -2)
- abort ();
+ __builtin_abort ();
#endif
i = -1;
j = -2;
v = 0;
-#pragma acc EXEC_DIRECTIVE /* copyout */ present_or_copyout (v) present_or_copyin (i, j)
+#pragma acc CONSTRUCT /* copyout */ present_or_copyout (v) present_or_copyin (i, j)
{
if (i != -1 || j != -2)
- abort ();
+ __builtin_abort ();
i = 2;
j = 1;
if (i != 2 || j != 1)
- abort ();
+ __builtin_abort ();
v = 1;
}
if (v != 1)
- abort ();
+ __builtin_abort ();
#if ACC_MEM_SHARED
if (v != 1 || i != 2 || j != 1)
- abort ();
+ __builtin_abort ();
#else
if (v != 1 || i != -1 || j != -2)
- abort ();
+ __builtin_abort ();
#endif
i = -1;
j = -2;
v = 0;
-#pragma acc EXEC_DIRECTIVE /* copyout */ present_or_copyout (v) present_or_copyout (i, j)
+#pragma acc CONSTRUCT /* copyout */ present_or_copyout (v) present_or_copyout (i, j)
{
i = 2;
j = 1;
if (i != 2 || j != 1)
- abort ();
+ __builtin_abort ();
v = 1;
}
if (v != 1 || i != 2 || j != 1)
- abort ();
+ __builtin_abort ();
i = -1;
j = -2;
v = 0;
-#pragma acc EXEC_DIRECTIVE /* copyout */ present_or_copyout (v) present_or_copy (i, j)
+#pragma acc CONSTRUCT /* copyout */ present_or_copyout (v) present_or_copy (i, j)
{
if (i != -1 || j != -2)
- abort ();
+ __builtin_abort ();
i = 2;
j = 1;
if (i != 2 || j != 1)
- abort ();
+ __builtin_abort ();
v = 1;
}
if (v != 1 || i != 2 || j != 1)
- abort ();
+ __builtin_abort ();
i = -1;
j = -2;
v = 0;
-#pragma acc EXEC_DIRECTIVE /* copyout */ present_or_copyout (v) present_or_create (i, j)
+#pragma acc CONSTRUCT /* copyout */ present_or_copyout (v) present_or_create (i, j)
{
i = 2;
j = 1;
if (i != 2 || j != 1)
- abort ();
+ __builtin_abort ();
v = 1;
}
if (v != 1)
- abort ();
+ __builtin_abort ();
#if ACC_MEM_SHARED
if (v != 1 || i != 2 || j != 1)
- abort ();
+ __builtin_abort ();
#else
if (v != 1 || i != -1 || j != -2)
- abort ();
+ __builtin_abort ();
#endif
i = -1;
@@ -154,23 +154,23 @@ int main(void)
#pragma acc data copyin (i, j)
{
-#pragma acc EXEC_DIRECTIVE /* copyout */ present_or_copyout (v) present (i, j)
+#pragma acc CONSTRUCT /* copyout */ present_or_copyout (v) present (i, j)
{
if (i != -1 || j != -2)
- abort ();
+ __builtin_abort ();
i = 2;
j = 1;
if (i != 2 || j != 1)
- abort ();
+ __builtin_abort ();
v = 1;
}
}
#if ACC_MEM_SHARED
if (v != 1 || i != 2 || j != 1)
- abort ();
+ __builtin_abort ();
#else
if (v != 1 || i != -1 || j != -2)
- abort ();
+ __builtin_abort ();
#endif
i = -1;
@@ -179,23 +179,23 @@ int main(void)
#pragma acc data copyin(i, j)
{
-#pragma acc EXEC_DIRECTIVE /* copyout */ present_or_copyout (v)
+#pragma acc CONSTRUCT /* copyout */ present_or_copyout (v)
{
if (i != -1 || j != -2)
- abort ();
+ __builtin_abort ();
i = 2;
j = 1;
if (i != 2 || j != 1)
- abort ();
+ __builtin_abort ();
v = 1;
}
}
#if ACC_MEM_SHARED
if (v != 1 || i != 2 || j != 1)
- abort ();
+ __builtin_abort ();
#else
if (v != 1 || i != -1 || j != -2)
- abort ();
+ __builtin_abort ();
#endif
return 0;
@@ -1,8 +1,7 @@
-/* { dg-do run } */
-
#include <openacc.h>
-int main ()
+
+void t1 ()
{
int ok = 1;
int val = 2;
@@ -28,14 +27,117 @@ int main ()
if (ondev)
{
if (!ok)
- return 1;
+ __builtin_abort ();
if (val != 2)
- return 1;
+ __builtin_abort ();
for (int i = 0; i < 32; i++)
if (ary[i] != 2 + i)
- return 1;
+ __builtin_abort ();
}
-
+}
+
+
+void t2 ()
+{
+ int ok = 1;
+ int val = 2;
+
+#pragma acc data copy(val)
+ {
+#pragma acc parallel present (val)
+ {
+ val = 7;
+ }
+
+#pragma acc parallel firstprivate (val) copy(ok)
+ {
+ ok = val == 7;
+ val = 9;
+ }
+ }
+
+ if (!ok)
+ __builtin_abort ();
+ if (val != 7)
+ __builtin_abort ();
+}
+
+
+#define N 100
+void t3 ()
+{
+ int a, b[N], c, d, i;
+ int n = acc_get_device_type () == acc_device_nvidia ? N : 1;
+
+ a = 5;
+ for (i = 0; i < n; i++)
+ b[i] = -1;
+
+ #pragma acc parallel num_gangs (n) firstprivate (a)
+ #pragma acc loop gang
+ for (i = 0; i < n; i++)
+ {
+ a = a + i;
+ b[i] = a;
+ }
+
+ for (i = 0; i < n; i++)
+ if (a + i != b[i])
+ __builtin_abort ();
+
+ #pragma acc data copy (a)
+ {
+ #pragma acc parallel firstprivate (a) copyout (c)
+ {
+ a = 10;
+ c = a;
+ }
+
+ /* This version of 'a' should still be 5. */
+ #pragma acc parallel copyout (d) present (a)
+ {
+ d = a;
+ }
+ }
+
+ if (c != 10)
+ __builtin_abort ();
+ if (d != 5)
+ __builtin_abort ();
+}
+#undef N
+
+
+void t4 ()
+{
+ int x = 5, i, arr[32];
+
+ for (i = 0; i < 32; i++)
+ arr[i] = 3;
+
+#pragma acc parallel firstprivate(x) copy(arr) num_gangs(32) num_workers(8) vector_length(32)
+ /* { dg-warning "region is worker partitioned but does not contain worker partitioned code" "worker" { target *-*-* } 119 } */
+ /* { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 119 } */
+ {
+#pragma acc loop gang
+ for (i = 0; i < 32; i++)
+ arr[i] += x;
+ }
+
+ for (i = 0; i < 32; i++)
+ if (arr[i] != 8)
+ __builtin_abort ();
+}
+
+
+int
+main()
+{
+ t1 ();
+ t2 ();
+ t3 ();
+ t4 ();
+
return 0;
}
deleted file mode 100644
@@ -1,29 +0,0 @@
-#include <openacc.h>
-
-int main ()
-{
- int ok = 1;
- int val = 2;
-
-#pragma acc data copy(val)
- {
-#pragma acc parallel present (val)
- {
- val = 7;
- }
-
-#pragma acc parallel firstprivate (val) copy(ok)
- {
- ok = val == 7;
- val = 9;
- }
-
- }
-
- if (!ok)
- return 1;
- if(val != 7)
- return 1;
-
- return 0;
-}
deleted file mode 100644
@@ -1,31 +0,0 @@
-/* { dg-do run } */
-/* { dg-xfail-run-if "TODO" { openacc_host_selected } { "*" } { "" } } */
-
-#include <stdlib.h>
-
-#define n 100
-
-int
-main()
-{
- int a, b[n], i;
-
- a = 5;
-
- for (i = 0; i < n; i++)
- b[i] = -1;
-
- #pragma acc parallel num_gangs (n) firstprivate (a)
- #pragma acc loop gang
- for (i = 0; i < n; i++)
- {
- a = a + i;
- b[i] = a;
- }
-
- for (i = 0; i < n; i++)
- if (a + i != b[i])
- abort ();
-
- return 0;
-}
deleted file mode 100644
@@ -1,54 +0,0 @@
-/* { dg-do run } */
-
-#include <stdlib.h>
-#include <openacc.h>
-
-#define N 100
-
-int
-main()
-{
- int a, old_a, b[N], c, d, i;
- int n = acc_get_device_type () == acc_device_nvidia ? N : 1;
-
- a = 5;
-
- for (i = 0; i < n; i++)
- b[i] = -1;
-
- #pragma acc parallel num_gangs (n) firstprivate (a)
- #pragma acc loop gang
- for (i = 0; i < n; i++)
- {
- a = a + i;
- b[i] = a;
- }
-
- for (i = 0; i < n; i++)
- if (a + i != b[i])
- abort ();
-
- #pragma acc data copy (a)
- {
- #pragma acc parallel firstprivate (a) copyout (c)
- {
- a = 10;
- c = a;
- }
-
- /* This version of 'a' should still be 5. */
- #pragma acc parallel copyout (d) present (a)
- {
- d = a;
- }
-
- }
-
- if (c != 10)
- abort ();
-
- if (d != 5)
- abort ();
-
- return 0;
-}
similarity index 90%
rename from libgomp/testsuite/libgomp.oacc-c-c++-common/kernels-loop-4.c
rename to libgomp/testsuite/libgomp.oacc-c-c++-common/kernels-loop-clauses.c
@@ -1,8 +1,6 @@
/* Exercise the auto, independent, seq and tile loop clauses inside
kernels regions. */
-/* { dg-prune-output "insufficient partitioning available to parallelize loop" } */
-
#include <assert.h>
#define N 100
@@ -1,4 +1,3 @@
-/* { dg-additional-options "-w" } */
/* This code uses nvptx inline assembly guarded with acc_on_device, which is
not optimized away at -O0, and then confuses the target assembler.
{ dg-skip-if "" { *-*-* } { "-O0" } { "" } } */
@@ -103,6 +102,7 @@ int vector_1 (int *ary, int size)
clear (ary, size);
#pragma acc parallel num_workers (32) vector_length(32) copy(ary[0:size]) firstprivate (size)
+ /* { dg-warning "region is worker partitioned but does not contain worker partitioned code" "worker" { target *-*-* } 104 } */
{
#pragma acc loop gang
for (int jx = 0; jx < 1; jx++)
@@ -153,6 +153,7 @@ int gang_1 (int *ary, int size)
clear (ary, size);
#pragma acc parallel num_gangs (32) num_workers (32) vector_length(32) copy(ary[0:size]) firstprivate (size)
+ /* { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 155 } */
{
#pragma acc loop auto
for (int jx = 0; jx < size / 64; jx++)
@@ -187,6 +188,7 @@ int gang_3 (int *ary, int size)
clear (ary, size);
#pragma acc parallel num_workers (32) vector_length(32) copy(ary[0:size]) firstprivate (size)
+ /* { dg-warning "region is worker partitioned but does not contain worker partitioned code" "worker" { target *-*-* } 190 } */
{
#pragma acc loop auto
for (int jx = 0; jx < size / 64; jx++)
deleted file mode 100644
@@ -1,45 +0,0 @@
-/* { dg-additional-options "-w" } */
-
-#include <assert.h>
-
-/* Test of reduction on loop directive (gangs, non-private reduction
- variable). */
-
-int
-main (int argc, char *argv[])
-{
- int i, arr[1024], res = 0, hres = 0;
-
- for (i = 0; i < 1024; i++)
- arr[i] = i;
-
- #pragma acc parallel num_gangs(32) num_workers(32) vector_length(32) \
- copy(res)
- {
- #pragma acc loop gang reduction(+:res)
- for (i = 0; i < 1024; i++)
- res += arr[i];
- }
-
- for (i = 0; i < 1024; i++)
- hres += arr[i];
-
- assert (res == hres);
-
- res = hres = 1;
-
- #pragma acc parallel num_gangs(32) num_workers(32) vector_length(32) \
- copy(res)
- {
- #pragma acc loop gang reduction(*:res)
- for (i = 0; i < 12; i++)
- res *= arr[i];
- }
-
- for (i = 0; i < 12; i++)
- hres *= arr[i];
-
- assert (res == hres);
-
- return 0;
-}
deleted file mode 100644
@@ -1,30 +0,0 @@
-/* { dg-additional-options "-w" } */
-
-#include <assert.h>
-
-/* Test of reduction on loop directive (gangs and vectors, non-private
- reduction variable). */
-
-int
-main (int argc, char *argv[])
-{
- int i, arr[1024], res = 0, hres = 0;
-
- for (i = 0; i < 1024; i++)
- arr[i] = i;
-
- #pragma acc parallel num_gangs(32) num_workers(32) vector_length(32) \
- copy(res)
- {
- #pragma acc loop gang vector reduction(+:res)
- for (i = 0; i < 1024; i++)
- res += arr[i];
- }
-
- for (i = 0; i < 1024; i++)
- hres += arr[i];
-
- assert (res == hres);
-
- return 0;
-}
deleted file mode 100644
@@ -1,30 +0,0 @@
-/* { dg-additional-options "-w" } */
-
-#include <assert.h>
-
-/* Test of reduction on loop directive (gangs and workers, non-private
- reduction variable). */
-
-int
-main (int argc, char *argv[])
-{
- int i, arr[1024], res = 0, hres = 0;
-
- for (i = 0; i < 1024; i++)
- arr[i] = i;
-
- #pragma acc parallel num_gangs(32) num_workers(32) vector_length(32) \
- copy(res)
- {
- #pragma acc loop gang worker reduction(+:res)
- for (i = 0; i < 1024; i++)
- res += arr[i];
- }
-
- for (i = 0; i < 1024; i++)
- hres += arr[i];
-
- assert (res == hres);
-
- return 0;
-}
deleted file mode 100644
@@ -1,28 +0,0 @@
-#include <assert.h>
-
-/* Test of reduction on loop directive (gangs, workers and vectors, non-private
- reduction variable). */
-
-int
-main (int argc, char *argv[])
-{
- int i, arr[1024], res = 0, hres = 0;
-
- for (i = 0; i < 1024; i++)
- arr[i] = i;
-
- #pragma acc parallel num_gangs(32) num_workers(32) vector_length(32) \
- copy(res)
- {
- #pragma acc loop gang worker vector reduction(+:res)
- for (i = 0; i < 1024; i++)
- res += arr[i];
- }
-
- for (i = 0; i < 1024; i++)
- hres += arr[i];
-
- assert (res == hres);
-
- return 0;
-}
deleted file mode 100644
@@ -1,34 +0,0 @@
-#include <assert.h>
-
-/* Test of reduction on loop directive (gangs, workers and vectors, non-private
- reduction variable: separate gang and worker/vector loops). */
-
-int
-main (int argc, char *argv[])
-{
- int i, j, arr[32768], res = 0, hres = 0;
-
- for (i = 0; i < 32768; i++)
- arr[i] = i;
-
- #pragma acc parallel num_gangs(32) num_workers(32) vector_length(32) \
- copy(res)
- {
- #pragma acc loop gang reduction(+:res)
- for (j = 0; j < 32; j++)
- {
- #pragma acc loop worker vector reduction(+:res)
- for (i = 0; i < 1024; i++)
- res += arr[j * 1024 + i];
- }
- /* "res" is non-private, and is not available until after the parallel
- region. */
- }
-
- for (i = 0; i < 32768; i++)
- hres += arr[i];
-
- assert (res == hres);
-
- return 0;
-}
deleted file mode 100644
@@ -1,33 +0,0 @@
-#include <assert.h>
-
-/* Test of reduction on loop directive (gangs, workers and vectors, non-private
- reduction variable: separate gang and worker/vector loops). */
-
-int
-main (int argc, char *argv[])
-{
- int i, j;
- double arr[32768], res = 0, hres = 0;
-
- for (i = 0; i < 32768; i++)
- arr[i] = i;
-
- #pragma acc parallel num_gangs(32) num_workers(32) vector_length(32) \
- copyin(arr) copy(res)
- {
- #pragma acc loop gang reduction(+:res)
- for (j = 0; j < 32; j++)
- {
- #pragma acc loop worker vector reduction(+:res)
- for (i = 0; i < 1024; i++)
- res += arr[j * 1024 + i];
- }
- }
-
- for (i = 0; i < 32768; i++)
- hres += arr[i];
-
- assert (res == hres);
-
- return 0;
-}
deleted file mode 100644
@@ -1,55 +0,0 @@
-#include <assert.h>
-
-/* Test of reduction on loop directive (gangs, workers and vectors, multiple
- non-private reduction variables, float type). */
-
-int
-main (int argc, char *argv[])
-{
- int i, j;
- float arr[32768];
- float res = 0, mres = 0, hres = 0, hmres = 0;
-
- for (i = 0; i < 32768; i++)
- arr[i] = i;
-
- #pragma acc parallel num_gangs(32) num_workers(32) vector_length(32) \
- copy(res, mres)
- {
- #pragma acc loop gang reduction(+:res) reduction(max:mres)
- for (j = 0; j < 32; j++)
- {
- #pragma acc loop worker vector reduction(+:res) reduction(max:mres)
- for (i = 0; i < 1024; i++)
- {
- res += arr[j * 1024 + i];
- if (arr[j * 1024 + i] > mres)
- mres = arr[j * 1024 + i];
- }
-
- #pragma acc loop worker vector reduction(+:res) reduction(max:mres)
- for (i = 0; i < 1024; i++)
- {
- res += arr[j * 1024 + (1023 - i)];
- if (arr[j * 1024 + (1023 - i)] > mres)
- mres = arr[j * 1024 + (1023 - i)];
- }
- }
- }
-
- for (j = 0; j < 32; j++)
- for (i = 0; i < 1024; i++)
- {
- hres += arr[j * 1024 + i];
- hres += arr[j * 1024 + (1023 - i)];
- if (arr[j * 1024 + i] > hmres)
- hmres = arr[j * 1024 + i];
- if (arr[j * 1024 + (1023 - i)] > hmres)
- hmres = arr[j * 1024 + (1023 - i)];
- }
-
- assert (res == hres);
- assert (mres == hmres);
-
- return 0;
-}
deleted file mode 100644
@@ -1,43 +0,0 @@
-/* { dg-additional-options "-w" } */
-
-#include <assert.h>
-
-/* Test of reduction on loop directive (vectors, private reduction
- variable). */
-
-int
-main (int argc, char *argv[])
-{
- int i, j, arr[1024], out[32], res = 0, hres = 0;
-
- for (i = 0; i < 1024; i++)
- arr[i] = i;
-
- #pragma acc parallel num_gangs(32) num_workers(32) vector_length(32) \
- private(res) copyout(out)
- {
- #pragma acc loop gang
- for (j = 0; j < 32; j++)
- {
- res = 0;
-
- #pragma acc loop vector reduction(+:res)
- for (i = 0; i < 32; i++)
- res += arr[j * 32 + i];
-
- out[j] = res;
- }
- }
-
- for (j = 0; j < 32; j++)
- {
- hres = 0;
-
- for (i = 0; i < 32; i++)
- hres += arr[j * 32 + i];
-
- assert (out[j] == hres);
- }
-
- return 0;
-}
deleted file mode 100644
@@ -1,41 +0,0 @@
-#include <assert.h>
-
-/* Test of reduction on loop directive (vector reduction in
- gang-partitioned/worker-partitioned mode, private reduction variable). */
-
-int
-main (int argc, char *argv[])
-{
- int i, j, k;
- double ina[1024], inb[1024], out[1024], acc;
-
- for (j = 0; j < 32; j++)
- for (i = 0; i < 32; i++)
- {
- ina[j * 32 + i] = (i == j) ? 2.0 : 0.0;
- inb[j * 32 + i] = (double) (i + j);
- }
-
- #pragma acc parallel num_gangs(32) num_workers(32) vector_length(32) \
- private(acc) copyin(ina, inb) copyout(out)
- {
- #pragma acc loop gang worker
- for (k = 0; k < 32; k++)
- for (j = 0; j < 32; j++)
- {
- acc = 0;
-
- #pragma acc loop vector reduction(+:acc)
- for (i = 0; i < 32; i++)
- acc += ina[k * 32 + i] * inb[i * 32 + j];
-
- out[k * 32 + j] = acc;
- }
- }
-
- for (j = 0; j < 32; j++)
- for (i = 0; i < 32; i++)
- assert (out[j * 32 + i] == (i + j) * 2);
-
- return 0;
-}
deleted file mode 100644
@@ -1,43 +0,0 @@
-/* { dg-additional-options "-w" } */
-
-#include <assert.h>
-
-/* Test of reduction on loop directive (workers, private reduction
- variable). */
-
-int
-main (int argc, char *argv[])
-{
- int i, j, arr[1024], out[32], res = 0, hres = 0;
-
- for (i = 0; i < 1024; i++)
- arr[i] = i;
-
- #pragma acc parallel num_gangs(32) num_workers(32) vector_length(32) \
- private(res) copyout(out)
- {
- #pragma acc loop gang
- for (j = 0; j < 32; j++)
- {
- res = 0;
-
- #pragma acc loop worker reduction(+:res)
- for (i = 0; i < 32; i++)
- res += arr[j * 32 + i];
-
- out[j] = res;
- }
- }
-
- for (j = 0; j < 32; j++)
- {
- hres = 0;
-
- for (i = 0; i < 32; i++)
- hres += arr[j * 32 + i];
-
- assert (out[j] == hres);
- }
-
- return 0;
-}
deleted file mode 100644
@@ -1,41 +0,0 @@
-#include <assert.h>
-
-/* Test of reduction on loop directive (workers and vectors, private reduction
- variable). */
-
-int
-main (int argc, char *argv[])
-{
- int i, j, arr[1024], out[32], res = 0, hres = 0;
-
- for (i = 0; i < 1024; i++)
- arr[i] = i;
-
- #pragma acc parallel num_gangs(32) num_workers(32) vector_length(32) \
- private(res) copyout(out)
- {
- #pragma acc loop gang
- for (j = 0; j < 32; j++)
- {
- res = 0;
-
- #pragma acc loop worker vector reduction(+:res)
- for (i = 0; i < 32; i++)
- res += arr[j * 32 + i];
-
- out[j] = res;
- }
- }
-
- for (j = 0; j < 32; j++)
- {
- hres = 0;
-
- for (i = 0; i < 32; i++)
- hres += arr[j * 32 + i];
-
- assert (out[j] == hres);
- }
-
- return 0;
-}
deleted file mode 100644
@@ -1,45 +0,0 @@
-#include <assert.h>
-
-/* Test of reduction on loop directive (workers and vectors, private reduction
- variable). */
-
-int
-main (int argc, char *argv[])
-{
- int i, j, arr[32768], out[32], res = 0, hres = 0;
-
- for (i = 0; i < 32768; i++)
- arr[i] = i;
-
- #pragma acc parallel num_gangs(32) num_workers(32) vector_length(32) \
- private(res) copyout(out)
- {
- #pragma acc loop gang
- for (j = 0; j < 32; j++)
- {
- res = j;
-
- #pragma acc loop worker reduction(+:res)
- for (i = 0; i < 1024; i++)
- res += arr[j * 1024 + i];
-
- #pragma acc loop vector reduction(+:res)
- for (i = 1023; i >= 0; i--)
- res += arr[j * 1024 + i];
-
- out[j] = res;
- }
- }
-
- for (j = 0; j < 32; j++)
- {
- hres = j;
-
- for (i = 0; i < 1024; i++)
- hres += arr[j * 1024 + i] * 2;
-
- assert (out[j] == hres);
- }
-
- return 0;
-}
deleted file mode 100644
@@ -1,38 +0,0 @@
-#include <assert.h>
-
-/* Test of reduction on loop directive (workers and vectors, private reduction
- variable: gang-redundant mode). */
-
-int
-main (int argc, char *argv[])
-{
- int i, arr[1024], out[32], res = 0, hres = 0;
-
- for (i = 0; i < 1024; i++)
- arr[i] = i ^ 33;
-
- #pragma acc parallel num_gangs(32) num_workers(32) vector_length(32) \
- private(res) copyin(arr) copyout(out)
- {
- /* Private variables aren't initialized by default in openacc. */
- res = 0;
-
- /* "res" should be available at the end of the following loop (and should
- have the same value redundantly in each gang). */
- #pragma acc loop worker vector reduction(+:res)
- for (i = 0; i < 1024; i++)
- res += arr[i];
-
- #pragma acc loop gang (static: 1)
- for (i = 0; i < 32; i++)
- out[i] = res;
- }
-
- for (i = 0; i < 1024; i++)
- hres += arr[i];
-
- for (i = 0; i < 32; i++)
- assert (out[i] == hres);
-
- return 0;
-}
@@ -1,4 +1,3 @@
-/* { dg-additional-options "-w" } */
/* This code uses nvptx inline assembly guarded with acc_on_device, which is
not optimized away at -O0, and then confuses the target assembler.
{ dg-skip-if "" { *-*-* } { "-O0" } { "" } } */
@@ -17,6 +16,7 @@ int main ()
ary[ix] = -1;
#pragma acc parallel num_workers(32) vector_length(32) copy(ary) copy(ondev)
+ /* { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 18 } */
{
#pragma acc loop worker
for (unsigned ix = 0; ix < N; ix++)
new file mode 100644
@@ -0,0 +1,1186 @@
+/* Miscellaneous test cases for gang/worker/vector mode transitions. */
+
+#include <assert.h>
+#include <stdbool.h>
+#include <stdlib.h>
+#include <math.h>
+#include <openacc.h>
+
+
+/* Test basic vector-partitioned mode transitions. */
+
+void t1()
+{
+ int n = 0, arr[32], i;
+
+ for (i = 0; i < 32; i++)
+ arr[i] = 0;
+
+ #pragma acc parallel copy(n, arr) \
+ num_gangs(1) num_workers(1) vector_length(32)
+ {
+ int j;
+ n++;
+ #pragma acc loop vector
+ for (j = 0; j < 32; j++)
+ arr[j]++;
+ n++;
+ }
+
+ assert (n == 2);
+
+ for (i = 0; i < 32; i++)
+ assert (arr[i] == 1);
+}
+
+
+/* Test vector-partitioned, gang-partitioned mode. */
+
+void t2()
+{
+ int n[32], arr[1024], i;
+
+ for (i = 0; i < 1024; i++)
+ arr[i] = 0;
+
+ for (i = 0; i < 32; i++)
+ n[i] = 0;
+
+ #pragma acc parallel copy(n, arr) \
+ num_gangs(32) num_workers(1) vector_length(32)
+ {
+ int j, k;
+
+ #pragma acc loop gang(static:*)
+ for (j = 0; j < 32; j++)
+ n[j]++;
+
+ #pragma acc loop gang
+ for (j = 0; j < 32; j++)
+ #pragma acc loop vector
+ for (k = 0; k < 32; k++)
+ arr[j * 32 + k]++;
+
+ #pragma acc loop gang(static:*)
+ for (j = 0; j < 32; j++)
+ n[j]++;
+ }
+
+ for (i = 0; i < 32; i++)
+ assert (n[i] == 2);
+
+ for (i = 0; i < 1024; i++)
+ assert (arr[i] == 1);
+}
+
+
+/* Test conditional vector-partitioned loops. */
+
+void t3()
+{
+ int n[32], arr[1024], i;
+
+ for (i = 0; i < 1024; i++)
+ arr[i] = 0;
+
+ for (i = 0; i < 32; i++)
+ n[i] = 0;
+
+ #pragma acc parallel copy(n, arr) \
+ num_gangs(32) num_workers(1) vector_length(32)
+ {
+ int j, k;
+
+ #pragma acc loop gang(static:*)
+ for (j = 0; j < 32; j++)
+ n[j]++;
+
+ #pragma acc loop gang
+ for (j = 0; j < 32; j++)
+ {
+ if ((j % 2) == 0)
+ {
+ #pragma acc loop vector
+ for (k = 0; k < 32; k++)
+ arr[j * 32 + k]++;
+ }
+ else
+ {
+ #pragma acc loop vector
+ for (k = 0; k < 32; k++)
+ arr[j * 32 + k]--;
+ }
+ }
+
+ #pragma acc loop gang(static:*)
+ for (j = 0; j < 32; j++)
+ n[j]++;
+ }
+
+ for (i = 0; i < 32; i++)
+ assert (n[i] == 2);
+
+ for (i = 0; i < 1024; i++)
+ assert (arr[i] == ((i % 64) < 32) ? 1 : -1);
+}
+
+
+/* Test conditions inside vector-partitioned loops. */
+
+void t4()
+{
+ int n[32], arr[1024], i;
+
+ for (i = 0; i < 1024; i++)
+ arr[i] = i;
+
+ for (i = 0; i < 32; i++)
+ n[i] = 0;
+
+ #pragma acc parallel copy(n, arr) \
+ num_gangs(32) num_workers(1) vector_length(32)
+ {
+ int j, k;
+
+ #pragma acc loop gang(static:*)
+ for (j = 0; j < 32; j++)
+ n[j]++;
+
+ #pragma acc loop gang
+ for (j = 0; j < 32; j++)
+ {
+ #pragma acc loop vector
+ for (k = 0; k < 32; k++)
+ if ((arr[j * 32 + k] % 2) != 0)
+ arr[j * 32 + k] *= 2;
+ }
+
+ #pragma acc loop gang(static:*)
+ for (j = 0; j < 32; j++)
+ n[j]++;
+ }
+
+ for (i = 0; i < 32; i++)
+ assert (n[i] == 2);
+
+ for (i = 0; i < 1024; i++)
+ assert (arr[i] == ((i % 2) == 0 ? i : i * 2));
+}
+
+
+/* Test conditions inside gang-partitioned/vector-partitioned loops. */
+
+void t5()
+{
+ int n[32], arr[1024], i;
+
+ for (i = 0; i < 1024; i++)
+ arr[i] = i;
+
+ for (i = 0; i < 32; i++)
+ n[i] = 0;
+
+ #pragma acc parallel copy(n, arr) \
+ num_gangs(32) num_workers(1) vector_length(32)
+ {
+ int j;
+
+ #pragma acc loop gang(static:*)
+ for (j = 0; j < 32; j++)
+ n[j]++;
+
+ #pragma acc loop gang vector
+ for (j = 0; j < 1024; j++)
+ if ((arr[j] % 2) != 0)
+ arr[j] *= 2;
+
+ #pragma acc loop gang(static:*)
+ for (j = 0; j < 32; j++)
+ n[j]++;
+ }
+
+ for (i = 0; i < 32; i++)
+ assert (n[i] == 2);
+
+ for (i = 0; i < 1024; i++)
+ assert (arr[i] == ((i % 2) == 0 ? i : i * 2));
+}
+
+
+/* Test switch containing vector-partitioned loops inside gang-partitioned
+ loops. */
+
+void t6()
+{
+ int n[32], arr[1024], i;
+
+ for (i = 0; i < 1024; i++)
+ arr[i] = 0;
+
+ for (i = 0; i < 32; i++)
+ n[i] = i % 5;
+
+ #pragma acc parallel copy(n, arr) \
+ num_gangs(32) num_workers(1) vector_length(32)
+ {
+ int j, k;
+
+ #pragma acc loop gang(static:*)
+ for (j = 0; j < 32; j++)
+ n[j]++;
+
+ #pragma acc loop gang(static:*)
+ for (j = 0; j < 32; j++)
+ switch (n[j])
+ {
+ case 1:
+ #pragma acc loop vector
+ for (k = 0; k < 32; k++)
+ arr[j * 32 + k] += 1;
+ break;
+
+ case 2:
+ #pragma acc loop vector
+ for (k = 0; k < 32; k++)
+ arr[j * 32 + k] += 2;
+ break;
+
+ case 3:
+ #pragma acc loop vector
+ for (k = 0; k < 32; k++)
+ arr[j * 32 + k] += 3;
+ break;
+
+ case 4:
+ #pragma acc loop vector
+ for (k = 0; k < 32; k++)
+ arr[j * 32 + k] += 4;
+ break;
+
+ case 5:
+ #pragma acc loop vector
+ for (k = 0; k < 32; k++)
+ arr[j * 32 + k] += 5;
+ break;
+
+ default:
+ abort ();
+ }
+
+ #pragma acc loop gang(static:*)
+ for (j = 0; j < 32; j++)
+ n[j]++;
+ }
+
+ for (i = 0; i < 32; i++)
+ assert (n[i] == (i % 5) + 2);
+
+ for (i = 0; i < 1024; i++)
+ assert (arr[i] == ((i / 32) % 5) + 1);
+}
+
+
+/* Test trivial operation of vector-single mode. */
+
+void t7()
+{
+ int n = 0;
+ #pragma acc parallel copy(n) \
+ num_gangs(1) num_workers(1) vector_length(32)
+ /* { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 288 } */
+ {
+ n++;
+ }
+ assert (n == 1);
+}
+
+
+/* Test vector-single, gang-partitioned mode. */
+
+void t8()
+{
+ int arr[1024];
+ int gangs;
+
+ for (gangs = 1; gangs <= 1024; gangs <<= 1)
+ {
+ int i;
+
+ for (i = 0; i < 1024; i++)
+ arr[i] = 0;
+
+ #pragma acc parallel copy(arr) \
+ num_gangs(gangs) num_workers(1) vector_length(32)
+ /* { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 312 } */
+ {
+ int j;
+ #pragma acc loop gang
+ for (j = 0; j < 1024; j++)
+ arr[j]++;
+ }
+
+ for (i = 0; i < 1024; i++)
+ assert (arr[i] == 1);
+ }
+}
+
+
+/* Test conditions in vector-single mode. */
+
+void t9()
+{
+ int arr[1024];
+ int gangs;
+
+ for (gangs = 1; gangs <= 1024; gangs <<= 1)
+ {
+ int i;
+
+ for (i = 0; i < 1024; i++)
+ arr[i] = 0;
+
+ #pragma acc parallel copy(arr) \
+ num_gangs(gangs) num_workers(1) vector_length(32)
+ /* { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 342 } */
+ {
+ int j;
+ #pragma acc loop gang
+ for (j = 0; j < 1024; j++)
+ if ((j % 3) == 0)
+ arr[j]++;
+ else
+ arr[j] += 2;
+ }
+
+ for (i = 0; i < 1024; i++)
+ assert (arr[i] == ((i % 3) == 0) ? 1 : 2);
+ }
+}
+
+
+/* Test switch in vector-single mode. */
+
+void t10()
+{
+ int arr[1024];
+ int gangs;
+
+ for (gangs = 1; gangs <= 1024; gangs <<= 1)
+ {
+ int i;
+
+ for (i = 0; i < 1024; i++)
+ arr[i] = 0;
+
+ #pragma acc parallel copy(arr) \
+ num_gangs(gangs) num_workers(1) vector_length(32)
+ /* { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 375 } */
+ {
+ int j;
+ #pragma acc loop gang
+ for (j = 0; j < 1024; j++)
+ switch (j % 5)
+ {
+ case 0: arr[j] += 1; break;
+ case 1: arr[j] += 2; break;
+ case 2: arr[j] += 3; break;
+ case 3: arr[j] += 4; break;
+ case 4: arr[j] += 5; break;
+ default: arr[j] += 99;
+ }
+ }
+
+ for (i = 0; i < 1024; i++)
+ assert (arr[i] == (i % 5) + 1);
+ }
+}
+
+
+/* Test switch in vector-single mode, initialise array on device. */
+
+void t11()
+{
+ int arr[1024];
+ int i;
+
+ for (i = 0; i < 1024; i++)
+ arr[i] = 99;
+
+ #pragma acc parallel copy(arr) \
+ num_gangs(1024) num_workers(1) vector_length(32)
+ /* { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 409 } */
+ {
+ int j;
+
+ /* This loop and the one following must be distributed to available gangs
+ in the same way to ensure data dependencies are not violated (hence the
+ "static" clauses). */
+ #pragma acc loop gang(static:*)
+ for (j = 0; j < 1024; j++)
+ arr[j] = 0;
+
+ #pragma acc loop gang(static:*)
+ for (j = 0; j < 1024; j++)
+ switch (j % 5)
+ {
+ case 0: arr[j] += 1; break;
+ case 1: arr[j] += 2; break;
+ case 2: arr[j] += 3; break;
+ case 3: arr[j] += 4; break;
+ case 4: arr[j] += 5; break;
+ default: arr[j] += 99;
+ }
+ }
+
+ for (i = 0; i < 1024; i++)
+ assert (arr[i] == (i % 5) + 1);
+}
+
+
+/* Test multiple conditions in vector-single mode. */
+
+#define NUM_GANGS 4096
+void t12()
+{
+ bool fizz[NUM_GANGS], buzz[NUM_GANGS], fizzbuzz[NUM_GANGS];
+ int i;
+
+ #pragma acc parallel copyout(fizz, buzz, fizzbuzz) \
+ num_gangs(NUM_GANGS) num_workers(1) vector_length(32)
+ /* { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 448 } */
+ {
+ int j;
+
+ /* This loop and the one following must be distributed to available gangs
+ in the same way to ensure data dependencies are not violated (hence the
+ "static" clauses). */
+ #pragma acc loop gang(static:*)
+ for (j = 0; j < NUM_GANGS; j++)
+ fizz[j] = buzz[j] = fizzbuzz[j] = 0;
+
+ #pragma acc loop gang(static:*)
+ for (j = 0; j < NUM_GANGS; j++)
+ {
+ if ((j % 3) == 0 && (j % 5) == 0)
+ fizzbuzz[j] = 1;
+ else
+ {
+ if ((j % 3) == 0)
+ fizz[j] = 1;
+ else if ((j % 5) == 0)
+ buzz[j] = 1;
+ }
+ }
+ }
+
+ for (i = 0; i < NUM_GANGS; i++)
+ {
+ assert (fizzbuzz[i] == ((i % 3) == 0 && (i % 5) == 0));
+ assert (fizz[i] == ((i % 3) == 0 && (i % 5) != 0));
+ assert (buzz[i] == ((i % 3) != 0 && (i % 5) == 0));
+ }
+}
+#undef NUM_GANGS
+
+
+/* Test worker-partitioned/vector-single mode. */
+
+void t13()
+{
+ int arr[32 * 8], i;
+
+ for (i = 0; i < 32 * 8; i++)
+ arr[i] = 0;
+
+ #pragma acc parallel copy(arr) \
+ num_gangs(8) num_workers(8) vector_length(32)
+ /* { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 495 } */
+ {
+ int j;
+ #pragma acc loop gang
+ for (j = 0; j < 32; j++)
+ {
+ int k;
+ #pragma acc loop worker
+ for (k = 0; k < 8; k++)
+ arr[j * 8 + k] += j * 8 + k;
+ }
+ }
+
+ for (i = 0; i < 32 * 8; i++)
+ assert (arr[i] == i);
+}
+
+
+/* Test condition in worker-partitioned mode. */
+
+void t14()
+{
+ int arr[32 * 32 * 8], i;
+
+ for (i = 0; i < 32 * 32 * 8; i++)
+ arr[i] = i;
+
+ #pragma acc parallel copy(arr) \
+ num_gangs(8) num_workers(8) vector_length(32)
+ {
+ int j;
+ #pragma acc loop gang
+ for (j = 0; j < 32; j++)
+ {
+ int k;
+ #pragma acc loop worker
+ for (k = 0; k < 8; k++)
+ {
+ int m;
+ if ((k % 2) == 0)
+ {
+ #pragma acc loop vector
+ for (m = 0; m < 32; m++)
+ arr[j * 32 * 8 + k * 32 + m]++;
+ }
+ else
+ {
+ #pragma acc loop vector
+ for (m = 0; m < 32; m++)
+ arr[j * 32 * 8 + k * 32 + m] += 2;
+ }
+ }
+ }
+ }
+
+ for (i = 0; i < 32 * 32 * 8; i++)
+ assert (arr[i] == i + ((i / 32) % 2) + 1);
+}
+
+
+/* Test switch in worker-partitioned mode. */
+
+void t15()
+{
+ int arr[32 * 32 * 8], i;
+
+ for (i = 0; i < 32 * 32 * 8; i++)
+ arr[i] = i;
+
+ #pragma acc parallel copy(arr) \
+ num_gangs(8) num_workers(8) vector_length(32)
+ {
+ int j;
+ #pragma acc loop gang
+ for (j = 0; j < 32; j++)
+ {
+ int k;
+ #pragma acc loop worker
+ for (k = 0; k < 8; k++)
+ {
+ int m;
+ switch ((j * 32 + k) % 3)
+ {
+ case 0:
+ #pragma acc loop vector
+ for (m = 0; m < 32; m++)
+ arr[j * 32 * 8 + k * 32 + m]++;
+ break;
+
+ case 1:
+ #pragma acc loop vector
+ for (m = 0; m < 32; m++)
+ arr[j * 32 * 8 + k * 32 + m] += 2;
+ break;
+
+ case 2:
+ #pragma acc loop vector
+ for (m = 0; m < 32; m++)
+ arr[j * 32 * 8 + k * 32 + m] += 3;
+ break;
+
+ default: ;
+ }
+ }
+ }
+ }
+
+ for (i = 0; i < 32 * 32 * 8; i++)
+ assert (arr[i] == i + ((i / 32) % 3) + 1);
+}
+
+
+/* Test worker-single/worker-partitioned transitions. */
+
+void t16()
+{
+ int n[32], arr[32 * 32], i;
+
+ for (i = 0; i < 32 * 32; i++)
+ arr[i] = 0;
+
+ for (i = 0; i < 32; i++)
+ n[i] = 0;
+
+ #pragma acc parallel copy(n, arr) \
+ num_gangs(8) num_workers(16) vector_length(32)
+ /* { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 621 } */
+ {
+ int j;
+ #pragma acc loop gang
+ for (j = 0; j < 32; j++)
+ {
+ int k;
+
+ n[j]++;
+
+ #pragma acc loop worker
+ for (k = 0; k < 32; k++)
+ arr[j * 32 + k]++;
+
+ n[j]++;
+
+ #pragma acc loop worker
+ for (k = 0; k < 32; k++)
+ arr[j * 32 + k]++;
+
+ n[j]++;
+
+ #pragma acc loop worker
+ for (k = 0; k < 32; k++)
+ arr[j * 32 + k]++;
+
+ n[j]++;
+ }
+ }
+
+ for (i = 0; i < 32; i++)
+ assert (n[i] == 4);
+
+ for (i = 0; i < 32 * 32; i++)
+ assert (arr[i] == 3);
+}
+
+
+/* Test correct synchronisation between worker-partitioned loops. */
+
+void t17()
+{
+ int arr_a[32 * 32], arr_b[32 * 32], i;
+ int num_workers, num_gangs;
+
+ for (num_workers = 1; num_workers <= 32; num_workers <<= 1)
+ for (num_gangs = 1; num_gangs <= 32; num_gangs <<= 1)
+ {
+ for (i = 0; i < 32 * 32; i++)
+ arr_a[i] = i;
+
+ #pragma acc parallel copyin(arr_a) copyout(arr_b) \
+ num_gangs(num_gangs) num_workers(num_workers) vector_length(32)
+ /* { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 674 } */
+ {
+ int j;
+ #pragma acc loop gang
+ for (j = 0; j < 32; j++)
+ {
+ int k;
+
+ #pragma acc loop worker
+ for (k = 0; k < 32; k++)
+ arr_b[j * 32 + (31 - k)] = arr_a[j * 32 + k] * 2;
+
+ #pragma acc loop worker
+ for (k = 0; k < 32; k++)
+ arr_a[j * 32 + (31 - k)] = arr_b[j * 32 + k] * 2;
+
+ #pragma acc loop worker
+ for (k = 0; k < 32; k++)
+ arr_b[j * 32 + (31 - k)] = arr_a[j * 32 + k] * 2;
+ }
+ }
+
+ for (i = 0; i < 32 * 32; i++)
+ assert (arr_b[i] == (i ^ 31) * 8);
+ }
+}
+
+
+/* Test correct synchronisation between worker+vector-partitioned loops. */
+
+void t18()
+{
+ int arr_a[32 * 32 * 32], arr_b[32 * 32 * 32], i;
+ int num_workers, num_gangs;
+
+ for (num_workers = 1; num_workers <= 32; num_workers <<= 1)
+ for (num_gangs = 1; num_gangs <= 32; num_gangs <<= 1)
+ {
+ for (i = 0; i < 32 * 32 * 32; i++)
+ arr_a[i] = i;
+
+ #pragma acc parallel copyin(arr_a) copyout(arr_b) \
+ num_gangs(num_gangs) num_workers(num_workers) vector_length(32)
+ {
+ int j;
+ #pragma acc loop gang
+ for (j = 0; j < 32; j++)
+ {
+ int k;
+
+ #pragma acc loop worker vector
+ for (k = 0; k < 32 * 32; k++)
+ arr_b[j * 32 * 32 + (1023 - k)] = arr_a[j * 32 * 32 + k] * 2;
+
+ #pragma acc loop worker vector
+ for (k = 0; k < 32 * 32; k++)
+ arr_a[j * 32 * 32 + (1023 - k)] = arr_b[j * 32 * 32 + k] * 2;
+
+ #pragma acc loop worker vector
+ for (k = 0; k < 32 * 32; k++)
+ arr_b[j * 32 * 32 + (1023 - k)] = arr_a[j * 32 * 32 + k] * 2;
+ }
+ }
+
+ for (i = 0; i < 32 * 32 * 32; i++)
+ assert (arr_b[i] == (i ^ 1023) * 8);
+ }
+}
+
+
+/* Test correct synchronisation between vector-partitioned loops in
+ worker-partitioned mode. */
+
+void t19()
+{
+ int n[32 * 32], arr_a[32 * 32 * 32], arr_b[32 * 32 * 32], i;
+ int num_workers, num_gangs;
+
+ for (num_workers = 1; num_workers <= 32; num_workers <<= 1)
+ for (num_gangs = 1; num_gangs <= 32; num_gangs <<= 1)
+ {
+ for (i = 0; i < 32 * 32 * 32; i++)
+ arr_a[i] = i;
+
+ for (i = 0; i < 32 * 32; i++)
+ n[i] = 0;
+
+ #pragma acc parallel copy (n) copyin(arr_a) copyout(arr_b) \
+ num_gangs(num_gangs) num_workers(num_workers) vector_length(32)
+ {
+ int j;
+ #pragma acc loop gang
+ for (j = 0; j < 32; j++)
+ {
+ int k;
+
+ #pragma acc loop worker
+ for (k = 0; k < 32; k++)
+ {
+ int m;
+
+ n[j * 32 + k]++;
+
+ #pragma acc loop vector
+ for (m = 0; m < 32; m++)
+ {
+ if (((j * 1024 + k * 32 + m) % 2) == 0)
+ arr_b[j * 1024 + k * 32 + (31 - m)]
+ = arr_a[j * 1024 + k * 32 + m] * 2;
+ else
+ arr_b[j * 1024 + k * 32 + (31 - m)]
+ = arr_a[j * 1024 + k * 32 + m] * 3;
+ }
+
+ /* Test returning to vector-single mode... */
+ n[j * 32 + k]++;
+
+ #pragma acc loop vector
+ for (m = 0; m < 32; m++)
+ {
+ if (((j * 1024 + k * 32 + m) % 3) == 0)
+ arr_a[j * 1024 + k * 32 + (31 - m)]
+ = arr_b[j * 1024 + k * 32 + m] * 5;
+ else
+ arr_a[j * 1024 + k * 32 + (31 - m)]
+ = arr_b[j * 1024 + k * 32 + m] * 7;
+ }
+
+ /* ...and back-to-back vector loops. */
+
+ #pragma acc loop vector
+ for (m = 0; m < 32; m++)
+ {
+ if (((j * 1024 + k * 32 + m) % 2) == 0)
+ arr_b[j * 1024 + k * 32 + (31 - m)]
+ = arr_a[j * 1024 + k * 32 + m] * 3;
+ else
+ arr_b[j * 1024 + k * 32 + (31 - m)]
+ = arr_a[j * 1024 + k * 32 + m] * 2;
+ }
+ }
+ }
+ }
+
+ for (i = 0; i < 32 * 32; i++)
+ assert (n[i] == 2);
+
+ for (i = 0; i < 32 * 32 * 32; i++)
+ {
+ int m = 6 * ((i % 3) == 0 ? 5 : 7);
+ assert (arr_b[i] == (i ^ 31) * m);
+ }
+ }
+}
+
+
+/* With -O0, variables are on the stack, not in registers. Check that worker
+ state propagation handles the stack frame. */
+
+void t20()
+{
+ int w0 = 0;
+ int w1 = 0;
+ int w2 = 0;
+ int w3 = 0;
+ int w4 = 0;
+ int w5 = 0;
+ int w6 = 0;
+ int w7 = 0;
+
+ int i;
+
+#pragma acc parallel copy (w0, w1, w2, w3, w4, w5, w6, w7) \
+ num_gangs (1) num_workers (8)
+ {
+ int internal = 100;
+
+#pragma acc loop worker
+ for (i = 0; i < 8; i++)
+ {
+ switch (i)
+ {
+ case 0: w0 = internal; break;
+ case 1: w1 = internal; break;
+ case 2: w2 = internal; break;
+ case 3: w3 = internal; break;
+ case 4: w4 = internal; break;
+ case 5: w5 = internal; break;
+ case 6: w6 = internal; break;
+ case 7: w7 = internal; break;
+ default: break;
+ }
+ }
+ }
+
+ if (w0 != 100
+ || w1 != 100
+ || w2 != 100
+ || w3 != 100
+ || w4 != 100
+ || w5 != 100
+ || w6 != 100
+ || w7 != 100)
+ __builtin_abort ();
+}
+
+
+/* Test worker-single/vector-single mode. */
+
+void t21()
+{
+ int arr[32], i;
+
+ for (i = 0; i < 32; i++)
+ arr[i] = 0;
+
+ #pragma acc parallel copy(arr) \
+ num_gangs(8) num_workers(8) vector_length(32)
+ /* { dg-warning "region is worker partitioned but does not contain worker partitioned code" "worker" { target *-*-* } 892 } */
+ /* { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 892 } */
+ {
+ int j;
+ #pragma acc loop gang
+ for (j = 0; j < 32; j++)
+ arr[j]++;
+ }
+
+ for (i = 0; i < 32; i++)
+ assert (arr[i] == 1);
+}
+
+
+/* Test worker-single/vector-single mode. */
+
+void t22()
+{
+ int arr[32], i;
+
+ for (i = 0; i < 32; i++)
+ arr[i] = 0;
+
+ #pragma acc parallel copy(arr) \
+ num_gangs(8) num_workers(8) vector_length(32)
+ /* { dg-warning "region is worker partitioned but does not contain worker partitioned code" "worker" { target *-*-* } 917 } */
+ /* { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 917 } */
+ {
+ int j;
+ #pragma acc loop gang
+ for (j = 0; j < 32; j++)
+ {
+ #pragma acc atomic
+ arr[j]++;
+ }
+ }
+
+ for (i = 0; i < 32; i++)
+ assert (arr[i] == 1);
+}
+
+
+/* Test condition in worker-single/vector-single mode. */
+
+void t23()
+{
+ int arr[32], i;
+
+ for (i = 0; i < 32; i++)
+ arr[i] = i;
+
+ #pragma acc parallel copy(arr) \
+ num_gangs(8) num_workers(8) vector_length(32)
+ /* { dg-warning "region is worker partitioned but does not contain worker partitioned code" "worker" { target *-*-* } 945 } */
+ /* { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 945 } */
+ {
+ int j;
+ #pragma acc loop gang
+ for (j = 0; j < 32; j++)
+ if ((arr[j] % 2) != 0)
+ arr[j]++;
+ else
+ arr[j] += 2;
+ }
+
+ for (i = 0; i < 32; i++)
+ assert (arr[i] == ((i % 2) != 0) ? i + 1 : i + 2);
+}
+
+
+/* Test switch in worker-single/vector-single mode. */
+
+void t24()
+{
+ int arr[32], i;
+
+ for (i = 0; i < 32; i++)
+ arr[i] = i;
+
+ #pragma acc parallel copy(arr) \
+ num_gangs(8) num_workers(8) vector_length(32)
+ /* { dg-warning "region is worker partitioned but does not contain worker partitioned code" "worker" { target *-*-* } 973 } */
+ /* { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 973 } */
+ {
+ int j;
+ #pragma acc loop gang
+ for (j = 0; j < 32; j++)
+ switch (arr[j] % 5)
+ {
+ case 0: arr[j] += 1; break;
+ case 1: arr[j] += 2; break;
+ case 2: arr[j] += 3; break;
+ case 3: arr[j] += 4; break;
+ case 4: arr[j] += 5; break;
+ default: arr[j] += 99;
+ }
+ }
+
+ for (i = 0; i < 32; i++)
+ assert (arr[i] == i + (i % 5) + 1);
+}
+
+
+/* Test worker-single/vector-partitioned mode. */
+
+void t25()
+{
+ int arr[32 * 32], i;
+
+ for (i = 0; i < 32 * 32; i++)
+ arr[i] = i;
+
+ #pragma acc parallel copy(arr) \
+ num_gangs(8) num_workers(8) vector_length(32)
+ /* { dg-warning "region is worker partitioned but does not contain worker partitioned code" "worker" { target *-*-* } 1006 } */
+ {
+ int j;
+ #pragma acc loop gang
+ for (j = 0; j < 32; j++)
+ {
+ int k;
+ #pragma acc loop vector
+ for (k = 0; k < 32; k++)
+ {
+ #pragma acc atomic
+ arr[j * 32 + k]++;
+ }
+ }
+ }
+
+ for (i = 0; i < 32 * 32; i++)
+ assert (arr[i] == i + 1);
+}
+
+
+/* Test multiple conditional vector-partitioned loops in worker-single
+ mode. */
+
+void t26()
+{
+ int arr[32 * 32], i;
+
+ for (i = 0; i < 32 * 32; i++)
+ arr[i] = i;
+
+ #pragma acc parallel copy(arr) \
+ num_gangs(8) num_workers(8) vector_length(32)
+ /* { dg-warning "region is worker partitioned but does not contain worker partitioned code" "worker" { target *-*-* } 1039 } */
+ {
+ int j;
+ #pragma acc loop gang
+ for (j = 0; j < 32; j++)
+ {
+ int k;
+ if ((j % 3) == 0)
+ {
+ #pragma acc loop vector
+ for (k = 0; k < 32; k++)
+ {
+ #pragma acc atomic
+ arr[j * 32 + k] += 3;
+ }
+ }
+ else if ((j % 3) == 1)
+ {
+ #pragma acc loop vector
+ for (k = 0; k < 32; k++)
+ {
+ #pragma acc atomic
+ arr[j * 32 + k] += 7;
+ }
+ }
+ }
+ }
+
+ for (i = 0; i < 32 * 32; i++)
+ {
+ int j = (i / 32) % 3;
+ assert (arr[i] == i + ((j == 0) ? 3 : (j == 1) ? 7 : 0));
+ }
+}
+
+
+/* Test worker-single, vector-partitioned, gang-redundant mode. */
+
+#define ACTUAL_GANGS 8
+void t27()
+{
+ int n, arr[32], i;
+ int ondev;
+
+ for (i = 0; i < 32; i++)
+ arr[i] = 0;
+
+ n = 0;
+
+ #pragma acc parallel copy(n, arr) copyout(ondev) \
+ num_gangs(ACTUAL_GANGS) num_workers(8) vector_length(32)
+ /* { dg-warning "region is gang partitioned but does not contain gang partitioned code" "gang" { target *-*-* } 1090 } */
+ /* { dg-warning "region is worker partitioned but does not contain worker partitioned code" "worker" { target *-*-* } 1090 } */
+ {
+ int j;
+
+ ondev = acc_on_device (acc_device_not_host);
+
+ #pragma acc atomic
+ n++;
+
+ #pragma acc loop vector
+ for (j = 0; j < 32; j++)
+ {
+ #pragma acc atomic
+ arr[j] += 1;
+ }
+
+ #pragma acc atomic
+ n++;
+ }
+
+ int m = ondev ? ACTUAL_GANGS : 1;
+
+ assert (n == m * 2);
+
+ for (i = 0; i < 32; i++)
+ assert (arr[i] == m);
+}
+#undef ACTUAL_GANGS
+
+
+/* Check if worker-single variables get broadcastd to vectors. */
+
+#pragma acc routine
+float t28_routine ()
+{
+ return 2.71;
+}
+
+#define N 32
+void t28()
+{
+ float threads[N], v1 = 3.14;
+
+ for (int i = 0; i < N; i++)
+ threads[i] = -1;
+
+#pragma acc parallel num_gangs (1) vector_length (32) copy (v1)
+ {
+ float val = t28_routine ();
+
+#pragma acc loop vector
+ for (int i = 0; i < N; i++)
+ threads[i] = val + v1*i;
+ }
+
+ for (int i = 0; i < N; i++)
+ assert (fabs (threads[i] - (t28_routine () + v1*i)) < 0.0001);
+}
+#undef N
+
+
+int main()
+{
+ t1();
+ t2();
+ t3();
+ t4();
+ t5();
+ t6();
+ t7();
+ t8();
+ t9();
+ t10();
+ t11();
+ t12();
+ t13();
+ t14();
+ t15();
+ t16();
+ t17();
+ t18();
+ t19();
+ t20();
+ t21();
+ t22();
+ t23();
+ t24();
+ t25();
+ t26();
+ t27();
+ t28();
+
+ return 0;
+}
@@ -1,3 +1,5 @@
+/* { dg-do run } */
+
#include <stdlib.h>
int
deleted file mode 100644
@@ -1,6 +0,0 @@
-/* { dg-do run } */
-
-#include <stdlib.h>
-
-#define EXEC_DIRECTIVE parallel
-#include "data-clauses.h"
new file mode 100644
@@ -0,0 +1,966 @@
+#include <assert.h>
+#include <openacc.h>
+
+typedef struct {
+ int x, y;
+} vec2;
+
+typedef struct {
+ int x, y, z;
+ int attr[13];
+} vec3_attr;
+
+
+/* Test of gang-private variables declared in local scope with parallel
+ directive. */
+
+void local_g_1()
+{
+ int i, arr[32];
+
+ for (i = 0; i < 32; i++)
+ arr[i] = 3;
+
+ #pragma acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
+ /* { dg-warning "region is worker partitioned but does not contain worker partitioned code" "worker" { target *-*-* } 24 } */
+ /* { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 24 } */
+ {
+ int x;
+
+ #pragma acc loop gang(static:1)
+ for (i = 0; i < 32; i++)
+ x = i * 2;
+
+ #pragma acc loop gang(static:1)
+ for (i = 0; i < 32; i++)
+ {
+ if (acc_on_device (acc_device_host))
+ x = i * 2;
+ arr[i] += x;
+ }
+ }
+
+ for (i = 0; i < 32; i++)
+ assert (arr[i] == 3 + i * 2);
+}
+
+
+/* Test of worker-private variables declared in a local scope, broadcasting
+ to vector-partitioned mode. Back-to-back worker loops. */
+
+void local_w_1()
+{
+ int i, arr[32 * 32 * 32];
+
+ for (i = 0; i < 32 * 32 * 32; i++)
+ arr[i] = i;
+
+ #pragma acc parallel copy(arr) num_gangs(32) num_workers(32) vector_length(32)
+ {
+ int j;
+
+ #pragma acc loop gang
+ for (i = 0; i < 32; i++)
+ {
+ #pragma acc loop worker
+ for (j = 0; j < 32; j++)
+ {
+ int k;
+ int x = i ^ j * 3;
+
+ #pragma acc loop vector
+ for (k = 0; k < 32; k++)
+ arr[i * 1024 + j * 32 + k] += x * k;
+ }
+
+ #pragma acc loop worker
+ for (j = 0; j < 32; j++)
+ {
+ int k;
+ int x = i | j * 5;
+
+ #pragma acc loop vector
+ for (k = 0; k < 32; k++)
+ arr[i * 1024 + j * 32 + k] += x * k;
+ }
+ }
+ }
+
+ for (i = 0; i < 32; i++)
+ for (int j = 0; j < 32; j++)
+ for (int k = 0; k < 32; k++)
+ {
+ int idx = i * 1024 + j * 32 + k;
+ assert (arr[idx] == idx + (i ^ j * 3) * k + (i | j * 5) * k);
+ }
+}
+
+
+/* Test of worker-private variables declared in a local scope, broadcasting
+ to vector-partitioned mode. Successive vector loops. */
+
+void local_w_2()
+{
+ int i, arr[32 * 32 * 32];
+
+ for (i = 0; i < 32 * 32 * 32; i++)
+ arr[i] = i;
+
+ #pragma acc parallel copy(arr) num_gangs(32) num_workers(32) vector_length(32)
+ {
+ int j;
+
+ #pragma acc loop gang
+ for (i = 0; i < 32; i++)
+ {
+ #pragma acc loop worker
+ for (j = 0; j < 32; j++)
+ {
+ int k;
+ int x = i ^ j * 3;
+
+ #pragma acc loop vector
+ for (k = 0; k < 32; k++)
+ arr[i * 1024 + j * 32 + k] += x * k;
+
+ x = i | j * 5;
+
+ #pragma acc loop vector
+ for (k = 0; k < 32; k++)
+ arr[i * 1024 + j * 32 + k] += x * k;
+ }
+ }
+ }
+
+ for (i = 0; i < 32; i++)
+ for (int j = 0; j < 32; j++)
+ for (int k = 0; k < 32; k++)
+ {
+ int idx = i * 1024 + j * 32 + k;
+ assert (arr[idx] == idx + (i ^ j * 3) * k + (i | j * 5) * k);
+ }
+}
+
+
+/* Test of worker-private variables declared in a local scope, broadcasting
+ to vector-partitioned mode. Aggregate worker variable. */
+
+void local_w_3()
+{
+ int i, arr[32 * 32 * 32];
+
+ for (i = 0; i < 32 * 32 * 32; i++)
+ arr[i] = i;
+
+ #pragma acc parallel copy(arr) num_gangs(32) num_workers(32) vector_length(32)
+ {
+ int j;
+
+ #pragma acc loop gang
+ for (i = 0; i < 32; i++)
+ {
+ #pragma acc loop worker
+ for (j = 0; j < 32; j++)
+ {
+ int k;
+ vec2 pt;
+
+ pt.x = i ^ j * 3;
+ pt.y = i | j * 5;
+
+ #pragma acc loop vector
+ for (k = 0; k < 32; k++)
+ arr[i * 1024 + j * 32 + k] += pt.x * k;
+
+ #pragma acc loop vector
+ for (k = 0; k < 32; k++)
+ arr[i * 1024 + j * 32 + k] += pt.y * k;
+ }
+ }
+ }
+
+ for (i = 0; i < 32; i++)
+ for (int j = 0; j < 32; j++)
+ for (int k = 0; k < 32; k++)
+ {
+ int idx = i * 1024 + j * 32 + k;
+ assert (arr[idx] == idx + (i ^ j * 3) * k + (i | j * 5) * k);
+ }
+}
+
+
+/* Test of worker-private variables declared in a local scope, broadcasting
+ to vector-partitioned mode. Addressable worker variable. */
+
+void local_w_4()
+{
+ int i, arr[32 * 32 * 32];
+
+ for (i = 0; i < 32 * 32 * 32; i++)
+ arr[i] = i;
+
+ #pragma acc parallel copy(arr) num_gangs(32) num_workers(32) vector_length(32)
+ {
+ int j;
+
+ #pragma acc loop gang
+ for (i = 0; i < 32; i++)
+ {
+ #pragma acc loop worker
+ for (j = 0; j < 32; j++)
+ {
+ int k;
+ vec2 pt, *ptp;
+
+ ptp = &pt;
+
+ pt.x = i ^ j * 3;
+
+ #pragma acc loop vector
+ for (k = 0; k < 32; k++)
+ arr[i * 1024 + j * 32 + k] += ptp->x * k;
+
+ ptp->y = i | j * 5;
+
+ #pragma acc loop vector
+ for (k = 0; k < 32; k++)
+ arr[i * 1024 + j * 32 + k] += pt.y * k;
+ }
+ }
+ }
+
+ for (i = 0; i < 32; i++)
+ for (int j = 0; j < 32; j++)
+ for (int k = 0; k < 32; k++)
+ {
+ int idx = i * 1024 + j * 32 + k;
+ assert (arr[idx] == idx + (i ^ j * 3) * k + (i | j * 5) * k);
+ }
+}
+
+
+/* Test of worker-private variables declared in a local scope, broadcasting
+ to vector-partitioned mode. Array worker variable. */
+
+void local_w_5()
+{
+ int i, arr[32 * 32 * 32];
+
+ for (i = 0; i < 32 * 32 * 32; i++)
+ arr[i] = i;
+
+ #pragma acc parallel copy(arr) num_gangs(32) num_workers(32) vector_length(32)
+ {
+ int j;
+
+ #pragma acc loop gang
+ for (i = 0; i < 32; i++)
+ {
+ #pragma acc loop worker
+ for (j = 0; j < 32; j++)
+ {
+ int k;
+ int pt[2];
+
+ pt[0] = i ^ j * 3;
+
+ #pragma acc loop vector
+ for (k = 0; k < 32; k++)
+ arr[i * 1024 + j * 32 + k] += pt[0] * k;
+
+ pt[1] = i | j * 5;
+
+ #pragma acc loop vector
+ for (k = 0; k < 32; k++)
+ arr[i * 1024 + j * 32 + k] += pt[1] * k;
+ }
+ }
+ }
+
+ for (i = 0; i < 32; i++)
+ for (int j = 0; j < 32; j++)
+ for (int k = 0; k < 32; k++)
+ {
+ int idx = i * 1024 + j * 32 + k;
+ assert (arr[idx] == idx + (i ^ j * 3) * k + (i | j * 5) * k);
+ }
+}
+
+
+/* Test of gang-private variables declared on loop directive. */
+
+void loop_g_1()
+{
+ int x = 5, i, arr[32];
+
+ for (i = 0; i < 32; i++)
+ arr[i] = i;
+
+ #pragma acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
+ /* { dg-warning "region is worker partitioned but does not contain worker partitioned code" "worker" { target *-*-* } 299 } */
+ /* { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 299 } */
+ {
+ #pragma acc loop gang private(x)
+ for (i = 0; i < 32; i++)
+ {
+ x = i * 2;
+ arr[i] += x;
+ }
+ }
+
+ for (i = 0; i < 32; i++)
+ assert (arr[i] == i * 3);
+}
+
+
+/* Test of gang-private variables declared on loop directive, with broadcasting
+ to partitioned workers. */
+
+void loop_g_2()
+{
+ int x = 5, i, arr[32 * 32];
+
+ for (i = 0; i < 32 * 32; i++)
+ arr[i] = i;
+
+ #pragma acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
+ /* { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 326 } */
+ {
+ #pragma acc loop gang private(x)
+ for (i = 0; i < 32; i++)
+ {
+ x = i * 2;
+
+ #pragma acc loop worker
+ for (int j = 0; j < 32; j++)
+ arr[i * 32 + j] += x;
+ }
+ }
+
+ for (i = 0; i < 32 * 32; i++)
+ assert (arr[i] == i + (i / 32) * 2);
+}
+
+
+/* Test of gang-private variables declared on loop directive, with broadcasting
+ to partitioned vectors. */
+
+void loop_g_3()
+{
+ int x = 5, i, arr[32 * 32];
+
+ for (i = 0; i < 32 * 32; i++)
+ arr[i] = i;
+
+ #pragma acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
+ /* { dg-warning "region is worker partitioned but does not contain worker partitioned code" "worker" { target *-*-* } 355 } */
+ {
+ #pragma acc loop gang private(x)
+ for (i = 0; i < 32; i++)
+ {
+ x = i * 2;
+
+ #pragma acc loop vector
+ for (int j = 0; j < 32; j++)
+ arr[i * 32 + j] += x;
+ }
+ }
+
+ for (i = 0; i < 32 * 32; i++)
+ assert (arr[i] == i + (i / 32) * 2);
+}
+
+
+/* Test of gang-private addressable variable declared on loop directive, with
+ broadcasting to partitioned workers. */
+
+void loop_g_4()
+{
+ int x = 5, i, arr[32 * 32];
+
+ for (i = 0; i < 32 * 32; i++)
+ arr[i] = i;
+
+ #pragma acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
+ /* { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 384 } */
+ {
+ #pragma acc loop gang private(x)
+ for (i = 0; i < 32; i++)
+ {
+ int *p = &x;
+
+ x = i * 2;
+
+ #pragma acc loop worker
+ for (int j = 0; j < 32; j++)
+ arr[i * 32 + j] += x;
+
+ (*p)--;
+ }
+ }
+
+ for (i = 0; i < 32 * 32; i++)
+ assert (arr[i] == i + (i / 32) * 2);
+}
+
+
+/* Test of gang-private array variable declared on loop directive, with
+ broadcasting to partitioned workers. */
+
+void loop_g_5()
+{
+ int x[8], i, arr[32 * 32];
+
+ for (i = 0; i < 32 * 32; i++)
+ arr[i] = i;
+
+ #pragma acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
+ /* { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 417 } */
+ {
+ #pragma acc loop gang private(x)
+ for (i = 0; i < 32; i++)
+ {
+ for (int j = 0; j < 8; j++)
+ x[j] = j * 2;
+
+ #pragma acc loop worker
+ for (int j = 0; j < 32; j++)
+ arr[i * 32 + j] += x[j % 8];
+ }
+ }
+
+ for (i = 0; i < 32 * 32; i++)
+ assert (arr[i] == i + (i % 8) * 2);
+}
+
+
+/* Test of gang-private aggregate variable declared on loop directive, with
+ broadcasting to partitioned workers. */
+
+void loop_g_6()
+{
+ int i, arr[32 * 32];
+ vec3_attr pt;
+
+ for (i = 0; i < 32 * 32; i++)
+ arr[i] = i;
+
+ #pragma acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
+ /* { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 448 } */
+ {
+ #pragma acc loop gang private(pt)
+ for (i = 0; i < 32; i++)
+ {
+ pt.x = i;
+ pt.y = i * 2;
+ pt.z = i * 4;
+ pt.attr[5] = i * 6;
+
+ #pragma acc loop worker
+ for (int j = 0; j < 32; j++)
+ arr[i * 32 + j] += pt.x + pt.y + pt.z + pt.attr[5];
+ }
+ }
+
+ for (i = 0; i < 32 * 32; i++)
+ assert (arr[i] == i + (i / 32) * 13);
+}
+
+
+/* Test of vector-private variables declared on loop directive. */
+
+void loop_v_1()
+{
+ int x, i, arr[32 * 32 * 32];
+
+ for (i = 0; i < 32 * 32 * 32; i++)
+ arr[i] = i;
+
+ #pragma acc parallel copy(arr) num_gangs(32) num_workers(32) vector_length(32)
+ {
+ int j;
+
+ #pragma acc loop gang
+ for (i = 0; i < 32; i++)
+ {
+ #pragma acc loop worker
+ for (j = 0; j < 32; j++)
+ {
+ int k;
+
+ #pragma acc loop vector private(x)
+ for (k = 0; k < 32; k++)
+ {
+ x = i ^ j * 3;
+ arr[i * 1024 + j * 32 + k] += x * k;
+ }
+
+ #pragma acc loop vector private(x)
+ for (k = 0; k < 32; k++)
+ {
+ x = i | j * 5;
+ arr[i * 1024 + j * 32 + k] += x * k;
+ }
+ }
+ }
+ }
+
+ for (i = 0; i < 32; i++)
+ for (int j = 0; j < 32; j++)
+ for (int k = 0; k < 32; k++)
+ {
+ int idx = i * 1024 + j * 32 + k;
+ assert (arr[idx] == idx + (i ^ j * 3) * k + (i | j * 5) * k);
+ }
+}
+
+
+/* Test of vector-private variables declared on loop directive. Array type. */
+
+void loop_v_2()
+{
+ int pt[2], i, arr[32 * 32 * 32];
+
+ for (i = 0; i < 32 * 32 * 32; i++)
+ arr[i] = i;
+
+ #pragma acc parallel copy(arr) num_gangs(32) num_workers(32) vector_length(32)
+ {
+ int j;
+
+ #pragma acc loop gang
+ for (i = 0; i < 32; i++)
+ {
+ #pragma acc loop worker
+ for (j = 0; j < 32; j++)
+ {
+ int k;
+
+ #pragma acc loop vector private(pt)
+ for (k = 0; k < 32; k++)
+ {
+ pt[0] = i ^ j * 3;
+ pt[1] = i | j * 5;
+ arr[i * 1024 + j * 32 + k] += pt[0] * k;
+ arr[i * 1024 + j * 32 + k] += pt[1] * k;
+ }
+ }
+ }
+ }
+
+ for (i = 0; i < 32; i++)
+ for (int j = 0; j < 32; j++)
+ for (int k = 0; k < 32; k++)
+ {
+ int idx = i * 1024 + j * 32 + k;
+ assert (arr[idx] == idx + (i ^ j * 3) * k + (i | j * 5) * k);
+ }
+}
+
+
+/* Test of worker-private variables declared on a loop directive. */
+
+void loop_w_1()
+{
+ int x = 5, i, arr[32 * 32];
+
+ for (i = 0; i < 32 * 32; i++)
+ arr[i] = i;
+
+ #pragma acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
+ /* { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 570 } */
+ {
+ int j;
+
+ #pragma acc loop gang
+ for (i = 0; i < 32; i++)
+ {
+ #pragma acc loop worker private(x)
+ for (j = 0; j < 32; j++)
+ {
+ x = i ^ j * 3;
+ /* Try to ensure 'x' accesses doesn't get optimized into a
+ temporary. */
+ __asm__ __volatile__ ("");
+ arr[i * 32 + j] += x;
+ }
+ }
+ }
+
+ for (i = 0; i < 32 * 32; i++)
+ assert (arr[i] == i + ((i / 32) ^ (i % 32) * 3));
+}
+
+
+/* Test of worker-private variables declared on a loop directive, broadcasting
+ to vector-partitioned mode. */
+
+void loop_w_2()
+{
+ int x = 5, i, arr[32 * 32 * 32];
+
+ for (i = 0; i < 32 * 32 * 32; i++)
+ arr[i] = i;
+
+ #pragma acc parallel copy(arr) num_gangs(32) num_workers(32) vector_length(32)
+ {
+ int j;
+
+ #pragma acc loop gang
+ for (i = 0; i < 32; i++)
+ {
+ #pragma acc loop worker private(x)
+ for (j = 0; j < 32; j++)
+ {
+ int k;
+ x = i ^ j * 3;
+
+ #pragma acc loop vector
+ for (k = 0; k < 32; k++)
+ arr[i * 1024 + j * 32 + k] += x * k;
+ }
+ }
+ }
+
+ for (i = 0; i < 32; i++)
+ for (int j = 0; j < 32; j++)
+ for (int k = 0; k < 32; k++)
+ {
+ int idx = i * 1024 + j * 32 + k;
+ assert (arr[idx] == idx + (i ^ j * 3) * k);
+ }
+}
+
+
+/* Test of worker-private variables declared on a loop directive, broadcasting
+ to vector-partitioned mode. Back-to-back worker loops. */
+
+void loop_w_3()
+{
+ int x = 5, i, arr[32 * 32 * 32];
+
+ for (i = 0; i < 32 * 32 * 32; i++)
+ arr[i] = i;
+
+ #pragma acc parallel copy(arr) num_gangs(32) num_workers(32) vector_length(32)
+ {
+ int j;
+
+ #pragma acc loop gang
+ for (i = 0; i < 32; i++)
+ {
+ #pragma acc loop worker private(x)
+ for (j = 0; j < 32; j++)
+ {
+ int k;
+ x = i ^ j * 3;
+
+ #pragma acc loop vector
+ for (k = 0; k < 32; k++)
+ arr[i * 1024 + j * 32 + k] += x * k;
+ }
+
+ #pragma acc loop worker private(x)
+ for (j = 0; j < 32; j++)
+ {
+ int k;
+ x = i | j * 5;
+
+ #pragma acc loop vector
+ for (k = 0; k < 32; k++)
+ arr[i * 1024 + j * 32 + k] += x * k;
+ }
+ }
+ }
+
+ for (i = 0; i < 32; i++)
+ for (int j = 0; j < 32; j++)
+ for (int k = 0; k < 32; k++)
+ {
+ int idx = i * 1024 + j * 32 + k;
+ assert (arr[idx] == idx + (i ^ j * 3) * k + (i | j * 5) * k);
+ }
+}
+
+
+/* Test of worker-private variables declared on a loop directive, broadcasting
+ to vector-partitioned mode. Successive vector loops. */
+
+void loop_w_4()
+{
+ int x = 5, i, arr[32 * 32 * 32];
+
+ for (i = 0; i < 32 * 32 * 32; i++)
+ arr[i] = i;
+
+ #pragma acc parallel copy(arr) num_gangs(32) num_workers(32) vector_length(32)
+ {
+ int j;
+
+ #pragma acc loop gang
+ for (i = 0; i < 32; i++)
+ {
+ #pragma acc loop worker private(x)
+ for (j = 0; j < 32; j++)
+ {
+ int k;
+ x = i ^ j * 3;
+
+ #pragma acc loop vector
+ for (k = 0; k < 32; k++)
+ arr[i * 1024 + j * 32 + k] += x * k;
+
+ x = i | j * 5;
+
+ #pragma acc loop vector
+ for (k = 0; k < 32; k++)
+ arr[i * 1024 + j * 32 + k] += x * k;
+ }
+ }
+ }
+
+ for (i = 0; i < 32; i++)
+ for (int j = 0; j < 32; j++)
+ for (int k = 0; k < 32; k++)
+ {
+ int idx = i * 1024 + j * 32 + k;
+ assert (arr[idx] == idx + (i ^ j * 3) * k + (i | j * 5) * k);
+ }
+}
+
+
+/* Test of worker-private variables declared on a loop directive, broadcasting
+ to vector-partitioned mode. Addressable worker variable. */
+
+void loop_w_5()
+{
+ int x = 5, i, arr[32 * 32 * 32];
+
+ for (i = 0; i < 32 * 32 * 32; i++)
+ arr[i] = i;
+
+ #pragma acc parallel copy(arr) num_gangs(32) num_workers(32) vector_length(32)
+ {
+ int j;
+
+ #pragma acc loop gang
+ for (i = 0; i < 32; i++)
+ {
+ #pragma acc loop worker private(x)
+ for (j = 0; j < 32; j++)
+ {
+ int k;
+ int *p = &x;
+
+ x = i ^ j * 3;
+
+ #pragma acc loop vector
+ for (k = 0; k < 32; k++)
+ arr[i * 1024 + j * 32 + k] += x * k;
+
+ *p = i | j * 5;
+
+ #pragma acc loop vector
+ for (k = 0; k < 32; k++)
+ arr[i * 1024 + j * 32 + k] += x * k;
+ }
+ }
+ }
+
+ for (i = 0; i < 32; i++)
+ for (int j = 0; j < 32; j++)
+ for (int k = 0; k < 32; k++)
+ {
+ int idx = i * 1024 + j * 32 + k;
+ assert (arr[idx] == idx + (i ^ j * 3) * k + (i | j * 5) * k);
+ }
+}
+
+
+/* Test of worker-private variables declared on a loop directive, broadcasting
+ to vector-partitioned mode. Aggregate worker variable. */
+
+void loop_w_6()
+{
+ int i, arr[32 * 32 * 32];
+ vec2 pt;
+
+ for (i = 0; i < 32 * 32 * 32; i++)
+ arr[i] = i;
+
+ #pragma acc parallel copy(arr) num_gangs(32) num_workers(32) vector_length(32)
+ {
+ int j;
+
+ #pragma acc loop gang
+ for (i = 0; i < 32; i++)
+ {
+ #pragma acc loop worker private(pt)
+ for (j = 0; j < 32; j++)
+ {
+ int k;
+
+ pt.x = i ^ j * 3;
+ pt.y = i | j * 5;
+
+ #pragma acc loop vector
+ for (k = 0; k < 32; k++)
+ arr[i * 1024 + j * 32 + k] += pt.x * k;
+
+ #pragma acc loop vector
+ for (k = 0; k < 32; k++)
+ arr[i * 1024 + j * 32 + k] += pt.y * k;
+ }
+ }
+ }
+
+ for (i = 0; i < 32; i++)
+ for (int j = 0; j < 32; j++)
+ for (int k = 0; k < 32; k++)
+ {
+ int idx = i * 1024 + j * 32 + k;
+ assert (arr[idx] == idx + (i ^ j * 3) * k + (i | j * 5) * k);
+ }
+}
+
+
+/* Test of worker-private variables declared on loop directive, broadcasting
+ to vector-partitioned mode. Array worker variable. */
+
+void loop_w_7()
+{
+ int i, arr[32 * 32 * 32];
+ int pt[2];
+
+ for (i = 0; i < 32 * 32 * 32; i++)
+ arr[i] = i;
+
+ /* "pt" is treated as "present_or_copy" on the parallel directive because it
+ is an array variable. */
+ #pragma acc parallel copy(arr) num_gangs(32) num_workers(32) vector_length(32)
+ {
+ int j;
+
+ #pragma acc loop gang
+ for (i = 0; i < 32; i++)
+ {
+ /* But here, it is made private per-worker. */
+ #pragma acc loop worker private(pt)
+ for (j = 0; j < 32; j++)
+ {
+ int k;
+
+ pt[0] = i ^ j * 3;
+
+ #pragma acc loop vector
+ for (k = 0; k < 32; k++)
+ arr[i * 1024 + j * 32 + k] += pt[0] * k;
+
+ pt[1] = i | j * 5;
+
+ #pragma acc loop vector
+ for (k = 0; k < 32; k++)
+ arr[i * 1024 + j * 32 + k] += pt[1] * k;
+ }
+ }
+ }
+
+ for (i = 0; i < 32; i++)
+ for (int j = 0; j < 32; j++)
+ for (int k = 0; k < 32; k++)
+ {
+ int idx = i * 1024 + j * 32 + k;
+ assert (arr[idx] == idx + (i ^ j * 3) * k + (i | j * 5) * k);
+ }
+}
+
+
+/* Test of gang-private variables declared on the parallel directive. */
+
+void parallel_g_1()
+{
+ int x = 5, i, arr[32];
+
+ for (i = 0; i < 32; i++)
+ arr[i] = 3;
+
+ #pragma acc parallel private(x) copy(arr) num_gangs(32) num_workers(8) vector_length(32)
+ /* { dg-warning "region is worker partitioned but does not contain worker partitioned code" "worker" { target *-*-* } 887 } */
+ /* { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 887 } */
+ {
+ #pragma acc loop gang(static:1)
+ for (i = 0; i < 32; i++)
+ x = i * 2;
+
+ #pragma acc loop gang(static:1)
+ for (i = 0; i < 32; i++)
+ {
+ if (acc_on_device (acc_device_host))
+ x = i * 2;
+ arr[i] += x;
+ }
+ }
+
+ for (i = 0; i < 32; i++)
+ assert (arr[i] == 3 + i * 2);
+}
+
+
+/* Test of gang-private array variable declared on the parallel directive. */
+
+void parallel_g_2()
+{
+ int x[32], i, arr[32 * 32];
+
+ for (i = 0; i < 32 * 32; i++)
+ arr[i] = i;
+
+ #pragma acc parallel private(x) copy(arr) num_gangs(32) num_workers(2) vector_length(32)
+ /* { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 918 } */
+ {
+ #pragma acc loop gang
+ for (i = 0; i < 32; i++)
+ {
+ int j;
+ for (j = 0; j < 32; j++)
+ x[j] = j * 2;
+
+ #pragma acc loop worker
+ for (j = 0; j < 32; j++)
+ arr[i * 32 + j] += x[31 - j];
+ }
+ }
+
+ for (i = 0; i < 32 * 32; i++)
+ assert (arr[i] == i + (31 - (i % 32)) * 2);
+}
+
+
+int main ()
+{
+ local_g_1();
+ local_w_1();
+ local_w_2();
+ local_w_3();
+ local_w_4();
+ local_w_5();
+ loop_g_1();
+ loop_g_2();
+ loop_g_3();
+ loop_g_4();
+ loop_g_5();
+ loop_g_6();
+ loop_v_1();
+ loop_v_2();
+ loop_w_1();
+ loop_w_2();
+ loop_w_3();
+ loop_w_4();
+ loop_w_5();
+ loop_w_6();
+ loop_w_7();
+ parallel_g_1();
+ parallel_g_2();
+
+ return 0;
+}
deleted file mode 100644
@@ -1,41 +0,0 @@
-/* { dg-additional-options "-w" } */
-
-#include <assert.h>
-#include <openacc.h>
-
-/* Test of gang-private variables declared in local scope with parallel
- directive. */
-
-#define ACTUAL_GANGS 32
-
-int
-main (int argc, char* argv[])
-{
- int x = 5, i, arr[ACTUAL_GANGS];
-
- for (i = 0; i < ACTUAL_GANGS; i++)
- arr[i] = 3;
-
- #pragma acc parallel copy(arr) num_gangs(ACTUAL_GANGS) num_workers(8) \
- vector_length(32)
- {
- int x;
-
- #pragma acc loop gang(static:1)
- for (i = 0; i < ACTUAL_GANGS; i++)
- x = i * 2;
-
- #pragma acc loop gang(static:1)
- for (i = 0; i < ACTUAL_GANGS; i++)
- {
- if (acc_on_device (acc_device_host))
- x = i * 2;
- arr[i] += x;
- }
- }
-
- for (i = 0; i < ACTUAL_GANGS; i++)
- assert (arr[i] == 3 + i * 2);
-
- return 0;
-}
deleted file mode 100644
@@ -1,54 +0,0 @@
-#include <assert.h>
-
-/* Test of worker-private variables declared in a local scope, broadcasting
- to vector-partitioned mode. Back-to-back worker loops. */
-
-int
-main (int argc, char* argv[])
-{
- int i, arr[32 * 32 * 32];
-
- for (i = 0; i < 32 * 32 * 32; i++)
- arr[i] = i;
-
- #pragma acc parallel copy(arr) num_gangs(32) num_workers(32) vector_length(32)
- {
- int j;
-
- #pragma acc loop gang
- for (i = 0; i < 32; i++)
- {
- #pragma acc loop worker
- for (j = 0; j < 32; j++)
- {
- int k;
- int x = i ^ j * 3;
-
- #pragma acc loop vector
- for (k = 0; k < 32; k++)
- arr[i * 1024 + j * 32 + k] += x * k;
- }
-
- #pragma acc loop worker
- for (j = 0; j < 32; j++)
- {
- int k;
- int x = i | j * 5;
-
- #pragma acc loop vector
- for (k = 0; k < 32; k++)
- arr[i * 1024 + j * 32 + k] += x * k;
- }
- }
- }
-
- for (i = 0; i < 32; i++)
- for (int j = 0; j < 32; j++)
- for (int k = 0; k < 32; k++)
- {
- int idx = i * 1024 + j * 32 + k;
- assert (arr[idx] == idx + (i ^ j * 3) * k + (i | j * 5) * k);
- }
-
- return 0;
-}
deleted file mode 100644
@@ -1,49 +0,0 @@
-#include <assert.h>
-
-/* Test of worker-private variables declared in a local scope, broadcasting
- to vector-partitioned mode. Successive vector loops. */
-
-int
-main (int argc, char* argv[])
-{
- int x = 5, i, arr[32 * 32 * 32];
-
- for (i = 0; i < 32 * 32 * 32; i++)
- arr[i] = i;
-
- #pragma acc parallel copy(arr) num_gangs(32) num_workers(32) vector_length(32)
- {
- int j;
-
- #pragma acc loop gang
- for (i = 0; i < 32; i++)
- {
- #pragma acc loop worker
- for (j = 0; j < 32; j++)
- {
- int k;
- int x = i ^ j * 3;
-
- #pragma acc loop vector
- for (k = 0; k < 32; k++)
- arr[i * 1024 + j * 32 + k] += x * k;
-
- x = i | j * 5;
-
- #pragma acc loop vector
- for (k = 0; k < 32; k++)
- arr[i * 1024 + j * 32 + k] += x * k;
- }
- }
- }
-
- for (i = 0; i < 32; i++)
- for (int j = 0; j < 32; j++)
- for (int k = 0; k < 32; k++)
- {
- int idx = i * 1024 + j * 32 + k;
- assert (arr[idx] == idx + (i ^ j * 3) * k + (i | j * 5) * k);
- }
-
- return 0;
-}
deleted file mode 100644
@@ -1,55 +0,0 @@
-#include <assert.h>
-
-/* Test of worker-private variables declared in a local scope, broadcasting
- to vector-partitioned mode. Aggregate worker variable. */
-
-typedef struct
-{
- int x, y;
-} vec2;
-
-int
-main (int argc, char* argv[])
-{
- int i, arr[32 * 32 * 32];
-
- for (i = 0; i < 32 * 32 * 32; i++)
- arr[i] = i;
-
- #pragma acc parallel copy(arr) num_gangs(32) num_workers(32) vector_length(32)
- {
- int j;
-
- #pragma acc loop gang
- for (i = 0; i < 32; i++)
- {
- #pragma acc loop worker
- for (j = 0; j < 32; j++)
- {
- int k;
- vec2 pt;
-
- pt.x = i ^ j * 3;
- pt.y = i | j * 5;
-
- #pragma acc loop vector
- for (k = 0; k < 32; k++)
- arr[i * 1024 + j * 32 + k] += pt.x * k;
-
- #pragma acc loop vector
- for (k = 0; k < 32; k++)
- arr[i * 1024 + j * 32 + k] += pt.y * k;
- }
- }
- }
-
- for (i = 0; i < 32; i++)
- for (int j = 0; j < 32; j++)
- for (int k = 0; k < 32; k++)
- {
- int idx = i * 1024 + j * 32 + k;
- assert (arr[idx] == idx + (i ^ j * 3) * k + (i | j * 5) * k);
- }
-
- return 0;
-}
deleted file mode 100644
@@ -1,58 +0,0 @@
-#include <assert.h>
-
-/* Test of worker-private variables declared in a local scope, broadcasting
- to vector-partitioned mode. Addressable worker variable. */
-
-typedef struct
-{
- int x, y;
-} vec2;
-
-int
-main (int argc, char* argv[])
-{
- int i, arr[32 * 32 * 32];
-
- for (i = 0; i < 32 * 32 * 32; i++)
- arr[i] = i;
-
- #pragma acc parallel copy(arr) num_gangs(32) num_workers(32) vector_length(32)
- {
- int j;
-
- #pragma acc loop gang
- for (i = 0; i < 32; i++)
- {
- #pragma acc loop worker
- for (j = 0; j < 32; j++)
- {
- int k;
- vec2 pt, *ptp;
-
- ptp = &pt;
-
- pt.x = i ^ j * 3;
-
- #pragma acc loop vector
- for (k = 0; k < 32; k++)
- arr[i * 1024 + j * 32 + k] += ptp->x * k;
-
- ptp->y = i | j * 5;
-
- #pragma acc loop vector
- for (k = 0; k < 32; k++)
- arr[i * 1024 + j * 32 + k] += pt.y * k;
- }
- }
- }
-
- for (i = 0; i < 32; i++)
- for (int j = 0; j < 32; j++)
- for (int k = 0; k < 32; k++)
- {
- int idx = i * 1024 + j * 32 + k;
- assert (arr[idx] == idx + (i ^ j * 3) * k + (i | j * 5) * k);
- }
-
- return 0;
-}
deleted file mode 100644
@@ -1,51 +0,0 @@
-#include <assert.h>
-
-/* Test of worker-private variables declared in a local scope, broadcasting
- to vector-partitioned mode. Array worker variable. */
-
-int
-main (int argc, char* argv[])
-{
- int i, arr[32 * 32 * 32];
-
- for (i = 0; i < 32 * 32 * 32; i++)
- arr[i] = i;
-
- #pragma acc parallel copy(arr) num_gangs(32) num_workers(32) vector_length(32)
- {
- int j;
-
- #pragma acc loop gang
- for (i = 0; i < 32; i++)
- {
- #pragma acc loop worker
- for (j = 0; j < 32; j++)
- {
- int k;
- int pt[2];
-
- pt[0] = i ^ j * 3;
-
- #pragma acc loop vector
- for (k = 0; k < 32; k++)
- arr[i * 1024 + j * 32 + k] += pt[0] * k;
-
- pt[1] = i | j * 5;
-
- #pragma acc loop vector
- for (k = 0; k < 32; k++)
- arr[i * 1024 + j * 32 + k] += pt[1] * k;
- }
- }
- }
-
- for (i = 0; i < 32; i++)
- for (int j = 0; j < 32; j++)
- for (int k = 0; k < 32; k++)
- {
- int idx = i * 1024 + j * 32 + k;
- assert (arr[idx] == idx + (i ^ j * 3) * k + (i | j * 5) * k);
- }
-
- return 0;
-}
deleted file mode 100644
@@ -1,29 +0,0 @@
-/* { dg-additional-options "-w" } */
-
-#include <assert.h>
-
-/* Test of gang-private variables declared on loop directive. */
-
-int
-main (int argc, char* argv[])
-{
- int x = 5, i, arr[32];
-
- for (i = 0; i < 32; i++)
- arr[i] = i;
-
- #pragma acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
- {
- #pragma acc loop gang private(x)
- for (i = 0; i < 32; i++)
- {
- x = i * 2;
- arr[i] += x;
- }
- }
-
- for (i = 0; i < 32; i++)
- assert (arr[i] == i * 3);
-
- return 0;
-}
deleted file mode 100644
@@ -1,33 +0,0 @@
-/* { dg-additional-options "-w" } */
-
-#include <assert.h>
-
-/* Test of gang-private variables declared on loop directive, with broadcasting
- to partitioned workers. */
-
-int
-main (int argc, char* argv[])
-{
- int x = 5, i, arr[32 * 32];
-
- for (i = 0; i < 32 * 32; i++)
- arr[i] = i;
-
- #pragma acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
- {
- #pragma acc loop gang private(x)
- for (i = 0; i < 32; i++)
- {
- x = i * 2;
-
- #pragma acc loop worker
- for (int j = 0; j < 32; j++)
- arr[i * 32 + j] += x;
- }
- }
-
- for (i = 0; i < 32 * 32; i++)
- assert (arr[i] == i + (i / 32) * 2);
-
- return 0;
-}
deleted file mode 100644
@@ -1,33 +0,0 @@
-/* { dg-additional-options "-w" } */
-
-#include <assert.h>
-
-/* Test of gang-private variables declared on loop directive, with broadcasting
- to partitioned vectors. */
-
-int
-main (int argc, char* argv[])
-{
- int x = 5, i, arr[32 * 32];
-
- for (i = 0; i < 32 * 32; i++)
- arr[i] = i;
-
- #pragma acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
- {
- #pragma acc loop gang private(x)
- for (i = 0; i < 32; i++)
- {
- x = i * 2;
-
- #pragma acc loop vector
- for (int j = 0; j < 32; j++)
- arr[i * 32 + j] += x;
- }
- }
-
- for (i = 0; i < 32 * 32; i++)
- assert (arr[i] == i + (i / 32) * 2);
-
- return 0;
-}
deleted file mode 100644
@@ -1,37 +0,0 @@
-/* { dg-additional-options "-w" } */
-
-#include <assert.h>
-
-/* Test of gang-private addressable variable declared on loop directive, with
- broadcasting to partitioned workers. */
-
-int
-main (int argc, char* argv[])
-{
- int x = 5, i, arr[32 * 32];
-
- for (i = 0; i < 32 * 32; i++)
- arr[i] = i;
-
- #pragma acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
- {
- #pragma acc loop gang private(x)
- for (i = 0; i < 32; i++)
- {
- int *p = &x;
-
- x = i * 2;
-
- #pragma acc loop worker
- for (int j = 0; j < 32; j++)
- arr[i * 32 + j] += x;
-
- (*p)--;
- }
- }
-
- for (i = 0; i < 32 * 32; i++)
- assert (arr[i] == i + (i / 32) * 2);
-
- return 0;
-}
deleted file mode 100644
@@ -1,34 +0,0 @@
-/* { dg-additional-options "-w" } */
-
-#include <assert.h>
-
-/* Test of gang-private array variable declared on loop directive, with
- broadcasting to partitioned workers. */
-
-int
-main (int argc, char* argv[])
-{
- int x[8], i, arr[32 * 32];
-
- for (i = 0; i < 32 * 32; i++)
- arr[i] = i;
-
- #pragma acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
- {
- #pragma acc loop gang private(x)
- for (i = 0; i < 32; i++)
- {
- for (int j = 0; j < 8; j++)
- x[j] = j * 2;
-
- #pragma acc loop worker
- for (int j = 0; j < 32; j++)
- arr[i * 32 + j] += x[j % 8];
- }
- }
-
- for (i = 0; i < 32 * 32; i++)
- assert (arr[i] == i + (i % 8) * 2);
-
- return 0;
-}
deleted file mode 100644
@@ -1,42 +0,0 @@
-/* { dg-additional-options "-w" } */
-
-#include <assert.h>
-
-/* Test of gang-private aggregate variable declared on loop directive, with
- broadcasting to partitioned workers. */
-
-typedef struct {
- int x, y, z;
- int attr[13];
-} vec3;
-
-int
-main (int argc, char* argv[])
-{
- int i, arr[32 * 32];
- vec3 pt;
-
- for (i = 0; i < 32 * 32; i++)
- arr[i] = i;
-
- #pragma acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
- {
- #pragma acc loop gang private(pt)
- for (i = 0; i < 32; i++)
- {
- pt.x = i;
- pt.y = i * 2;
- pt.z = i * 4;
- pt.attr[5] = i * 6;
-
- #pragma acc loop worker
- for (int j = 0; j < 32; j++)
- arr[i * 32 + j] += pt.x + pt.y + pt.z + pt.attr[5];
- }
- }
-
- for (i = 0; i < 32 * 32; i++)
- assert (arr[i] == i + (i / 32) * 13);
-
- return 0;
-}
deleted file mode 100644
@@ -1,51 +0,0 @@
-#include <assert.h>
-
-/* Test of vector-private variables declared on loop directive. */
-
-int
-main (int argc, char* argv[])
-{
- int x, i, arr[32 * 32 * 32];
-
- for (i = 0; i < 32 * 32 * 32; i++)
- arr[i] = i;
-
- #pragma acc parallel copy(arr) num_gangs(32) num_workers(32) vector_length(32)
- {
- int j;
-
- #pragma acc loop gang
- for (i = 0; i < 32; i++)
- {
- #pragma acc loop worker
- for (j = 0; j < 32; j++)
- {
- int k;
-
- #pragma acc loop vector private(x)
- for (k = 0; k < 32; k++)
- {
- x = i ^ j * 3;
- arr[i * 1024 + j * 32 + k] += x * k;
- }
-
- #pragma acc loop vector private(x)
- for (k = 0; k < 32; k++)
- {
- x = i | j * 5;
- arr[i * 1024 + j * 32 + k] += x * k;
- }
- }
- }
- }
-
- for (i = 0; i < 32; i++)
- for (int j = 0; j < 32; j++)
- for (int k = 0; k < 32; k++)
- {
- int idx = i * 1024 + j * 32 + k;
- assert (arr[idx] == idx + (i ^ j * 3) * k + (i | j * 5) * k);
- }
-
- return 0;
-}
deleted file mode 100644
@@ -1,46 +0,0 @@
-#include <assert.h>
-
-/* Test of vector-private variables declared on loop directive. Array type. */
-
-int
-main (int argc, char* argv[])
-{
- int pt[2], i, arr[32 * 32 * 32];
-
- for (i = 0; i < 32 * 32 * 32; i++)
- arr[i] = i;
-
- #pragma acc parallel copy(arr) num_gangs(32) num_workers(32) vector_length(32)
- {
- int j;
-
- #pragma acc loop gang
- for (i = 0; i < 32; i++)
- {
- #pragma acc loop worker
- for (j = 0; j < 32; j++)
- {
- int k;
-
- #pragma acc loop vector private(pt)
- for (k = 0; k < 32; k++)
- {
- pt[0] = i ^ j * 3;
- pt[1] = i | j * 5;
- arr[i * 1024 + j * 32 + k] += pt[0] * k;
- arr[i * 1024 + j * 32 + k] += pt[1] * k;
- }
- }
- }
- }
-
- for (i = 0; i < 32; i++)
- for (int j = 0; j < 32; j++)
- for (int k = 0; k < 32; k++)
- {
- int idx = i * 1024 + j * 32 + k;
- assert (arr[idx] == idx + (i ^ j * 3) * k + (i | j * 5) * k);
- }
-
- return 0;
-}
deleted file mode 100644
@@ -1,38 +0,0 @@
-/* { dg-additional-options "-w" } */
-
-#include <assert.h>
-
-/* Test of worker-private variables declared on a loop directive. */
-
-int
-main (int argc, char* argv[])
-{
- int x = 5, i, arr[32 * 32];
-
- for (i = 0; i < 32 * 32; i++)
- arr[i] = i;
-
- #pragma acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
- {
- int j;
-
- #pragma acc loop gang
- for (i = 0; i < 32; i++)
- {
- #pragma acc loop worker private(x)
- for (j = 0; j < 32; j++)
- {
- x = i ^ j * 3;
- /* Try to ensure 'x' accesses doesn't get optimized into a
- temporary. */
- __asm__ __volatile__ ("");
- arr[i * 32 + j] += x;
- }
- }
- }
-
- for (i = 0; i < 32 * 32; i++)
- assert (arr[i] == i + ((i / 32) ^ (i % 32) * 3));
-
- return 0;
-}
deleted file mode 100644
@@ -1,43 +0,0 @@
-#include <assert.h>
-
-/* Test of worker-private variables declared on a loop directive, broadcasting
- to vector-partitioned mode. */
-
-int
-main (int argc, char* argv[])
-{
- int x = 5, i, arr[32 * 32 * 32];
-
- for (i = 0; i < 32 * 32 * 32; i++)
- arr[i] = i;
-
- #pragma acc parallel copy(arr) num_gangs(32) num_workers(32) vector_length(32)
- {
- int j;
-
- #pragma acc loop gang
- for (i = 0; i < 32; i++)
- {
- #pragma acc loop worker private(x)
- for (j = 0; j < 32; j++)
- {
- int k;
- x = i ^ j * 3;
-
- #pragma acc loop vector
- for (k = 0; k < 32; k++)
- arr[i * 1024 + j * 32 + k] += x * k;
- }
- }
- }
-
- for (i = 0; i < 32; i++)
- for (int j = 0; j < 32; j++)
- for (int k = 0; k < 32; k++)
- {
- int idx = i * 1024 + j * 32 + k;
- assert (arr[idx] == idx + (i ^ j * 3) * k);
- }
-
- return 0;
-}
deleted file mode 100644
@@ -1,54 +0,0 @@
-#include <assert.h>
-
-/* Test of worker-private variables declared on a loop directive, broadcasting
- to vector-partitioned mode. Back-to-back worker loops. */
-
-int
-main (int argc, char* argv[])
-{
- int x = 5, i, arr[32 * 32 * 32];
-
- for (i = 0; i < 32 * 32 * 32; i++)
- arr[i] = i;
-
- #pragma acc parallel copy(arr) num_gangs(32) num_workers(32) vector_length(32)
- {
- int j;
-
- #pragma acc loop gang
- for (i = 0; i < 32; i++)
- {
- #pragma acc loop worker private(x)
- for (j = 0; j < 32; j++)
- {
- int k;
- x = i ^ j * 3;
-
- #pragma acc loop vector
- for (k = 0; k < 32; k++)
- arr[i * 1024 + j * 32 + k] += x * k;
- }
-
- #pragma acc loop worker private(x)
- for (j = 0; j < 32; j++)
- {
- int k;
- x = i | j * 5;
-
- #pragma acc loop vector
- for (k = 0; k < 32; k++)
- arr[i * 1024 + j * 32 + k] += x * k;
- }
- }
- }
-
- for (i = 0; i < 32; i++)
- for (int j = 0; j < 32; j++)
- for (int k = 0; k < 32; k++)
- {
- int idx = i * 1024 + j * 32 + k;
- assert (arr[idx] == idx + (i ^ j * 3) * k + (i | j * 5) * k);
- }
-
- return 0;
-}
deleted file mode 100644
@@ -1,49 +0,0 @@
-#include <assert.h>
-
-/* Test of worker-private variables declared on a loop directive, broadcasting
- to vector-partitioned mode. Successive vector loops. */
-
-int
-main (int argc, char* argv[])
-{
- int x = 5, i, arr[32 * 32 * 32];
-
- for (i = 0; i < 32 * 32 * 32; i++)
- arr[i] = i;
-
- #pragma acc parallel copy(arr) num_gangs(32) num_workers(32) vector_length(32)
- {
- int j;
-
- #pragma acc loop gang
- for (i = 0; i < 32; i++)
- {
- #pragma acc loop worker private(x)
- for (j = 0; j < 32; j++)
- {
- int k;
- x = i ^ j * 3;
-
- #pragma acc loop vector
- for (k = 0; k < 32; k++)
- arr[i * 1024 + j * 32 + k] += x * k;
-
- x = i | j * 5;
-
- #pragma acc loop vector
- for (k = 0; k < 32; k++)
- arr[i * 1024 + j * 32 + k] += x * k;
- }
- }
- }
-
- for (i = 0; i < 32; i++)
- for (int j = 0; j < 32; j++)
- for (int k = 0; k < 32; k++)
- {
- int idx = i * 1024 + j * 32 + k;
- assert (arr[idx] == idx + (i ^ j * 3) * k + (i | j * 5) * k);
- }
-
- return 0;
-}
deleted file mode 100644
@@ -1,51 +0,0 @@
-#include <assert.h>
-
-/* Test of worker-private variables declared on a loop directive, broadcasting
- to vector-partitioned mode. Addressable worker variable. */
-
-int
-main (int argc, char* argv[])
-{
- int x = 5, i, arr[32 * 32 * 32];
-
- for (i = 0; i < 32 * 32 * 32; i++)
- arr[i] = i;
-
- #pragma acc parallel copy(arr) num_gangs(32) num_workers(32) vector_length(32)
- {
- int j;
-
- #pragma acc loop gang
- for (i = 0; i < 32; i++)
- {
- #pragma acc loop worker private(x)
- for (j = 0; j < 32; j++)
- {
- int k;
- int *p = &x;
-
- x = i ^ j * 3;
-
- #pragma acc loop vector
- for (k = 0; k < 32; k++)
- arr[i * 1024 + j * 32 + k] += x * k;
-
- *p = i | j * 5;
-
- #pragma acc loop vector
- for (k = 0; k < 32; k++)
- arr[i * 1024 + j * 32 + k] += x * k;
- }
- }
- }
-
- for (i = 0; i < 32; i++)
- for (int j = 0; j < 32; j++)
- for (int k = 0; k < 32; k++)
- {
- int idx = i * 1024 + j * 32 + k;
- assert (arr[idx] == idx + (i ^ j * 3) * k + (i | j * 5) * k);
- }
-
- return 0;
-}
deleted file mode 100644
@@ -1,55 +0,0 @@
-#include <assert.h>
-
-/* Test of worker-private variables declared on a loop directive, broadcasting
- to vector-partitioned mode. Aggregate worker variable. */
-
-typedef struct
-{
- int x, y;
-} vec2;
-
-int
-main (int argc, char* argv[])
-{
- int i, arr[32 * 32 * 32];
- vec2 pt;
-
- for (i = 0; i < 32 * 32 * 32; i++)
- arr[i] = i;
-
- #pragma acc parallel copy(arr) num_gangs(32) num_workers(32) vector_length(32)
- {
- int j;
-
- #pragma acc loop gang
- for (i = 0; i < 32; i++)
- {
- #pragma acc loop worker private(pt)
- for (j = 0; j < 32; j++)
- {
- int k;
-
- pt.x = i ^ j * 3;
- pt.y = i | j * 5;
-
- #pragma acc loop vector
- for (k = 0; k < 32; k++)
- arr[i * 1024 + j * 32 + k] += pt.x * k;
-
- #pragma acc loop vector
- for (k = 0; k < 32; k++)
- arr[i * 1024 + j * 32 + k] += pt.y * k;
- }
- }
- }
-
- for (i = 0; i < 32; i++)
- for (int j = 0; j < 32; j++)
- for (int k = 0; k < 32; k++)
- {
- int idx = i * 1024 + j * 32 + k;
- assert (arr[idx] == idx + (i ^ j * 3) * k + (i | j * 5) * k);
- }
-
- return 0;
-}
deleted file mode 100644
@@ -1,54 +0,0 @@
-#include <assert.h>
-
-/* Test of worker-private variables declared on loop directive, broadcasting
- to vector-partitioned mode. Array worker variable. */
-
-int
-main (int argc, char* argv[])
-{
- int i, arr[32 * 32 * 32];
- int pt[2];
-
- for (i = 0; i < 32 * 32 * 32; i++)
- arr[i] = i;
-
- /* "pt" is treated as "present_or_copy" on the parallel directive because it
- is an array variable. */
- #pragma acc parallel copy(arr) num_gangs(32) num_workers(32) vector_length(32)
- {
- int j;
-
- #pragma acc loop gang
- for (i = 0; i < 32; i++)
- {
- /* But here, it is made private per-worker. */
- #pragma acc loop worker private(pt)
- for (j = 0; j < 32; j++)
- {
- int k;
-
- pt[0] = i ^ j * 3;
-
- #pragma acc loop vector
- for (k = 0; k < 32; k++)
- arr[i * 1024 + j * 32 + k] += pt[0] * k;
-
- pt[1] = i | j * 5;
-
- #pragma acc loop vector
- for (k = 0; k < 32; k++)
- arr[i * 1024 + j * 32 + k] += pt[1] * k;
- }
- }
- }
-
- for (i = 0; i < 32; i++)
- for (int j = 0; j < 32; j++)
- for (int k = 0; k < 32; k++)
- {
- int idx = i * 1024 + j * 32 + k;
- assert (arr[idx] == idx + (i ^ j * 3) * k + (i | j * 5) * k);
- }
-
- return 0;
-}
deleted file mode 100644
@@ -1,27 +0,0 @@
-/* { dg-additional-options "-w" } */
-
-#include <assert.h>
-
-/* Basic test of firstprivate variable. */
-
-int
-main (int argc, char* argv[])
-{
- int x = 5, i, arr[32];
-
- for (i = 0; i < 32; i++)
- arr[i] = 3;
-
- #pragma acc parallel firstprivate(x) copy(arr) num_gangs(32) num_workers(8) \
- vector_length(32)
- {
- #pragma acc loop gang
- for (i = 0; i < 32; i++)
- arr[i] += x;
- }
-
- for (i = 0; i < 32; i++)
- assert (arr[i] == 8);
-
- return 0;
-}
deleted file mode 100644
@@ -1,38 +0,0 @@
-/* { dg-additional-options "-w" } */
-
-#include <assert.h>
-#include <openacc.h>
-
-/* Test of gang-private variables declared on the parallel directive. */
-
-#define ACTUAL_GANGS 32
-
-int
-main (int argc, char* argv[])
-{
- int x = 5, i, arr[ACTUAL_GANGS];
-
- for (i = 0; i < ACTUAL_GANGS; i++)
- arr[i] = 3;
-
- #pragma acc parallel private(x) copy(arr) num_gangs(ACTUAL_GANGS) \
- num_workers(8) vector_length(32)
- {
- #pragma acc loop gang(static:1)
- for (i = 0; i < ACTUAL_GANGS; i++)
- x = i * 2;
-
- #pragma acc loop gang(static:1)
- for (i = 0; i < ACTUAL_GANGS; i++)
- {
- if (acc_on_device (acc_device_host))
- x = i * 2;
- arr[i] += x;
- }
- }
-
- for (i = 0; i < ACTUAL_GANGS; i++)
- assert (arr[i] == 3 + i * 2);
-
- return 0;
-}
deleted file mode 100644
@@ -1,35 +0,0 @@
-/* { dg-additional-options "-w" } */
-
-#include <assert.h>
-
-/* Test of gang-private array variable declared on the parallel directive. */
-
-int
-main (int argc, char* argv[])
-{
- int x[32], i, arr[32 * 32];
-
- for (i = 0; i < 32 * 32; i++)
- arr[i] = i;
-
- #pragma acc parallel private(x) copy(arr) num_gangs(32) num_workers(2) \
- vector_length(32)
- {
- #pragma acc loop gang
- for (i = 0; i < 32; i++)
- {
- int j;
- for (j = 0; j < 32; j++)
- x[j] = j * 2;
-
- #pragma acc loop worker
- for (j = 0; j < 32; j++)
- arr[i * 32 + j] += x[31 - j];
- }
- }
-
- for (i = 0; i < 32 * 32; i++)
- assert (arr[i] == i + (31 - (i % 32)) * 2);
-
- return 0;
-}
new file mode 100644
@@ -0,0 +1,488 @@
+/* Tests of reduction on loop directive. */
+
+#include <assert.h>
+
+
+/* Test of reduction on loop directive (gangs, non-private reduction
+ variable). */
+
+void g_np_1()
+{
+ int i, arr[1024], res = 0, hres = 0;
+
+ for (i = 0; i < 1024; i++)
+ arr[i] = i;
+
+ #pragma acc parallel num_gangs(32) num_workers(32) vector_length(32) \
+ copy(res)
+ /* { dg-warning "region is worker partitioned but does not contain worker partitioned code" "worker" { target *-*-* } 16 } */
+ /* { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 16 } */
+ {
+ #pragma acc loop gang reduction(+:res)
+ for (i = 0; i < 1024; i++)
+ res += arr[i];
+ }
+
+ for (i = 0; i < 1024; i++)
+ hres += arr[i];
+
+ assert (res == hres);
+
+ res = hres = 1;
+
+ #pragma acc parallel num_gangs(32) num_workers(32) vector_length(32) \
+ copy(res)
+ /* { dg-warning "region is worker partitioned but does not contain worker partitioned code" "worker" { target *-*-* } 33 } */
+ /* { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 33 } */
+ {
+ #pragma acc loop gang reduction(*:res)
+ for (i = 0; i < 12; i++)
+ res *= arr[i];
+ }
+
+ for (i = 0; i < 12; i++)
+ hres *= arr[i];
+
+ assert (res == hres);
+}
+
+
+/* Test of reduction on loop directive (gangs and vectors, non-private
+ reduction variable). */
+
+void gv_np_1()
+{
+ int i, arr[1024], res = 0, hres = 0;
+
+ for (i = 0; i < 1024; i++)
+ arr[i] = i;
+
+ #pragma acc parallel num_gangs(32) num_workers(32) vector_length(32) \
+ copy(res)
+ /* { dg-warning "region is worker partitioned but does not contain worker partitioned code" "worker" { target *-*-* } 60 } */
+ {
+ #pragma acc loop gang vector reduction(+:res)
+ for (i = 0; i < 1024; i++)
+ res += arr[i];
+ }
+
+ for (i = 0; i < 1024; i++)
+ hres += arr[i];
+
+ assert (res == hres);
+}
+
+
+/* Test of reduction on loop directive (gangs and workers, non-private
+ reduction variable). */
+
+void gw_np_1()
+{
+ int i, arr[1024], res = 0, hres = 0;
+
+ for (i = 0; i < 1024; i++)
+ arr[i] = i;
+
+ #pragma acc parallel num_gangs(32) num_workers(32) vector_length(32) \
+ copy(res)
+ /* { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 86 } */
+ {
+ #pragma acc loop gang worker reduction(+:res)
+ for (i = 0; i < 1024; i++)
+ res += arr[i];
+ }
+
+ for (i = 0; i < 1024; i++)
+ hres += arr[i];
+
+ assert (res == hres);
+}
+
+
+/* Test of reduction on loop directive (gangs, workers and vectors, non-private
+ reduction variable). */
+
+void gwv_np_1()
+{
+ int i, arr[1024], res = 0, hres = 0;
+
+ for (i = 0; i < 1024; i++)
+ arr[i] = i;
+
+ #pragma acc parallel num_gangs(32) num_workers(32) vector_length(32) \
+ copy(res)
+ {
+ #pragma acc loop gang worker vector reduction(+:res)
+ for (i = 0; i < 1024; i++)
+ res += arr[i];
+ }
+
+ for (i = 0; i < 1024; i++)
+ hres += arr[i];
+
+ assert (res == hres);
+}
+
+
+/* Test of reduction on loop directive (gangs, workers and vectors, non-private
+ reduction variable: separate gang and worker/vector loops). */
+
+void gwv_np_2()
+{
+ int i, j, arr[32768], res = 0, hres = 0;
+
+ for (i = 0; i < 32768; i++)
+ arr[i] = i;
+
+ #pragma acc parallel num_gangs(32) num_workers(32) vector_length(32) \
+ copy(res)
+ {
+ #pragma acc loop gang reduction(+:res)
+ for (j = 0; j < 32; j++)
+ {
+ #pragma acc loop worker vector reduction(+:res)
+ for (i = 0; i < 1024; i++)
+ res += arr[j * 1024 + i];
+ }
+ /* "res" is non-private, and is not available until after the parallel
+ region. */
+ }
+
+ for (i = 0; i < 32768; i++)
+ hres += arr[i];
+
+ assert (res == hres);
+}
+
+
+/* Test of reduction on loop directive (gangs, workers and vectors, non-private
+ reduction variable: separate gang and worker/vector loops). */
+
+void gwv_np_3()
+{
+ int i, j;
+ double arr[32768], res = 0, hres = 0;
+
+ for (i = 0; i < 32768; i++)
+ arr[i] = i;
+
+ #pragma acc parallel num_gangs(32) num_workers(32) vector_length(32) \
+ copyin(arr) copy(res)
+ {
+ #pragma acc loop gang reduction(+:res)
+ for (j = 0; j < 32; j++)
+ {
+ #pragma acc loop worker vector reduction(+:res)
+ for (i = 0; i < 1024; i++)
+ res += arr[j * 1024 + i];
+ }
+ }
+
+ for (i = 0; i < 32768; i++)
+ hres += arr[i];
+
+ assert (res == hres);
+}
+
+
+/* Test of reduction on loop directive (gangs, workers and vectors, multiple
+ non-private reduction variables, float type). */
+
+void gwv_np_4()
+{
+ int i, j;
+ float arr[32768];
+ float res = 0, mres = 0, hres = 0, hmres = 0;
+
+ for (i = 0; i < 32768; i++)
+ arr[i] = i;
+
+ #pragma acc parallel num_gangs(32) num_workers(32) vector_length(32) \
+ copy(res, mres)
+ {
+ #pragma acc loop gang reduction(+:res) reduction(max:mres)
+ for (j = 0; j < 32; j++)
+ {
+ #pragma acc loop worker vector reduction(+:res) reduction(max:mres)
+ for (i = 0; i < 1024; i++)
+ {
+ res += arr[j * 1024 + i];
+ if (arr[j * 1024 + i] > mres)
+ mres = arr[j * 1024 + i];
+ }
+
+ #pragma acc loop worker vector reduction(+:res) reduction(max:mres)
+ for (i = 0; i < 1024; i++)
+ {
+ res += arr[j * 1024 + (1023 - i)];
+ if (arr[j * 1024 + (1023 - i)] > mres)
+ mres = arr[j * 1024 + (1023 - i)];
+ }
+ }
+ }
+
+ for (j = 0; j < 32; j++)
+ for (i = 0; i < 1024; i++)
+ {
+ hres += arr[j * 1024 + i];
+ hres += arr[j * 1024 + (1023 - i)];
+ if (arr[j * 1024 + i] > hmres)
+ hmres = arr[j * 1024 + i];
+ if (arr[j * 1024 + (1023 - i)] > hmres)
+ hmres = arr[j * 1024 + (1023 - i)];
+ }
+
+ assert (res == hres);
+ assert (mres == hmres);
+}
+
+
+/* Test of reduction on loop directive (vectors, private reduction
+ variable). */
+
+void v_p_1()
+{
+ int i, j, arr[1024], out[32], res = 0, hres = 0;
+
+ for (i = 0; i < 1024; i++)
+ arr[i] = i;
+
+ #pragma acc parallel num_gangs(32) num_workers(32) vector_length(32) \
+ private(res) copyout(out)
+ /* { dg-warning "region is worker partitioned but does not contain worker partitioned code" "worker" { target *-*-* } 250 } */
+ {
+ #pragma acc loop gang
+ for (j = 0; j < 32; j++)
+ {
+ res = 0;
+
+ #pragma acc loop vector reduction(+:res)
+ for (i = 0; i < 32; i++)
+ res += arr[j * 32 + i];
+
+ out[j] = res;
+ }
+ }
+
+ for (j = 0; j < 32; j++)
+ {
+ hres = 0;
+
+ for (i = 0; i < 32; i++)
+ hres += arr[j * 32 + i];
+
+ assert (out[j] == hres);
+ }
+}
+
+
+/* Test of reduction on loop directive (vector reduction in
+ gang-partitioned/worker-partitioned mode, private reduction variable). */
+
+void v_p_2()
+{
+ int i, j, k;
+ double ina[1024], inb[1024], out[1024], acc;
+
+ for (j = 0; j < 32; j++)
+ for (i = 0; i < 32; i++)
+ {
+ ina[j * 32 + i] = (i == j) ? 2.0 : 0.0;
+ inb[j * 32 + i] = (double) (i + j);
+ }
+
+ #pragma acc parallel num_gangs(32) num_workers(32) vector_length(32) \
+ private(acc) copyin(ina, inb) copyout(out)
+ {
+ #pragma acc loop gang worker
+ for (k = 0; k < 32; k++)
+ for (j = 0; j < 32; j++)
+ {
+ acc = 0;
+
+ #pragma acc loop vector reduction(+:acc)
+ for (i = 0; i < 32; i++)
+ acc += ina[k * 32 + i] * inb[i * 32 + j];
+
+ out[k * 32 + j] = acc;
+ }
+ }
+
+ for (j = 0; j < 32; j++)
+ for (i = 0; i < 32; i++)
+ assert (out[j * 32 + i] == (i + j) * 2);
+}
+
+
+/* Test of reduction on loop directive (workers, private reduction
+ variable). */
+
+void w_p_1()
+{
+ int i, j, arr[1024], out[32], res = 0, hres = 0;
+
+ for (i = 0; i < 1024; i++)
+ arr[i] = i;
+
+ #pragma acc parallel num_gangs(32) num_workers(32) vector_length(32) \
+ private(res) copyout(out)
+ /* { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 327 } */
+ {
+ #pragma acc loop gang
+ for (j = 0; j < 32; j++)
+ {
+ res = 0;
+
+ #pragma acc loop worker reduction(+:res)
+ for (i = 0; i < 32; i++)
+ res += arr[j * 32 + i];
+
+ out[j] = res;
+ }
+ }
+
+ for (j = 0; j < 32; j++)
+ {
+ hres = 0;
+
+ for (i = 0; i < 32; i++)
+ hres += arr[j * 32 + i];
+
+ assert (out[j] == hres);
+ }
+}
+
+
+/* Test of reduction on loop directive (workers and vectors, private reduction
+ variable). */
+
+void wv_p_1()
+{
+ int i, j, arr[1024], out[32], res = 0, hres = 0;
+
+ for (i = 0; i < 1024; i++)
+ arr[i] = i;
+
+ #pragma acc parallel num_gangs(32) num_workers(32) vector_length(32) \
+ private(res) copyout(out)
+ {
+ #pragma acc loop gang
+ for (j = 0; j < 32; j++)
+ {
+ res = 0;
+
+ #pragma acc loop worker vector reduction(+:res)
+ for (i = 0; i < 32; i++)
+ res += arr[j * 32 + i];
+
+ out[j] = res;
+ }
+ }
+
+ for (j = 0; j < 32; j++)
+ {
+ hres = 0;
+
+ for (i = 0; i < 32; i++)
+ hres += arr[j * 32 + i];
+
+ assert (out[j] == hres);
+ }
+}
+
+
+/* Test of reduction on loop directive (workers and vectors, private reduction
+ variable). */
+
+void wv_p_2()
+{
+ int i, j, arr[32768], out[32], res = 0, hres = 0;
+
+ for (i = 0; i < 32768; i++)
+ arr[i] = i;
+
+ #pragma acc parallel num_gangs(32) num_workers(32) vector_length(32) \
+ private(res) copyout(out)
+ {
+ #pragma acc loop gang
+ for (j = 0; j < 32; j++)
+ {
+ res = j;
+
+ #pragma acc loop worker reduction(+:res)
+ for (i = 0; i < 1024; i++)
+ res += arr[j * 1024 + i];
+
+ #pragma acc loop vector reduction(+:res)
+ for (i = 1023; i >= 0; i--)
+ res += arr[j * 1024 + i];
+
+ out[j] = res;
+ }
+ }
+
+ for (j = 0; j < 32; j++)
+ {
+ hres = j;
+
+ for (i = 0; i < 1024; i++)
+ hres += arr[j * 1024 + i] * 2;
+
+ assert (out[j] == hres);
+ }
+}
+
+
+/* Test of reduction on loop directive (workers and vectors, private reduction
+ variable: gang-redundant mode). */
+
+void wv_p_3()
+{
+ int i, arr[1024], out[32], res = 0, hres = 0;
+
+ for (i = 0; i < 1024; i++)
+ arr[i] = i ^ 33;
+
+ #pragma acc parallel num_gangs(32) num_workers(32) vector_length(32) \
+ private(res) copyin(arr) copyout(out)
+ {
+ /* Private variables aren't initialized by default in openacc. */
+ res = 0;
+
+ /* "res" should be available at the end of the following loop (and should
+ have the same value redundantly in each gang). */
+ #pragma acc loop worker vector reduction(+:res)
+ for (i = 0; i < 1024; i++)
+ res += arr[i];
+
+ #pragma acc loop gang (static: 1)
+ for (i = 0; i < 32; i++)
+ out[i] = res;
+ }
+
+ for (i = 0; i < 1024; i++)
+ hres += arr[i];
+
+ for (i = 0; i < 32; i++)
+ assert (out[i] == hres);
+}
+
+
+int main()
+{
+ g_np_1();
+ gv_np_1();
+ gw_np_1();
+ gwv_np_1();
+ gwv_np_2();
+ gwv_np_3();
+ gwv_np_4();
+ v_p_1();
+ v_p_2();
+ w_p_1();
+ wv_p_1();
+ wv_p_2();
+ wv_p_3();
+
+ return 0;
+}
@@ -1,40 +1,86 @@
-/* FIXME: remove -fno-var-tracking from dg-aditional-options. */
-
-/* { dg-do run { target openacc_nvidia_accel_selected } } */
-/* { dg-additional-options "-fno-inline -fno-var-tracking" } */
-
#include <stdio.h>
#include <stdlib.h>
#pragma acc routine
-int
-fact (int n)
+int fact(int n)
{
if (n == 0 || n == 1)
return 1;
-
- return n * fact (n - 1);
+ else
+ return n * fact (n - 1);
}
-int
-main()
+int main()
{
- int *a, i, n = 10;
+ int *s, *g, *w, *v, *gw, *gv, *wv, *gwv, i, n = 10;
- a = (int *)malloc (sizeof (int) * n);
+ s = (int *) malloc (sizeof (int) * n);
+ g = (int *) malloc (sizeof (int) * n);
+ w = (int *) malloc (sizeof (int) * n);
+ v = (int *) malloc (sizeof (int) * n);
+ gw = (int *) malloc (sizeof (int) * n);
+ gv = (int *) malloc (sizeof (int) * n);
+ wv = (int *) malloc (sizeof (int) * n);
+ gwv = (int *) malloc (sizeof (int) * n);
-#pragma acc parallel copy (a[0:n]) vector_length (32)
- {
-#pragma acc loop vector
- for (i = 0; i < n; i++)
- a[i] = fact (i);
- }
+#pragma acc parallel loop async copyout(s[0:n]) seq
+ for (i = 0; i < n; i++)
+ s[i] = fact (i);
+#pragma acc parallel loop async copyout(g[0:n]) gang
for (i = 0; i < n; i++)
- if (a[i] != fact (i))
- abort ();
+ g[i] = fact (i);
+
+#pragma acc parallel loop async copyout(w[0:n]) worker
+ for (i = 0; i < n; i++)
+ w[i] = fact (i);
+
+#pragma acc parallel loop async copyout(v[0:n]) vector
+ for (i = 0; i < n; i++)
+ v[i] = fact (i);
+
+#pragma acc parallel loop async copyout(gw[0:n]) gang worker
+ for (i = 0; i < n; i++)
+ gw[i] = fact (i);
+
+#pragma acc parallel loop async copyout(gv[0:n]) gang vector
+ for (i = 0; i < n; i++)
+ gv[i] = fact (i);
+
+#pragma acc parallel loop async copyout(wv[0:n]) worker vector
+ for (i = 0; i < n; i++)
+ wv[i] = fact (i);
+
+#pragma acc parallel loop async copyout(gwv[0:n]) gang worker vector
+ for (i = 0; i < n; i++)
+ gwv[i] = fact (i);
- free (a);
+#pragma acc wait
+
+ for (i = 0; i < n; i++)
+ if (s[i] != fact (i))
+ abort ();
+ for (i = 0; i < n; i++)
+ if (g[i] != s[i])
+ abort ();
+ for (i = 0; i < n; i++)
+ if (w[i] != s[i])
+ abort ();
+ for (i = 0; i < n; i++)
+ if (v[i] != s[i])
+ abort ();
+ for (i = 0; i < n; i++)
+ if (gw[i] != s[i])
+ abort ();
+ for (i = 0; i < n; i++)
+ if (gv[i] != s[i])
+ abort ();
+ for (i = 0; i < n; i++)
+ if (wv[i] != s[i])
+ abort ();
+ for (i = 0; i < n; i++)
+ if (gwv[i] != s[i])
+ abort ();
return 0;
}
deleted file mode 100644
@@ -1,42 +0,0 @@
-/* FIXME: remove -fno-var-tracking from dg-additional-options. */
-
-/* { dg-do run { target openacc_nvidia_accel_selected } } */
-/* { dg-additional-options "-fno-inline -fno-var-tracking" } */
-
-#include <stdio.h>
-#include <stdlib.h>
-
-int fact (int);
-
-#pragma acc routine (fact)
-
-int fact (int n)
-{
- if (n == 0 || n == 1)
- return 1;
-
- return n * fact (n - 1);
-}
-
-int
-main()
-{
- int *a, i, n = 10;
-
- a = (int *)malloc (sizeof (int) * n);
-
-#pragma acc parallel copy (a[0:n]) vector_length (32)
- {
-#pragma acc loop vector
- for (i = 0; i < n; i++)
- a[i] = fact (i);
- }
-
- for (i = 0; i < n; i++)
- if (a[i] != fact (i))
- abort ();
-
- free (a);
-
- return 0;
-}
@@ -1,7 +1,5 @@
-/* { dg-do run } */
-/* { dg-additional-options "-w" } */
-
#include <stdlib.h>
+#include <stdio.h>
#define M 8
#define N 32
@@ -38,7 +36,7 @@ gang (int *a)
{
int i;
-#pragma acc loop gang
+#pragma acc loop gang worker vector
for (i = 0; i < N; i++)
a[i] -= i;
}
@@ -53,8 +51,6 @@ seq (int *a)
a[i] += 1;
}
-#include <stdio.h>
-
int
main(int argc, char **argv)
{
@@ -1,4 +1,3 @@
-/* { dg-additional-options "-w" } */
/* This code uses nvptx inline assembly guarded with acc_on_device, which is
not optimized away at -O0, and then confuses the target assembler.
{ dg-skip-if "" { *-*-* } { "-O0" } { "" } } */
@@ -9,6 +8,8 @@
#pragma acc routine gang
void __attribute__ ((noinline)) gang (int ary[N])
+/* { dg-warning "region is worker partitioned but does not contain worker partitioned code" "worker" { target *-*-* } 10 } */
+/* { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 10 } */
{
#pragma acc loop gang
for (unsigned ix = 0; ix < N; ix++)
deleted file mode 100644
@@ -1,47 +0,0 @@
-/* This code uses nvptx inline assembly guarded with acc_on_device, which is
- not optimized away at -O0, and then confuses the target assembler.
- { dg-skip-if "" { *-*-* } { "-O0" } { "" } } */
-
-#include <stdio.h>
-#include <openacc.h>
-
-#define VEC_ID(I, N) \
- (acc_on_device (acc_device_nvidia) \
- ? ({unsigned __r; \
- __asm__ volatile ("mov.u32 %0,%%tid.x;" : "=r" (__r)); \
- __r; }) : (I % N))
-
-#pragma acc routine vector
-void Vec (int *ptr, int lim, int N)
-{
-#pragma acc loop vector
- for (int i = 0; i < lim; i++)
- ptr[i] = VEC_ID(i, N);
-}
-
-#define LEN 32
-
-int main ()
-{
- int ary[LEN];
- int err = 0;
-
- for (int ix = 0; ix != LEN; ix++)
- ary[ix] = 0xdeadbeef;
-
-#pragma acc parallel vector_length(32) copy (ary)
- {
- Vec (ary, LEN, 32);
- }
-
- for (int ix = 0; ix != LEN; ix++)
- {
- if (ary[ix] != ix % 32)
- {
- printf ("ary[%d] = %d expected %d\n", ix, ary[ix], ix % 32);
- err = 1;
- }
- }
-
- return err;
-}
@@ -1,4 +1,3 @@
-/* { dg-additional-options "-w" } */
/* This code uses nvptx inline assembly guarded with acc_on_device, which is
not optimized away at -O0, and then confuses the target assembler.
{ dg-skip-if "" { *-*-* } { "-O0" } { "" } } */
@@ -9,6 +8,7 @@
#pragma acc routine worker
void __attribute__ ((noinline)) worker (int ary[N])
+/* { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 10 } */
{
#pragma acc loop worker
for (unsigned ix = 0; ix < N; ix++)
deleted file mode 100644
@@ -1,55 +0,0 @@
-/* { dg-additional-options "-w" } */
-/* This code uses nvptx inline assembly guarded with acc_on_device, which is
- not optimized away at -O0, and then confuses the target assembler.
- { dg-skip-if "" { *-*-* } { "-O0" } { "" } } */
-
-#include <stdio.h>
-#include <openacc.h>
-
-#define WORK_ID(I,N) \
- (acc_on_device (acc_device_nvidia) \
- ? ({unsigned __r; \
- __asm__ volatile ("mov.u32 %0,%%tid.y;" : "=r" (__r)); \
- __r; }) : (I % N))
-
-#pragma acc routine worker
-void Work (int *ptr, int lim, int N)
-{
-#pragma acc loop worker
- for (int i = 0; i < lim; i++)
- ptr[i] = WORK_ID(i, N);
-}
-
-#define LEN 32
-
-int DoWork (int err, int N)
-{
- int ary[LEN];
-
- for (int ix = 0; ix != LEN; ix++)
- ary[ix] = 0xdeadbeef;
-
-#pragma acc parallel num_workers(N) copy (ary)
- {
- Work (ary, LEN, N);
- }
-
- for (int ix = 0; ix != LEN; ix++)
- if (ary[ix] != ix % N)
- {
- printf ("ary[%d] = %d expected %d\n", ix, ary[ix], ix % N);
- err = 1;
- }
- return err;
-}
-
-
-int main ()
-{
- int err = 0;
-
- for (int W = 1; W <= LEN; W <<= 1)
- err = DoWork (err, W);
-
- return err;
-}
deleted file mode 100644
@@ -1,361 +0,0 @@
-/* Copy of update-1.c with self exchanged with host for #pragma acc update. */
-
-/* { dg-do run } */
-/* { dg-skip-if "" { *-*-* } { "*" } { "-DACC_MEM_SHARED=0" } } */
-
-#include <openacc.h>
-#include <string.h>
-#include <stdio.h>
-#include <stdlib.h>
-#include <stdbool.h>
-
-int
-main (int argc, char **argv)
-{
- int N = 8;
- int NDIV2 = N / 2;
- float *a, *b, *c;
- float *d_a, *d_b, *d_c;
- int i;
-
- a = (float *) malloc (N * sizeof (float));
- b = (float *) malloc (N * sizeof (float));
- c = (float *) malloc (N * sizeof (float));
-
- d_a = (float *) acc_malloc (N * sizeof (float));
- d_b = (float *) acc_malloc (N * sizeof (float));
- d_c = (float *) acc_malloc (N * sizeof (float));
-
- for (i = 0; i < N; i++)
- {
- a[i] = 3.0;
- b[i] = 0.0;
- }
-
- acc_map_data (a, d_a, N * sizeof (float));
- acc_map_data (b, d_b, N * sizeof (float));
- acc_map_data (c, d_c, N * sizeof (float));
-
-#pragma acc update device (a[0:N], b[0:N])
-
-#pragma acc parallel present (a[0:N], b[0:N])
- {
- int ii;
-
- for (ii = 0; ii < N; ii++)
- b[ii] = a[ii];
- }
-
-#pragma acc update self (a[0:N], b[0:N])
-
- for (i = 0; i < N; i++)
- {
- if (a[i] != 3.0)
- abort ();
-
- if (b[i] != 3.0)
- abort ();
- }
-
- if (!acc_is_present (&a[0], (N * sizeof (float))))
- abort ();
-
- if (!acc_is_present (&b[0], (N * sizeof (float))))
- abort ();
-
- for (i = 0; i < N; i++)
- {
- a[i] = 5.0;
- b[i] = 1.0;
- }
-
-#pragma acc update device (a[0:N], b[0:N])
-
-#pragma acc parallel present (a[0:N], b[0:N])
- {
- int ii;
-
- for (ii = 0; ii < N; ii++)
- b[ii] = a[ii];
- }
-
-#pragma acc update self (a[0:N], b[0:N])
-
- for (i = 0; i < N; i++)
- {
- if (a[i] != 5.0)
- abort ();
-
- if (b[i] != 5.0)
- abort ();
- }
-
- if (!acc_is_present (&a[0], (N * sizeof (float))))
- abort ();
-
- if (!acc_is_present (&b[0], (N * sizeof (float))))
- abort ();
-
- for (i = 0; i < N; i++)
- {
- a[i] = 5.0;
- b[i] = 1.0;
- }
-
-#pragma acc update device (a[0:N], b[0:N])
-
-#pragma acc parallel present (a[0:N], b[0:N])
- {
- int ii;
-
- for (ii = 0; ii < N; ii++)
- b[ii] = a[ii];
- }
-
-#pragma acc update host (a[0:N], b[0:N])
-
- for (i = 0; i < N; i++)
- {
- if (a[i] != 5.0)
- abort ();
-
- if (b[i] != 5.0)
- abort ();
- }
-
- if (!acc_is_present (&a[0], (N * sizeof (float))))
- abort ();
-
- if (!acc_is_present (&b[0], (N * sizeof (float))))
- abort ();
-
- for (i = 0; i < N; i++)
- {
- a[i] = 6.0;
- b[i] = 0.0;
- }
-
-#pragma acc update device (a[0:N], b[0:N])
-
- for (i = 0; i < N; i++)
- {
- a[i] = 9.0;
- }
-
-#pragma acc parallel present (a[0:N], b[0:N])
- {
- int ii;
-
- for (ii = 0; ii < N; ii++)
- b[ii] = a[ii];
- }
-
-#pragma acc update self (a[0:N], b[0:N])
-
- for (i = 0; i < N; i++)
- {
- if (a[i] != 6.0)
- abort ();
-
- if (b[i] != 6.0)
- abort ();
- }
-
- if (!acc_is_present (&a[0], (N * sizeof (float))))
- abort ();
-
- if (!acc_is_present (&b[0], (N * sizeof (float))))
- abort ();
-
- for (i = 0; i < N; i++)
- {
- a[i] = 7.0;
- b[i] = 2.0;
- }
-
-#pragma acc update device (a[0:N], b[0:N])
-
- for (i = 0; i < N; i++)
- {
- a[i] = 9.0;
- }
-
-#pragma acc parallel present (a[0:N], b[0:N])
- {
- int ii;
-
- for (ii = 0; ii < N; ii++)
- b[ii] = a[ii];
- }
-
-#pragma acc update self (a[0:N], b[0:N])
-
- for (i = 0; i < N; i++)
- {
- if (a[i] != 7.0)
- abort ();
-
- if (b[i] != 7.0)
- abort ();
- }
-
- for (i = 0; i < N; i++)
- {
- a[i] = 9.0;
- }
-
-#pragma acc update device (a[0:N])
-
-#pragma acc parallel present (a[0:N], b[0:N])
- {
- int ii;
-
- for (ii = 0; ii < N; ii++)
- b[ii] = a[ii];
- }
-
-#pragma acc update self (a[0:N], b[0:N])
-
- for (i = 0; i < N; i++)
- {
- if (a[i] != 9.0)
- abort ();
-
- if (b[i] != 9.0)
- abort ();
- }
-
- if (!acc_is_present (&a[0], (N * sizeof (float))))
- abort ();
-
- if (!acc_is_present (&b[0], (N * sizeof (float))))
- abort ();
-
- for (i = 0; i < N; i++)
- {
- a[i] = 5.0;
- }
-
-#pragma acc update device (a[0:N])
-
- for (i = 0; i < N; i++)
- {
- a[i] = 6.0;
- }
-
-#pragma acc update device (a[0:NDIV2])
-
-#pragma acc parallel present (a[0:N], b[0:N])
- {
- int ii;
-
- for (ii = 0; ii < N; ii++)
- b[ii] = a[ii];
- }
-
-#pragma acc update self (a[0:N], b[0:N])
-
- for (i = 0; i < NDIV2; i++)
- {
- if (a[i] != 6.0)
- abort ();
-
- if (b[i] != 6.0)
- abort ();
- }
-
- for (i = NDIV2; i < N; i++)
- {
- if (a[i] != 5.0)
- abort ();
-
- if (b[i] != 5.0)
- abort ();
- }
-
- if (!acc_is_present (&a[0], (N * sizeof (float))))
- abort ();
-
- if (!acc_is_present (&b[0], (N * sizeof (float))))
- abort ();
-
- for (i = 0; i < N; i++)
- {
- a[i] = 0.0;
- }
-
-#pragma acc update device (a[0:4])
-
-#pragma acc parallel present (a[0:N])
- {
- int ii;
-
- for (ii = 0; ii < N; ii++)
- a[ii] = a[ii] + 1.0;
- }
-
-#pragma acc update self (a[4:4])
-
- for (i = 0; i < NDIV2; i++)
- {
- if (a[i] != 0.0)
- abort ();
- }
-
- for (i = NDIV2; i < N; i++)
- {
- if (a[i] != 6.0)
- abort ();
- }
-
-#pragma acc update self (a[0:4])
-
- for (i = 0; i < NDIV2; i++)
- {
- if (a[i] != 1.0)
- abort ();
- }
-
- for (i = NDIV2; i < N; i++)
- {
- if (a[i] != 6.0)
- abort ();
- }
-
- a[2] = 9;
- a[3] = 9;
- a[4] = 9;
- a[5] = 9;
-
-#pragma acc update device (a[2:4])
-
-#pragma acc parallel present (a[0:N])
- {
- int ii;
-
- for (ii = 0; ii < N; ii++)
- a[ii] = a[ii] + 1.0;
- }
-
-#pragma acc update self (a[2:4])
-
- for (i = 0; i < 2; i++)
- {
- if (a[i] != 1.0)
- abort ();
- }
-
- for (i = 2; i < 6; i++)
- {
- if (a[i] != 10.0)
- abort ();
- }
-
- for (i = 6; i < N; i++)
- {
- if (a[i] != 6.0)
- abort ();
- }
-
- return 0;
-}
deleted file mode 100644
@@ -1,30 +0,0 @@
-#include <assert.h>
-
-/* Test basic vector-partitioned mode transitions. */
-
-int
-main (int argc, char *argv[])
-{
- int n = 0, arr[32], i;
-
- for (i = 0; i < 32; i++)
- arr[i] = 0;
-
- #pragma acc parallel copy(n, arr) num_gangs(1) num_workers(1) \
- vector_length(32)
- {
- int j;
- n++;
- #pragma acc loop vector
- for (j = 0; j < 32; j++)
- arr[j]++;
- n++;
- }
-
- assert (n == 2);
-
- for (i = 0; i < 32; i++)
- assert (arr[i] == 1);
-
- return 0;
-}
deleted file mode 100644
@@ -1,43 +0,0 @@
-#include <assert.h>
-
-/* Test vector-partitioned, gang-partitioned mode. */
-
-int
-main (int argc, char *argv[])
-{
- int n[32], arr[1024], i;
-
- for (i = 0; i < 1024; i++)
- arr[i] = 0;
-
- for (i = 0; i < 32; i++)
- n[i] = 0;
-
- #pragma acc parallel copy(n, arr) num_gangs(32) num_workers(1) \
- vector_length(32)
- {
- int j, k;
-
- #pragma acc loop gang(static:*)
- for (j = 0; j < 32; j++)
- n[j]++;
-
- #pragma acc loop gang
- for (j = 0; j < 32; j++)
- #pragma acc loop vector
- for (k = 0; k < 32; k++)
- arr[j * 32 + k]++;
-
- #pragma acc loop gang(static:*)
- for (j = 0; j < 32; j++)
- n[j]++;
- }
-
- for (i = 0; i < 32; i++)
- assert (n[i] == 2);
-
- for (i = 0; i < 1024; i++)
- assert (arr[i] == 1);
-
- return 0;
-}
deleted file mode 100644
@@ -1,54 +0,0 @@
-#include <assert.h>
-
-/* Test conditional vector-partitioned loops. */
-
-int
-main (int argc, char *argv[])
-{
- int n[32], arr[1024], i;
-
- for (i = 0; i < 1024; i++)
- arr[i] = 0;
-
- for (i = 0; i < 32; i++)
- n[i] = 0;
-
- #pragma acc parallel copy(n, arr) num_gangs(32) num_workers(1) \
- vector_length(32)
- {
- int j, k;
-
- #pragma acc loop gang(static:*)
- for (j = 0; j < 32; j++)
- n[j]++;
-
- #pragma acc loop gang
- for (j = 0; j < 32; j++)
- {
- if ((j % 2) == 0)
- {
- #pragma acc loop vector
- for (k = 0; k < 32; k++)
- arr[j * 32 + k]++;
- }
- else
- {
- #pragma acc loop vector
- for (k = 0; k < 32; k++)
- arr[j * 32 + k]--;
- }
- }
-
- #pragma acc loop gang(static:*)
- for (j = 0; j < 32; j++)
- n[j]++;
- }
-
- for (i = 0; i < 32; i++)
- assert (n[i] == 2);
-
- for (i = 0; i < 1024; i++)
- assert (arr[i] == ((i % 64) < 32) ? 1 : -1);
-
- return 0;
-}
deleted file mode 100644
@@ -1,46 +0,0 @@
-#include <assert.h>
-
-/* Test conditions inside vector-partitioned loops. */
-
-int
-main (int argc, char *argv[])
-{
- int n[32], arr[1024], i;
-
- for (i = 0; i < 1024; i++)
- arr[i] = i;
-
- for (i = 0; i < 32; i++)
- n[i] = 0;
-
- #pragma acc parallel copy(n, arr) num_gangs(32) num_workers(1) \
- vector_length(32)
- {
- int j, k;
-
- #pragma acc loop gang(static:*)
- for (j = 0; j < 32; j++)
- n[j]++;
-
- #pragma acc loop gang
- for (j = 0; j < 32; j++)
- {
- #pragma acc loop vector
- for (k = 0; k < 32; k++)
- if ((arr[j * 32 + k] % 2) != 0)
- arr[j * 32 + k] *= 2;
- }
-
- #pragma acc loop gang(static:*)
- for (j = 0; j < 32; j++)
- n[j]++;
- }
-
- for (i = 0; i < 32; i++)
- assert (n[i] == 2);
-
- for (i = 0; i < 1024; i++)
- assert (arr[i] == ((i % 2) == 0 ? i : i * 2));
-
- return 0;
-}
deleted file mode 100644
@@ -1,42 +0,0 @@
-#include <assert.h>
-
-/* Test conditions inside gang-partitioned/vector-partitioned loops. */
-
-int
-main (int argc, char *argv[])
-{
- int n[32], arr[1024], i;
-
- for (i = 0; i < 1024; i++)
- arr[i] = i;
-
- for (i = 0; i < 32; i++)
- n[i] = 0;
-
- #pragma acc parallel copy(n, arr) num_gangs(32) num_workers(1) \
- vector_length(32)
- {
- int j, k;
-
- #pragma acc loop gang(static:*)
- for (j = 0; j < 32; j++)
- n[j]++;
-
- #pragma acc loop gang vector
- for (j = 0; j < 1024; j++)
- if ((arr[j] % 2) != 0)
- arr[j] *= 2;
-
- #pragma acc loop gang(static:*)
- for (j = 0; j < 32; j++)
- n[j]++;
- }
-
- for (i = 0; i < 32; i++)
- assert (n[i] == 2);
-
- for (i = 0; i < 1024; i++)
- assert (arr[i] == ((i % 2) == 0 ? i : i * 2));
-
- return 0;
-}
deleted file mode 100644
@@ -1,77 +0,0 @@
-#include <assert.h>
-#include <stdlib.h>
-
-/* Test switch containing vector-partitioned loops inside gang-partitioned
- loops. */
-
-int
-main (int argc, char *argv[])
-{
- int n[32], arr[1024], i;
-
- for (i = 0; i < 1024; i++)
- arr[i] = 0;
-
- for (i = 0; i < 32; i++)
- n[i] = i % 5;
-
- #pragma acc parallel copy(n, arr) num_gangs(32) num_workers(1) \
- vector_length(32)
- {
- int j, k;
-
- #pragma acc loop gang(static:*)
- for (j = 0; j < 32; j++)
- n[j]++;
-
- #pragma acc loop gang(static:*)
- for (j = 0; j < 32; j++)
- switch (n[j])
- {
- case 1:
- #pragma acc loop vector
- for (k = 0; k < 32; k++)
- arr[j * 32 + k] += 1;
- break;
-
- case 2:
- #pragma acc loop vector
- for (k = 0; k < 32; k++)
- arr[j * 32 + k] += 2;
- break;
-
- case 3:
- #pragma acc loop vector
- for (k = 0; k < 32; k++)
- arr[j * 32 + k] += 3;
- break;
-
- case 4:
- #pragma acc loop vector
- for (k = 0; k < 32; k++)
- arr[j * 32 + k] += 4;
- break;
-
- case 5:
- #pragma acc loop vector
- for (k = 0; k < 32; k++)
- arr[j * 32 + k] += 5;
- break;
-
- default:
- abort ();
- }
-
- #pragma acc loop gang(static:*)
- for (j = 0; j < 32; j++)
- n[j]++;
- }
-
- for (i = 0; i < 32; i++)
- assert (n[i] == (i % 5) + 2);
-
- for (i = 0; i < 1024; i++)
- assert (arr[i] == ((i / 32) % 5) + 1);
-
- return 0;
-}
deleted file mode 100644
@@ -1,17 +0,0 @@
-/* { dg-additional-options "-w" } */
-
-#include <assert.h>
-
-/* Test trivial operation of vector-single mode. */
-
-int
-main (int argc, char *argv[])
-{
- int n = 0;
- #pragma acc parallel copy(n) num_gangs(1) num_workers(1) vector_length(32)
- {
- n++;
- }
- assert (n == 1);
- return 0;
-}
deleted file mode 100644
@@ -1,34 +0,0 @@
-/* { dg-additional-options "-w" } */
-
-#include <assert.h>
-
-/* Test vector-single, gang-partitioned mode. */
-
-int
-main (int argc, char *argv[])
-{
- int arr[1024];
- int gangs;
-
- for (gangs = 1; gangs <= 1024; gangs <<= 1)
- {
- int i;
-
- for (i = 0; i < 1024; i++)
- arr[i] = 0;
-
- #pragma acc parallel copy(arr) num_gangs(gangs) num_workers(1) \
- vector_length(32)
- {
- int j;
- #pragma acc loop gang
- for (j = 0; j < 1024; j++)
- arr[j]++;
- }
-
- for (i = 0; i < 1024; i++)
- assert (arr[i] == 1);
- }
-
- return 0;
-}
deleted file mode 100644
@@ -1,37 +0,0 @@
-/* { dg-additional-options "-w" } */
-
-#include <assert.h>
-
-/* Test conditions in vector-single mode. */
-
-int
-main (int argc, char *argv[])
-{
- int arr[1024];
- int gangs;
-
- for (gangs = 1; gangs <= 1024; gangs <<= 1)
- {
- int i;
-
- for (i = 0; i < 1024; i++)
- arr[i] = 0;
-
- #pragma acc parallel copy(arr) num_gangs(gangs) num_workers(1) \
- vector_length(32)
- {
- int j;
- #pragma acc loop gang
- for (j = 0; j < 1024; j++)
- if ((j % 3) == 0)
- arr[j]++;
- else
- arr[j] += 2;
- }
-
- for (i = 0; i < 1024; i++)
- assert (arr[i] == ((i % 3) == 0) ? 1 : 2);
- }
-
- return 0;
-}
deleted file mode 100644
@@ -1,42 +0,0 @@
-/* { dg-additional-options "-w" } */
-
-#include <assert.h>
-
-/* Test switch in vector-single mode. */
-
-int
-main (int argc, char *argv[])
-{
- int arr[1024];
- int gangs;
-
- for (gangs = 1; gangs <= 1024; gangs <<= 1)
- {
- int i;
-
- for (i = 0; i < 1024; i++)
- arr[i] = 0;
-
- #pragma acc parallel copy(arr) num_gangs(gangs) num_workers(1) \
- vector_length(32)
- {
- int j;
- #pragma acc loop gang
- for (j = 0; j < 1024; j++)
- switch (j % 5)
- {
- case 0: arr[j] += 1; break;
- case 1: arr[j] += 2; break;
- case 2: arr[j] += 3; break;
- case 3: arr[j] += 4; break;
- case 4: arr[j] += 5; break;
- default: arr[j] += 99;
- }
- }
-
- for (i = 0; i < 1024; i++)
- assert (arr[i] == (i % 5) + 1);
- }
-
- return 0;
-}
deleted file mode 100644
@@ -1,45 +0,0 @@
-/* { dg-additional-options "-w" } */
-
-#include <assert.h>
-
-/* Test switch in vector-single mode, initialise array on device. */
-
-int
-main (int argc, char *argv[])
-{
- int arr[1024];
- int i;
-
- for (i = 0; i < 1024; i++)
- arr[i] = 99;
-
- #pragma acc parallel copy(arr) num_gangs(1024) num_workers(1) \
- vector_length(32)
- {
- int j;
-
- /* This loop and the one following must be distributed to available gangs
- in the same way to ensure data dependencies are not violated (hence the
- "static" clauses). */
- #pragma acc loop gang(static:*)
- for (j = 0; j < 1024; j++)
- arr[j] = 0;
-
- #pragma acc loop gang(static:*)
- for (j = 0; j < 1024; j++)
- switch (j % 5)
- {
- case 0: arr[j] += 1; break;
- case 1: arr[j] += 2; break;
- case 2: arr[j] += 3; break;
- case 3: arr[j] += 4; break;
- case 4: arr[j] += 5; break;
- default: arr[j] += 99;
- }
- }
-
- for (i = 0; i < 1024; i++)
- assert (arr[i] == (i % 5) + 1);
-
- return 0;
-}
deleted file mode 100644
@@ -1,51 +0,0 @@
-/* { dg-additional-options "-w" } */
-
-#include <assert.h>
-#include <stdbool.h>
-
-#define NUM_GANGS 4096
-
-/* Test multiple conditions in vector-single mode. */
-
-int
-main (int argc, char *argv[])
-{
- bool fizz[NUM_GANGS], buzz[NUM_GANGS], fizzbuzz[NUM_GANGS];
- int i;
-
- #pragma acc parallel copyout(fizz, buzz, fizzbuzz) \
- num_gangs(NUM_GANGS) num_workers(1) vector_length(32)
- {
- int j;
-
- /* This loop and the one following must be distributed to available gangs
- in the same way to ensure data dependencies are not violated (hence the
- "static" clauses). */
- #pragma acc loop gang(static:*)
- for (j = 0; j < NUM_GANGS; j++)
- fizz[j] = buzz[j] = fizzbuzz[j] = 0;
-
- #pragma acc loop gang(static:*)
- for (j = 0; j < NUM_GANGS; j++)
- {
- if ((j % 3) == 0 && (j % 5) == 0)
- fizzbuzz[j] = 1;
- else
- {
- if ((j % 3) == 0)
- fizz[j] = 1;
- else if ((j % 5) == 0)
- buzz[j] = 1;
- }
- }
- }
-
- for (i = 0; i < NUM_GANGS; i++)
- {
- assert (fizzbuzz[i] == ((i % 3) == 0 && (i % 5) == 0));
- assert (fizz[i] == ((i % 3) == 0 && (i % 5) != 0));
- assert (buzz[i] == ((i % 3) != 0 && (i % 5) == 0));
- }
-
- return 0;
-}
deleted file mode 100644
@@ -1,38 +0,0 @@
-/* Check if worker-single variables get broadcastd to vectors. */
-
-/* { dg-do run } */
-
-#include <assert.h>
-#include <math.h>
-
-#define N 32
-
-#pragma acc routine
-float
-some_val ()
-{
- return 2.71;
-}
-
-int
-main ()
-{
- float threads[N], v1 = 3.14;
-
- for (int i = 0; i < N; i++)
- threads[i] = -1;
-
-#pragma acc parallel num_gangs (1) vector_length (32) copy (v1)
- {
- float val = some_val ();
-
-#pragma acc loop vector
- for (int i = 0; i < N; i++)
- threads[i] = val + v1*i;
- }
-
- for (int i = 0; i < N; i++)
- assert (fabs (threads[i] - (some_val () + v1*i)) < 0.0001);
-
- return 0;
-}
deleted file mode 100644
@@ -1,32 +0,0 @@
-/* { dg-additional-options "-w" } */
-
-#include <assert.h>
-
-/* Test worker-partitioned/vector-single mode. */
-
-int
-main (int argc, char *argv[])
-{
- int arr[32 * 8], i;
-
- for (i = 0; i < 32 * 8; i++)
- arr[i] = 0;
-
- #pragma acc parallel copy(arr) num_gangs(8) num_workers(8) vector_length(32)
- {
- int j;
- #pragma acc loop gang
- for (j = 0; j < 32; j++)
- {
- int k;
- #pragma acc loop worker
- for (k = 0; k < 8; k++)
- arr[j * 8 + k] += j * 8 + k;
- }
- }
-
- for (i = 0; i < 32 * 8; i++)
- assert (arr[i] == i);
-
- return 0;
-}
deleted file mode 100644
@@ -1,44 +0,0 @@
-#include <assert.h>
-
-/* Test condition in worker-partitioned mode. */
-
-int
-main (int argc, char *argv[])
-{
- int arr[32 * 32 * 8], i;
-
- for (i = 0; i < 32 * 32 * 8; i++)
- arr[i] = i;
-
- #pragma acc parallel copy(arr) num_gangs(8) num_workers(8) vector_length(32)
- {
- int j;
- #pragma acc loop gang
- for (j = 0; j < 32; j++)
- {
- int k;
- #pragma acc loop worker
- for (k = 0; k < 8; k++)
- {
- int m;
- if ((k % 2) == 0)
- {
- #pragma acc loop vector
- for (m = 0; m < 32; m++)
- arr[j * 32 * 8 + k * 32 + m]++;
- }
- else
- {
- #pragma acc loop vector
- for (m = 0; m < 32; m++)
- arr[j * 32 * 8 + k * 32 + m] += 2;
- }
- }
- }
- }
-
- for (i = 0; i < 32 * 32 * 8; i++)
- assert (arr[i] == i + ((i / 32) % 2) + 1);
-
- return 0;
-}
deleted file mode 100644
@@ -1,54 +0,0 @@
-#include <assert.h>
-
-/* Test switch in worker-partitioned mode. */
-
-int
-main (int argc, char *argv[])
-{
- int arr[32 * 32 * 8], i;
-
- for (i = 0; i < 32 * 32 * 8; i++)
- arr[i] = i;
-
- #pragma acc parallel copy(arr) num_gangs(8) num_workers(8) vector_length(32)
- {
- int j;
- #pragma acc loop gang
- for (j = 0; j < 32; j++)
- {
- int k;
- #pragma acc loop worker
- for (k = 0; k < 8; k++)
- {
- int m;
- switch ((j * 32 + k) % 3)
- {
- case 0:
- #pragma acc loop vector
- for (m = 0; m < 32; m++)
- arr[j * 32 * 8 + k * 32 + m]++;
- break;
-
- case 1:
- #pragma acc loop vector
- for (m = 0; m < 32; m++)
- arr[j * 32 * 8 + k * 32 + m] += 2;
- break;
-
- case 2:
- #pragma acc loop vector
- for (m = 0; m < 32; m++)
- arr[j * 32 * 8 + k * 32 + m] += 3;
- break;
-
- default: ;
- }
- }
- }
- }
-
- for (i = 0; i < 32 * 32 * 8; i++)
- assert (arr[i] == i + ((i / 32) % 3) + 1);
-
- return 0;
-}
deleted file mode 100644
@@ -1,56 +0,0 @@
-/* { dg-additional-options "-w" } */
-
-#include <assert.h>
-
-/* Test worker-single/worker-partitioned transitions. */
-
-int
-main (int argc, char *argv[])
-{
- int n[32], arr[32 * 32], i;
-
- for (i = 0; i < 32 * 32; i++)
- arr[i] = 0;
-
- for (i = 0; i < 32; i++)
- n[i] = 0;
-
- #pragma acc parallel copy(n, arr) num_gangs(8) num_workers(16) \
- vector_length(32)
- {
- int j;
- #pragma acc loop gang
- for (j = 0; j < 32; j++)
- {
- int k;
-
- n[j]++;
-
- #pragma acc loop worker
- for (k = 0; k < 32; k++)
- arr[j * 32 + k]++;
-
- n[j]++;
-
- #pragma acc loop worker
- for (k = 0; k < 32; k++)
- arr[j * 32 + k]++;
-
- n[j]++;
-
- #pragma acc loop worker
- for (k = 0; k < 32; k++)
- arr[j * 32 + k]++;
-
- n[j]++;
- }
- }
-
- for (i = 0; i < 32; i++)
- assert (n[i] == 4);
-
- for (i = 0; i < 32 * 32; i++)
- assert (arr[i] == 3);
-
- return 0;
-}
deleted file mode 100644
@@ -1,47 +0,0 @@
-/* { dg-additional-options "-w" } */
-
-#include <assert.h>
-
-/* Test correct synchronisation between worker-partitioned loops. */
-
-int
-main (int argc, char *argv[])
-{
- int arr_a[32 * 32], arr_b[32 * 32], i;
- int num_workers, num_gangs;
-
- for (num_workers = 1; num_workers <= 32; num_workers <<= 1)
- for (num_gangs = 1; num_gangs <= 32; num_gangs <<= 1)
- {
- for (i = 0; i < 32 * 32; i++)
- arr_a[i] = i;
-
- #pragma acc parallel copyin(arr_a) copyout(arr_b) num_gangs(num_gangs) \
- num_workers(num_workers) vector_length(32)
- {
- int j;
- #pragma acc loop gang
- for (j = 0; j < 32; j++)
- {
- int k;
-
- #pragma acc loop worker
- for (k = 0; k < 32; k++)
- arr_b[j * 32 + (31 - k)] = arr_a[j * 32 + k] * 2;
-
- #pragma acc loop worker
- for (k = 0; k < 32; k++)
- arr_a[j * 32 + (31 - k)] = arr_b[j * 32 + k] * 2;
-
- #pragma acc loop worker
- for (k = 0; k < 32; k++)
- arr_b[j * 32 + (31 - k)] = arr_a[j * 32 + k] * 2;
- }
- }
-
- for (i = 0; i < 32 * 32; i++)
- assert (arr_b[i] == (i ^ 31) * 8);
- }
-
- return 0;
-}
deleted file mode 100644
@@ -1,45 +0,0 @@
-#include <assert.h>
-
-/* Test correct synchronisation between worker+vector-partitioned loops. */
-
-int
-main (int argc, char *argv[])
-{
- int arr_a[32 * 32 * 32], arr_b[32 * 32 * 32], i;
- int num_workers, num_gangs;
-
- for (num_workers = 1; num_workers <= 32; num_workers <<= 1)
- for (num_gangs = 1; num_gangs <= 32; num_gangs <<= 1)
- {
- for (i = 0; i < 32 * 32 * 32; i++)
- arr_a[i] = i;
-
- #pragma acc parallel copyin(arr_a) copyout(arr_b) num_gangs(num_gangs) \
- num_workers(num_workers) vector_length(32)
- {
- int j;
- #pragma acc loop gang
- for (j = 0; j < 32; j++)
- {
- int k;
-
- #pragma acc loop worker vector
- for (k = 0; k < 32 * 32; k++)
- arr_b[j * 32 * 32 + (1023 - k)] = arr_a[j * 32 * 32 + k] * 2;
-
- #pragma acc loop worker vector
- for (k = 0; k < 32 * 32; k++)
- arr_a[j * 32 * 32 + (1023 - k)] = arr_b[j * 32 * 32 + k] * 2;
-
- #pragma acc loop worker vector
- for (k = 0; k < 32 * 32; k++)
- arr_b[j * 32 * 32 + (1023 - k)] = arr_a[j * 32 * 32 + k] * 2;
- }
- }
-
- for (i = 0; i < 32 * 32 * 32; i++)
- assert (arr_b[i] == (i ^ 1023) * 8);
- }
-
- return 0;
-}
deleted file mode 100644
@@ -1,90 +0,0 @@
-#include <assert.h>
-
-/* Test correct synchronisation between vector-partitioned loops in
- worker-partitioned mode. */
-
-int
-main (int argc, char *argv[])
-{
- int n[32 * 32], arr_a[32 * 32 * 32], arr_b[32 * 32 * 32], i;
- int num_workers, num_gangs;
-
- for (num_workers = 1; num_workers <= 32; num_workers <<= 1)
- for (num_gangs = 1; num_gangs <= 32; num_gangs <<= 1)
- {
- for (i = 0; i < 32 * 32 * 32; i++)
- arr_a[i] = i;
-
- for (i = 0; i < 32 * 32; i++)
- n[i] = 0;
-
- #pragma acc parallel copy (n) copyin(arr_a) copyout(arr_b) \
- num_gangs(num_gangs) num_workers(num_workers) vector_length(32)
- {
- int j;
- #pragma acc loop gang
- for (j = 0; j < 32; j++)
- {
- int k;
-
- #pragma acc loop worker
- for (k = 0; k < 32; k++)
- {
- int m;
-
- n[j * 32 + k]++;
-
- #pragma acc loop vector
- for (m = 0; m < 32; m++)
- {
- if (((j * 1024 + k * 32 + m) % 2) == 0)
- arr_b[j * 1024 + k * 32 + (31 - m)]
- = arr_a[j * 1024 + k * 32 + m] * 2;
- else
- arr_b[j * 1024 + k * 32 + (31 - m)]
- = arr_a[j * 1024 + k * 32 + m] * 3;
- }
-
- /* Test returning to vector-single mode... */
- n[j * 32 + k]++;
-
- #pragma acc loop vector
- for (m = 0; m < 32; m++)
- {
- if (((j * 1024 + k * 32 + m) % 3) == 0)
- arr_a[j * 1024 + k * 32 + (31 - m)]
- = arr_b[j * 1024 + k * 32 + m] * 5;
- else
- arr_a[j * 1024 + k * 32 + (31 - m)]
- = arr_b[j * 1024 + k * 32 + m] * 7;
- }
-
- /* ...and back-to-back vector loops. */
-
- #pragma acc loop vector
- for (m = 0; m < 32; m++)
- {
- if (((j * 1024 + k * 32 + m) % 2) == 0)
- arr_b[j * 1024 + k * 32 + (31 - m)]
- = arr_a[j * 1024 + k * 32 + m] * 3;
- else
- arr_b[j * 1024 + k * 32 + (31 - m)]
- = arr_a[j * 1024 + k * 32 + m] * 2;
- }
- }
- }
- }
-
- for (i = 0; i < 32 * 32; i++)
- assert (n[i] == 2);
-
- for (i = 0; i < 32 * 32 * 32; i++)
- {
- int m = 6 * ((i % 3) == 0 ? 5 : 7);
- assert (arr_b[i] == (i ^ 31) * m);
- }
- }
-
-
- return 0;
-}
deleted file mode 100644
@@ -1,51 +0,0 @@
-/* With -O0, variables are on the stack, not in registers. Check that worker
- state propagation handles the stack frame. */
-
-int
-main (int argc, char *argv[])
-{
- int w0 = 0;
- int w1 = 0;
- int w2 = 0;
- int w3 = 0;
- int w4 = 0;
- int w5 = 0;
- int w6 = 0;
- int w7 = 0;
-
- int i;
-
-#pragma acc parallel num_gangs (1) num_workers (8) copy (w0, w1, w2, w3, w4, w5, w6, w7)
- {
- int internal = 100;
-
-#pragma acc loop worker
- for (i = 0; i < 8; i++)
- {
- switch (i)
- {
- case 0: w0 = internal; break;
- case 1: w1 = internal; break;
- case 2: w2 = internal; break;
- case 3: w3 = internal; break;
- case 4: w4 = internal; break;
- case 5: w5 = internal; break;
- case 6: w6 = internal; break;
- case 7: w7 = internal; break;
- default: break;
- }
- }
- }
-
- if (w0 != 100
- || w1 != 100
- || w2 != 100
- || w3 != 100
- || w4 != 100
- || w5 != 100
- || w6 != 100
- || w7 != 100)
- __builtin_abort ();
-
- return 0;
-}
deleted file mode 100644
@@ -1,27 +0,0 @@
-/* { dg-additional-options "-w" } */
-
-#include <assert.h>
-
-/* Test worker-single/vector-single mode. */
-
-int
-main (int argc, char *argv[])
-{
- int arr[32], i;
-
- for (i = 0; i < 32; i++)
- arr[i] = 0;
-
- #pragma acc parallel copy(arr) num_gangs(8) num_workers(8) vector_length(32)
- {
- int j;
- #pragma acc loop gang
- for (j = 0; j < 32; j++)
- arr[j]++;
- }
-
- for (i = 0; i < 32; i++)
- assert (arr[i] == 1);
-
- return 0;
-}
deleted file mode 100644
@@ -1,30 +0,0 @@
-/* { dg-additional-options "-w" } */
-
-#include <assert.h>
-
-/* Test worker-single/vector-single mode. */
-
-int
-main (int argc, char *argv[])
-{
- int arr[32], i;
-
- for (i = 0; i < 32; i++)
- arr[i] = 0;
-
- #pragma acc parallel copy(arr) num_gangs(8) num_workers(8) vector_length(32)
- {
- int j;
- #pragma acc loop gang
- for (j = 0; j < 32; j++)
- {
- #pragma acc atomic
- arr[j]++;
- }
- }
-
- for (i = 0; i < 32; i++)
- assert (arr[i] == 1);
-
- return 0;
-}
deleted file mode 100644
@@ -1,30 +0,0 @@
-/* { dg-additional-options "-w" } */
-
-#include <assert.h>
-
-/* Test condition in worker-single/vector-single mode. */
-
-int
-main (int argc, char *argv[])
-{
- int arr[32], i;
-
- for (i = 0; i < 32; i++)
- arr[i] = i;
-
- #pragma acc parallel copy(arr) num_gangs(8) num_workers(8) vector_length(32)
- {
- int j;
- #pragma acc loop gang
- for (j = 0; j < 32; j++)
- if ((arr[j] % 2) != 0)
- arr[j]++;
- else
- arr[j] += 2;
- }
-
- for (i = 0; i < 32; i++)
- assert (arr[i] == ((i % 2) != 0) ? i + 1 : i + 2);
-
- return 0;
-}
deleted file mode 100644
@@ -1,35 +0,0 @@
-/* { dg-additional-options "-w" } */
-
-#include <assert.h>
-
-/* Test switch in worker-single/vector-single mode. */
-
-int
-main (int argc, char *argv[])
-{
- int arr[32], i;
-
- for (i = 0; i < 32; i++)
- arr[i] = i;
-
- #pragma acc parallel copy(arr) num_gangs(8) num_workers(8) vector_length(32)
- {
- int j;
- #pragma acc loop gang
- for (j = 0; j < 32; j++)
- switch (arr[j] % 5)
- {
- case 0: arr[j] += 1; break;
- case 1: arr[j] += 2; break;
- case 2: arr[j] += 3; break;
- case 3: arr[j] += 4; break;
- case 4: arr[j] += 5; break;
- default: arr[j] += 99;
- }
- }
-
- for (i = 0; i < 32; i++)
- assert (arr[i] == i + (i % 5) + 1);
-
- return 0;
-}
deleted file mode 100644
@@ -1,35 +0,0 @@
-/* { dg-additional-options "-w" } */
-
-#include <assert.h>
-
-/* Test worker-single/vector-partitioned mode. */
-
-int
-main (int argc, char *argv[])
-{
- int arr[32 * 32], i;
-
- for (i = 0; i < 32 * 32; i++)
- arr[i] = i;
-
- #pragma acc parallel copy(arr) num_gangs(8) num_workers(8) vector_length(32)
- {
- int j;
- #pragma acc loop gang
- for (j = 0; j < 32; j++)
- {
- int k;
- #pragma acc loop vector
- for (k = 0; k < 32; k++)
- {
- #pragma acc atomic
- arr[j * 32 + k]++;
- }
- }
- }
-
- for (i = 0; i < 32 * 32; i++)
- assert (arr[i] == i + 1);
-
- return 0;
-}
deleted file mode 100644
@@ -1,51 +0,0 @@
-/* { dg-additional-options "-w" } */
-
-#include <assert.h>
-
-/* Test multiple conditional vector-partitioned loops in worker-single
- mode. */
-
-int
-main (int argc, char *argv[])
-{
- int arr[32 * 32], i;
-
- for (i = 0; i < 32 * 32; i++)
- arr[i] = i;
-
- #pragma acc parallel copy(arr) num_gangs(8) num_workers(8) vector_length(32)
- {
- int j;
- #pragma acc loop gang
- for (j = 0; j < 32; j++)
- {
- int k;
- if ((j % 3) == 0)
- {
- #pragma acc loop vector
- for (k = 0; k < 32; k++)
- {
- #pragma acc atomic
- arr[j * 32 + k] += 3;
- }
- }
- else if ((j % 3) == 1)
- {
- #pragma acc loop vector
- for (k = 0; k < 32; k++)
- {
- #pragma acc atomic
- arr[j * 32 + k] += 7;
- }
- }
- }
- }
-
- for (i = 0; i < 32 * 32; i++)
- {
- int j = (i / 32) % 3;
- assert (arr[i] == i + ((j == 0) ? 3 : (j == 1) ? 7 : 0));
- }
-
- return 0;
-}
deleted file mode 100644
@@ -1,50 +0,0 @@
-/* { dg-additional-options "-w" } */
-
-#include <assert.h>
-#include <openacc.h>
-
-#define ACTUAL_GANGS 8
-
-/* Test worker-single, vector-partitioned, gang-redundant mode. */
-
-int
-main (int argc, char *argv[])
-{
- int n, arr[32], i;
- int ondev;
-
- for (i = 0; i < 32; i++)
- arr[i] = 0;
-
- n = 0;
-
- #pragma acc parallel copy(n, arr) num_gangs(ACTUAL_GANGS) num_workers(8) \
- vector_length(32) copyout(ondev)
- {
- int j;
-
- ondev = acc_on_device (acc_device_not_host);
-
- #pragma acc atomic
- n++;
-
- #pragma acc loop vector
- for (j = 0; j < 32; j++)
- {
- #pragma acc atomic
- arr[j] += 1;
- }
-
- #pragma acc atomic
- n++;
- }
-
- int m = ondev ? ACTUAL_GANGS : 1;
-
- assert (n == m * 2);
-
- for (i = 0; i < 32; i++)
- assert (arr[i] == m);
-
- return 0;
-}
@@ -15,7 +15,7 @@ program collapse5
v4 = 4
v5 = 13
v6 = 18
- !$acc parallel copy (l)
+ !$acc parallel
!$acc loop collapse (3) reduction (.or.:l)
do i = v1, v2
do j = v3, v4
@@ -18,7 +18,7 @@ program collapse6
v7 = 1
v8 = 1
v9 = 1
- !$acc parallel copy (l)
+ !$acc parallel
!$acc loop collapse (3) reduction (.or.:l)
do i = v1, v2, v7
do j = v3, v4, v8
@@ -8,7 +8,7 @@ program collapse7
r = .false.
a(:, :, :) = 0
b(:, :, :) = 0
- !$acc parallel copy (l)
+ !$acc parallel
!$acc loop collapse (3) reduction (.or.:l)
do i = 2, 6
do j = -2, 4
@@ -15,7 +15,7 @@ program collapse8
v4 = 4
v5 = 13
v6 = 18
- !$acc parallel copy (l)
+ !$acc parallel
!$acc loop collapse (3) reduction (.or.:l)
do i = v1, v2
do j = v3, v4
new file mode 100644
@@ -0,0 +1,552 @@
+! Miscellaneous tests for private variables.
+
+! { dg-do run }
+
+
+! Test of gang-private variables declared on loop directive.
+
+subroutine t1()
+ integer :: x, i, arr(32)
+
+ do i = 1, 32
+ arr(i) = i
+ end do
+
+ !$acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
+ ! { dg-warning "region is worker partitioned but does not contain worker partitioned code" "worker" { target *-*-* } 15 }
+ ! { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 15 }
+ !$acc loop gang private(x)
+ do i = 1, 32
+ x = i * 2;
+ arr(i) = arr(i) + x
+ end do
+ !$acc end parallel
+
+ do i = 1, 32
+ if (arr(i) .ne. i * 3) call abort
+ end do
+end subroutine t1
+
+
+! Test of gang-private variables declared on loop directive, with broadcasting
+! to partitioned workers.
+
+subroutine t2()
+ integer :: x, i, j, arr(0:32*32)
+
+ do i = 0, 32*32-1
+ arr(i) = i
+ end do
+
+ !$acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
+ ! { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 41 }
+ !$acc loop gang private(x)
+ do i = 0, 31
+ x = i * 2;
+
+ !$acc loop worker
+ do j = 0, 31
+ arr(i * 32 + j) = arr(i * 32 + j) + x
+ end do
+ end do
+ !$acc end parallel
+
+ do i = 0, 32 * 32 - 1
+ if (arr(i) .ne. i + (i / 32) * 2) call abort
+ end do
+end subroutine t2
+
+
+! Test of gang-private variables declared on loop directive, with broadcasting
+! to partitioned vectors.
+
+subroutine t3()
+ integer :: x, i, j, arr(0:32*32)
+
+ do i = 0, 32*32-1
+ arr(i) = i
+ end do
+
+ !$acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
+ ! { dg-warning "region is worker partitioned but does not contain worker partitioned code" "worker" { target *-*-* } 70 }
+ !$acc loop gang private(x)
+ do i = 0, 31
+ x = i * 2;
+
+ !$acc loop vector
+ do j = 0, 31
+ arr(i * 32 + j) = arr(i * 32 + j) + x
+ end do
+ end do
+ !$acc end parallel
+
+ do i = 0, 32 * 32 - 1
+ if (arr(i) .ne. i + (i / 32) * 2) call abort
+ end do
+end subroutine t3
+
+
+! Test of gang-private addressable variable declared on loop directive, with
+! broadcasting to partitioned workers.
+
+subroutine t4()
+ type vec3
+ integer x, y, z, attr(13)
+ end type vec3
+
+ integer i, j, arr(0:32*32)
+ type(vec3) pt
+
+ do i = 0, 32*32-1
+ arr(i) = i
+ end do
+
+ !$acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
+ ! { dg-warning "region is worker partitioned but does not contain worker partitioned code" "worker" { target *-*-* } 104 }
+ !$acc loop gang private(pt)
+ do i = 0, 31
+ pt%x = i
+ pt%y = i * 2
+ pt%z = i * 4
+ pt%attr(5) = i * 6
+
+ !$acc loop vector
+ do j = 0, 31
+ arr(i * 32 + j) = arr(i * 32 + j) + pt%x + pt%y + pt%z + pt%attr(5);
+ end do
+ end do
+ !$acc end parallel
+
+ do i = 0, 32 * 32 - 1
+ if (arr(i) .ne. i + (i / 32) * 13) call abort
+ end do
+end subroutine t4
+
+
+! Test of vector-private variables declared on loop directive.
+
+subroutine t5()
+ integer :: x, i, j, k, idx, arr(0:32*32*32)
+
+ do i = 0, 32*32*32-1
+ arr(i) = i
+ end do
+
+ !$acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
+ !$acc loop gang
+ do i = 0, 31
+ !$acc loop worker
+ do j = 0, 31
+ !$acc loop vector private(x)
+ do k = 0, 31
+ x = ieor(i, j * 3)
+ arr(i * 1024 + j * 32 + k) = arr(i * 1024 + j * 32 + k) + x * k
+ end do
+ !$acc loop vector private(x)
+ do k = 0, 31
+ x = ior(i, j * 5)
+ arr(i * 1024 + j * 32 + k) = arr(i * 1024 + j * 32 + k) + x * k
+ end do
+ end do
+ end do
+ !$acc end parallel
+
+ do i = 0, 32 - 1
+ do j = 0, 32 -1
+ do k = 0, 32 - 1
+ idx = i * 1024 + j * 32 + k
+ if (arr(idx) .ne. idx + ieor(i, j * 3) * k + ior(i, j * 5) * k) then
+ call abort
+ end if
+ end do
+ end do
+ end do
+end subroutine t5
+
+
+! Test of vector-private variables declared on loop directive. Array type.
+
+subroutine t6()
+ integer :: i, j, k, idx, arr(0:32*32*32), pt(2)
+
+ do i = 0, 32*32*32-1
+ arr(i) = i
+ end do
+
+ !$acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
+ !$acc loop gang
+ do i = 0, 31
+ !$acc loop worker
+ do j = 0, 31
+ !$acc loop vector private(x, pt)
+ do k = 0, 31
+ pt(1) = ieor(i, j * 3)
+ pt(2) = ior(i, j * 5)
+ arr(i * 1024 + j * 32 + k) = arr(i * 1024 + j * 32 + k) + pt(1) * k
+ arr(i * 1024 + j * 32 + k) = arr(i * 1024 + j * 32 + k) + pt(2) * k
+ end do
+ end do
+ end do
+ !$acc end parallel
+
+ do i = 0, 32 - 1
+ do j = 0, 32 -1
+ do k = 0, 32 - 1
+ idx = i * 1024 + j * 32 + k
+ if (arr(idx) .ne. idx + ieor(i, j * 3) * k + ior(i, j * 5) * k) then
+ call abort
+ end if
+ end do
+ end do
+ end do
+end subroutine t6
+
+
+! Test of worker-private variables declared on a loop directive.
+
+subroutine t7()
+ integer :: x, i, j, arr(0:32*32)
+ common x
+
+ do i = 0, 32*32-1
+ arr(i) = i
+ end do
+
+ !$acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
+ ! { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 215 }
+ !$acc loop gang private(x)
+ do i = 0, 31
+ !$acc loop worker private(x)
+ do j = 0, 31
+ x = ieor(i, j * 3)
+ arr(i * 32 + j) = arr(i * 32 + j) + x
+ end do
+ end do
+ !$acc end parallel
+
+ do i = 0, 32 * 32 - 1
+ if (arr(i) .ne. i + ieor(i / 32, mod(i, 32) * 3)) call abort
+ end do
+end subroutine t7
+
+
+! Test of worker-private variables declared on a loop directive, broadcasting
+! to vector-partitioned mode.
+
+subroutine t8()
+ integer :: x, i, j, k, idx, arr(0:32*32*32)
+
+ do i = 0, 32*32*32-1
+ arr(i) = i
+ end do
+
+ !$acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
+ !$acc loop gang
+ do i = 0, 31
+ !$acc loop worker private(x)
+ do j = 0, 31
+ x = ieor(i, j * 3)
+
+ !$acc loop vector
+ do k = 0, 31
+ arr(i * 1024 + j * 32 + k) = arr(i * 1024 + j * 32 + k) + x * k
+ end do
+ end do
+ end do
+ !$acc end parallel
+
+ do i = 0, 32 - 1
+ do j = 0, 32 -1
+ do k = 0, 32 - 1
+ idx = i * 1024 + j * 32 + k
+ if (arr(idx) .ne. idx + ieor(i, j * 3) * k) call abort
+ end do
+ end do
+ end do
+end subroutine t8
+
+
+! Test of worker-private variables declared on a loop directive, broadcasting
+! to vector-partitioned mode. Back-to-back worker loops.
+
+subroutine t9()
+ integer :: x, i, j, k, idx, arr(0:32*32*32)
+
+ do i = 0, 32*32*32-1
+ arr(i) = i
+ end do
+
+ !$acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
+ !$acc loop gang
+ do i = 0, 31
+ !$acc loop worker private(x)
+ do j = 0, 31
+ x = ieor(i, j * 3)
+
+ !$acc loop vector
+ do k = 0, 31
+ arr(i * 1024 + j * 32 + k) = arr(i * 1024 + j * 32 + k) + x * k
+ end do
+ end do
+
+ !$acc loop worker private(x)
+ do j = 0, 31
+ x = ior(i, j * 5)
+
+ !$acc loop vector
+ do k = 0, 31
+ arr(i * 1024 + j * 32 + k) = arr(i * 1024 + j * 32 + k) + x * k
+ end do
+ end do
+ end do
+ !$acc end parallel
+
+ do i = 0, 32 - 1
+ do j = 0, 32 -1
+ do k = 0, 32 - 1
+ idx = i * 1024 + j * 32 + k
+ if (arr(idx) .ne. idx + ieor(i, j * 3) * k + ior(i, j * 5) * k) then
+ call abort
+ end if
+ end do
+ end do
+ end do
+end subroutine t9
+
+
+! Test of worker-private variables declared on a loop directive, broadcasting
+! to vector-partitioned mode. Successive vector loops. */
+
+subroutine t10()
+ integer :: x, i, j, k, idx, arr(0:32*32*32)
+
+ do i = 0, 32*32*32-1
+ arr(i) = i
+ end do
+
+ !$acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
+ !$acc loop gang
+ do i = 0, 31
+ !$acc loop worker private(x)
+ do j = 0, 31
+ x = ieor(i, j * 3)
+
+ !$acc loop vector
+ do k = 0, 31
+ arr(i * 1024 + j * 32 + k) = arr(i * 1024 + j * 32 + k) + x * k
+ end do
+
+ x = ior(i, j * 5)
+
+ !$acc loop vector
+ do k = 0, 31
+ arr(i * 1024 + j * 32 + k) = arr(i * 1024 + j * 32 + k) + x * k
+ end do
+ end do
+ end do
+ !$acc end parallel
+
+ do i = 0, 32 - 1
+ do j = 0, 32 -1
+ do k = 0, 32 - 1
+ idx = i * 1024 + j * 32 + k
+ if (arr(idx) .ne. idx + ieor(i, j * 3) * k + ior(i, j * 5) * k) then
+ call abort
+ end if
+ end do
+ end do
+ end do
+end subroutine t10
+
+
+! Test of worker-private variables declared on a loop directive, broadcasting
+! to vector-partitioned mode. Addressable worker variable.
+
+subroutine t11()
+ integer :: i, j, k, idx, arr(0:32*32*32)
+ integer, target :: x
+ integer, pointer :: p
+
+ do i = 0, 32*32*32-1
+ arr(i) = i
+ end do
+
+ !$acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
+ !$acc loop gang
+ do i = 0, 31
+ !$acc loop worker private(x, p)
+ do j = 0, 31
+ p => x
+ x = ieor(i, j * 3)
+
+ !$acc loop vector
+ do k = 0, 31
+ arr(i * 1024 + j * 32 + k) = arr(i * 1024 + j * 32 + k) + x * k
+ end do
+
+ p = ior(i, j * 5)
+
+ !$acc loop vector
+ do k = 0, 31
+ arr(i * 1024 + j * 32 + k) = arr(i * 1024 + j * 32 + k) + x * k
+ end do
+ end do
+ end do
+ !$acc end parallel
+
+ do i = 0, 32 - 1
+ do j = 0, 32 -1
+ do k = 0, 32 - 1
+ idx = i * 1024 + j * 32 + k
+ if (arr(idx) .ne. idx + ieor(i, j * 3) * k + ior(i, j * 5) * k) then
+ call abort
+ end if
+ end do
+ end do
+ end do
+end subroutine t11
+
+
+! Test of worker-private variables declared on a loop directive, broadcasting
+! to vector-partitioned mode. Aggregate worker variable.
+
+subroutine t12()
+ type vec2
+ integer x, y
+ end type vec2
+
+ integer :: i, j, k, idx, arr(0:32*32*32)
+ type(vec2) :: pt
+
+ do i = 0, 32*32*32-1
+ arr(i) = i
+ end do
+
+ !$acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
+ !$acc loop gang
+ do i = 0, 31
+ !$acc loop worker private(pt)
+ do j = 0, 31
+ pt%x = ieor(i, j * 3)
+ pt%y = ior(i, j * 5)
+
+ !$acc loop vector
+ do k = 0, 31
+ arr(i * 1024 + j * 32 + k) = arr(i * 1024 + j * 32 + k) + pt%x * k
+ end do
+
+ !$acc loop vector
+ do k = 0, 31
+ arr(i * 1024 + j * 32 + k) = arr(i * 1024 + j * 32 + k) + pt%y * k
+ end do
+ end do
+ end do
+ !$acc end parallel
+
+ do i = 0, 32 - 1
+ do j = 0, 32 -1
+ do k = 0, 32 - 1
+ idx = i * 1024 + j * 32 + k
+ if (arr(idx) .ne. idx + ieor(i, j * 3) * k + ior(i, j * 5) * k) then
+ call abort
+ end if
+ end do
+ end do
+ end do
+end subroutine t12
+
+
+! Test of worker-private variables declared on loop directive, broadcasting
+! to vector-partitioned mode. Array worker variable.
+
+subroutine t13()
+ integer :: i, j, k, idx, arr(0:32*32*32), pt(2)
+
+ do i = 0, 32*32*32-1
+ arr(i) = i
+ end do
+
+ !$acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
+ !$acc loop gang
+ do i = 0, 31
+ !$acc loop worker private(pt)
+ do j = 0, 31
+ pt(1) = ieor(i, j * 3)
+ pt(2) = ior(i, j * 5)
+
+ !$acc loop vector
+ do k = 0, 31
+ arr(i * 1024 + j * 32 + k) = arr(i * 1024 + j * 32 + k) + pt(1) * k
+ end do
+
+ !$acc loop vector
+ do k = 0, 31
+ arr(i * 1024 + j * 32 + k) = arr(i * 1024 + j * 32 + k) + pt(2) * k
+ end do
+ end do
+ end do
+ !$acc end parallel
+
+ do i = 0, 32 - 1
+ do j = 0, 32 -1
+ do k = 0, 32 - 1
+ idx = i * 1024 + j * 32 + k
+ if (arr(idx) .ne. idx + ieor(i, j * 3) * k + ior(i, j * 5) * k) then
+ call abort
+ end if
+ end do
+ end do
+ end do
+end subroutine t13
+
+
+! Test of gang-private variables declared on the parallel directive.
+
+subroutine t14()
+ use openacc
+ integer :: x = 5
+ integer, parameter :: n = 32
+ integer :: arr(n)
+
+ do i = 1, n
+ arr(i) = 3
+ end do
+
+ !$acc parallel private(x) copy(arr) num_gangs(n) num_workers(8) vector_length(32)
+ ! { dg-warning "region is worker partitioned but does not contain worker partitioned code" "worker" { target *-*-* } 515 }
+ ! { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 515 }
+ !$acc loop gang(static:1)
+ do i = 1, n
+ x = i * 2;
+ end do
+
+ !$acc loop gang(static:1)
+ do i = 1, n
+ if (acc_on_device (acc_device_host) .eqv. .TRUE.) x = i * 2
+ arr(i) = arr(i) + x
+ end do
+ !$acc end parallel
+
+ do i = 1, n
+ if (arr(i) .ne. (3 + i * 2)) call abort
+ end do
+
+end subroutine t14
+
+
+program main
+ call t1()
+ call t2()
+ call t3()
+ call t4()
+ call t5()
+ call t6()
+ call t7()
+ call t8()
+ call t9()
+ call t10()
+ call t11()
+ call t12()
+ call t13()
+ call t14()
+end program main
deleted file mode 100644
@@ -1,23 +0,0 @@
-! { dg-additional-options "-w" }
-
-! Test of gang-private variables declared on loop directive.
-
-program main
- integer :: x, i, arr(32)
-
- do i = 1, 32
- arr(i) = i
- end do
-
- !$acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
- !$acc loop gang private(x)
- do i = 1, 32
- x = i * 2;
- arr(i) = arr(i) + x
- end do
- !$acc end parallel
-
- do i = 1, 32
- if (arr(i) .ne. i * 3) call abort
- end do
-end program main
deleted file mode 100644
@@ -1,28 +0,0 @@
-! { dg-additional-options "-cpp -w" }
-
-! Test of gang-private variables declared on loop directive, with broadcasting
-! to partitioned workers.
-
-program main
- integer :: x, i, j, arr(0:32*32)
-
- do i = 0, 32*32-1
- arr(i) = i
- end do
-
- !$acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
- !$acc loop gang private(x)
- do i = 0, 31
- x = i * 2;
-
- !$acc loop worker
- do j = 0, 31
- arr(i * 32 + j) = arr(i * 32 + j) + x
- end do
- end do
- !$acc end parallel
-
- do i = 0, 32 * 32 - 1
- if (arr(i) .ne. i + (i / 32) * 2) call abort
- end do
-end program main
deleted file mode 100644
@@ -1,28 +0,0 @@
-! { dg-additional-options "-w" }
-
-! Test of gang-private variables declared on loop directive, with broadcasting
-! to partitioned vectors.
-
-program main
- integer :: x, i, j, arr(0:32*32)
-
- do i = 0, 32*32-1
- arr(i) = i
- end do
-
- !$acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
- !$acc loop gang private(x)
- do i = 0, 31
- x = i * 2;
-
- !$acc loop vector
- do j = 0, 31
- arr(i * 32 + j) = arr(i * 32 + j) + x
- end do
- end do
- !$acc end parallel
-
- do i = 0, 32 * 32 - 1
- if (arr(i) .ne. i + (i / 32) * 2) call abort
- end do
-end program main
deleted file mode 100644
@@ -1,36 +0,0 @@
-! { dg-additional-options "-w" }
-
-! Test of gang-private addressable variable declared on loop directive, with
-! broadcasting to partitioned workers.
-
-program main
- type vec3
- integer x, y, z, attr(13)
- end type vec3
-
- integer x, i, j, arr(0:32*32)
- type(vec3) pt
-
- do i = 0, 32*32-1
- arr(i) = i
- end do
-
- !$acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
- !$acc loop gang private(pt)
- do i = 0, 31
- pt%x = i
- pt%y = i * 2
- pt%z = i * 4
- pt%attr(5) = i * 6
-
- !$acc loop vector
- do j = 0, 31
- arr(i * 32 + j) = arr(i * 32 + j) + pt%x + pt%y + pt%z + pt%attr(5);
- end do
- end do
- !$acc end parallel
-
- do i = 0, 32 * 32 - 1
- if (arr(i) .ne. i + (i / 32) * 13) call abort
- end do
-end program main
deleted file mode 100644
@@ -1,39 +0,0 @@
-! Test of vector-private variables declared on loop directive.
-
-program main
- integer :: x, i, j, k, idx, arr(0:32*32*32)
-
- do i = 0, 32*32*32-1
- arr(i) = i
- end do
-
- !$acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
- !$acc loop gang
- do i = 0, 31
- !$acc loop worker
- do j = 0, 31
- !$acc loop vector private(x)
- do k = 0, 31
- x = ieor(i, j * 3)
- arr(i * 1024 + j * 32 + k) = arr(i * 1024 + j * 32 + k) + x * k
- end do
- !$acc loop vector private(x)
- do k = 0, 31
- x = ior(i, j * 5)
- arr(i * 1024 + j * 32 + k) = arr(i * 1024 + j * 32 + k) + x * k
- end do
- end do
- end do
- !$acc end parallel
-
- do i = 0, 32 - 1
- do j = 0, 32 -1
- do k = 0, 32 - 1
- idx = i * 1024 + j * 32 + k
- if (arr(idx) .ne. idx + ieor(i, j * 3) * k + ior(i, j * 5) * k) then
- call abort
- end if
- end do
- end do
- end do
-end program main
deleted file mode 100644
@@ -1,36 +0,0 @@
-! Test of vector-private variables declared on loop directive. Array type.
-
-program main
- integer :: i, j, k, idx, arr(0:32*32*32), pt(2)
-
- do i = 0, 32*32*32-1
- arr(i) = i
- end do
-
- !$acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
- !$acc loop gang
- do i = 0, 31
- !$acc loop worker
- do j = 0, 31
- !$acc loop vector private(x, pt)
- do k = 0, 31
- pt(1) = ieor(i, j * 3)
- pt(2) = ior(i, j * 5)
- arr(i * 1024 + j * 32 + k) = arr(i * 1024 + j * 32 + k) + pt(1) * k
- arr(i * 1024 + j * 32 + k) = arr(i * 1024 + j * 32 + k) + pt(2) * k
- end do
- end do
- end do
- !$acc end parallel
-
- do i = 0, 32 - 1
- do j = 0, 32 -1
- do k = 0, 32 - 1
- idx = i * 1024 + j * 32 + k
- if (arr(idx) .ne. idx + ieor(i, j * 3) * k + ior(i, j * 5) * k) then
- call abort
- end if
- end do
- end do
- end do
-end program main
deleted file mode 100644
@@ -1,27 +0,0 @@
-! { dg-additional-options "-cpp -w" }
-
-! Test of worker-private variables declared on a loop directive.
-
-program main
- integer :: x, i, j, arr(0:32*32)
- common x
-
- do i = 0, 32*32-1
- arr(i) = i
- end do
-
- !$acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
- !$acc loop gang private(x)
- do i = 0, 31
- !$acc loop worker private(x)
- do j = 0, 31
- x = ieor(i, j * 3)
- arr(i * 32 + j) = arr(i * 32 + j) + x
- end do
- end do
- !$acc end parallel
-
- do i = 0, 32 * 32 - 1
- if (arr(i) .ne. i + ieor(i / 32, mod(i, 32) * 3)) call abort
- end do
-end program main
deleted file mode 100644
@@ -1,34 +0,0 @@
-! Test of worker-private variables declared on a loop directive, broadcasting
-! to vector-partitioned mode.
-
-program main
- integer :: x, i, j, k, idx, arr(0:32*32*32)
-
- do i = 0, 32*32*32-1
- arr(i) = i
- end do
-
- !$acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
- !$acc loop gang
- do i = 0, 31
- !$acc loop worker private(x)
- do j = 0, 31
- x = ieor(i, j * 3)
-
- !$acc loop vector
- do k = 0, 31
- arr(i * 1024 + j * 32 + k) = arr(i * 1024 + j * 32 + k) + x * k
- end do
- end do
- end do
- !$acc end parallel
-
- do i = 0, 32 - 1
- do j = 0, 32 -1
- do k = 0, 32 - 1
- idx = i * 1024 + j * 32 + k
- if (arr(idx) .ne. idx + ieor(i, j * 3) * k) call abort
- end do
- end do
- end do
-end program main
deleted file mode 100644
@@ -1,46 +0,0 @@
-! Test of worker-private variables declared on a loop directive, broadcasting
-! to vector-partitioned mode. Back-to-back worker loops.
-
-program main
- integer :: x, i, j, k, idx, arr(0:32*32*32)
-
- do i = 0, 32*32*32-1
- arr(i) = i
- end do
-
- !$acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
- !$acc loop gang
- do i = 0, 31
- !$acc loop worker private(x)
- do j = 0, 31
- x = ieor(i, j * 3)
-
- !$acc loop vector
- do k = 0, 31
- arr(i * 1024 + j * 32 + k) = arr(i * 1024 + j * 32 + k) + x * k
- end do
- end do
-
- !$acc loop worker private(x)
- do j = 0, 31
- x = ior(i, j * 5)
-
- !$acc loop vector
- do k = 0, 31
- arr(i * 1024 + j * 32 + k) = arr(i * 1024 + j * 32 + k) + x * k
- end do
- end do
- end do
- !$acc end parallel
-
- do i = 0, 32 - 1
- do j = 0, 32 -1
- do k = 0, 32 - 1
- idx = i * 1024 + j * 32 + k
- if (arr(idx) .ne. idx + ieor(i, j * 3) * k + ior(i, j * 5) * k) then
- call abort
- end if
- end do
- end do
- end do
-end program main
deleted file mode 100644
@@ -1,43 +0,0 @@
-! Test of worker-private variables declared on a loop directive, broadcasting
-! to vector-partitioned mode. Successive vector loops. */
-
-program main
- integer :: x, i, j, k, idx, arr(0:32*32*32)
-
- do i = 0, 32*32*32-1
- arr(i) = i
- end do
-
- !$acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
- !$acc loop gang
- do i = 0, 31
- !$acc loop worker private(x)
- do j = 0, 31
- x = ieor(i, j * 3)
-
- !$acc loop vector
- do k = 0, 31
- arr(i * 1024 + j * 32 + k) = arr(i * 1024 + j * 32 + k) + x * k
- end do
-
- x = ior(i, j * 5)
-
- !$acc loop vector
- do k = 0, 31
- arr(i * 1024 + j * 32 + k) = arr(i * 1024 + j * 32 + k) + x * k
- end do
- end do
- end do
- !$acc end parallel
-
- do i = 0, 32 - 1
- do j = 0, 32 -1
- do k = 0, 32 - 1
- idx = i * 1024 + j * 32 + k
- if (arr(idx) .ne. idx + ieor(i, j * 3) * k + ior(i, j * 5) * k) then
- call abort
- end if
- end do
- end do
- end do
-end program main
deleted file mode 100644
@@ -1,46 +0,0 @@
-! Test of worker-private variables declared on a loop directive, broadcasting
-! to vector-partitioned mode. Addressable worker variable.
-
-program main
- integer :: i, j, k, idx, arr(0:32*32*32)
- integer, target :: x
- integer, pointer :: p
-
- do i = 0, 32*32*32-1
- arr(i) = i
- end do
-
- !$acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
- !$acc loop gang
- do i = 0, 31
- !$acc loop worker private(x, p)
- do j = 0, 31
- p => x
- x = ieor(i, j * 3)
-
- !$acc loop vector
- do k = 0, 31
- arr(i * 1024 + j * 32 + k) = arr(i * 1024 + j * 32 + k) + x * k
- end do
-
- p = ior(i, j * 5)
-
- !$acc loop vector
- do k = 0, 31
- arr(i * 1024 + j * 32 + k) = arr(i * 1024 + j * 32 + k) + x * k
- end do
- end do
- end do
- !$acc end parallel
-
- do i = 0, 32 - 1
- do j = 0, 32 -1
- do k = 0, 32 - 1
- idx = i * 1024 + j * 32 + k
- if (arr(idx) .ne. idx + ieor(i, j * 3) * k + ior(i, j * 5) * k) then
- call abort
- end if
- end do
- end do
- end do
-end program main
deleted file mode 100644
@@ -1,47 +0,0 @@
-! Test of worker-private variables declared on a loop directive, broadcasting
-! to vector-partitioned mode. Aggregate worker variable.
-
-program main
- type vec2
- integer x, y
- end type vec2
-
- integer :: i, j, k, idx, arr(0:32*32*32)
- type(vec2) :: pt
-
- do i = 0, 32*32*32-1
- arr(i) = i
- end do
-
- !$acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
- !$acc loop gang
- do i = 0, 31
- !$acc loop worker private(pt)
- do j = 0, 31
- pt%x = ieor(i, j * 3)
- pt%y = ior(i, j * 5)
-
- !$acc loop vector
- do k = 0, 31
- arr(i * 1024 + j * 32 + k) = arr(i * 1024 + j * 32 + k) + pt%x * k
- end do
-
- !$acc loop vector
- do k = 0, 31
- arr(i * 1024 + j * 32 + k) = arr(i * 1024 + j * 32 + k) + pt%y * k
- end do
- end do
- end do
- !$acc end parallel
-
- do i = 0, 32 - 1
- do j = 0, 32 -1
- do k = 0, 32 - 1
- idx = i * 1024 + j * 32 + k
- if (arr(idx) .ne. idx + ieor(i, j * 3) * k + ior(i, j * 5) * k) then
- call abort
- end if
- end do
- end do
- end do
-end program main
deleted file mode 100644
@@ -1,42 +0,0 @@
-! Test of worker-private variables declared on loop directive, broadcasting
-! to vector-partitioned mode. Array worker variable.
-
-program main
- integer :: i, j, k, idx, arr(0:32*32*32), pt(2)
-
- do i = 0, 32*32*32-1
- arr(i) = i
- end do
-
- !$acc parallel copy(arr) num_gangs(32) num_workers(8) vector_length(32)
- !$acc loop gang
- do i = 0, 31
- !$acc loop worker private(pt)
- do j = 0, 31
- pt(1) = ieor(i, j * 3)
- pt(2) = ior(i, j * 5)
-
- !$acc loop vector
- do k = 0, 31
- arr(i * 1024 + j * 32 + k) = arr(i * 1024 + j * 32 + k) + pt(1) * k
- end do
-
- !$acc loop vector
- do k = 0, 31
- arr(i * 1024 + j * 32 + k) = arr(i * 1024 + j * 32 + k) + pt(2) * k
- end do
- end do
- end do
- !$acc end parallel
-
- do i = 0, 32 - 1
- do j = 0, 32 -1
- do k = 0, 32 - 1
- idx = i * 1024 + j * 32 + k
- if (arr(idx) .ne. idx + ieor(i, j * 3) * k + ior(i, j * 5) * k) then
- call abort
- end if
- end do
- end do
- end do
-end program main
deleted file mode 100644
@@ -1,32 +0,0 @@
-! { dg-additional-options "-w" }
-
-! Test of gang-private variables declared on the parallel directive.
-
-program main
- use openacc
- integer :: x = 5
- integer, parameter :: n = 32
- integer :: arr(n)
-
- do i = 1, n
- arr(i) = 3
- end do
-
- !$acc parallel private(x) copy(arr) num_gangs(n) num_workers(8) vector_length(32)
- !$acc loop gang(static:1)
- do i = 1, n
- x = i * 2;
- end do
-
- !$acc loop gang(static:1)
- do i = 1, n
- if (acc_on_device (acc_device_host) .eqv. .TRUE.) x = i * 2
- arr(i) = arr(i) + x
- end do
- !$acc end parallel
-
- do i = 1, n
- if (arr(i) .ne. (3 + i * 2)) call abort
- end do
-
-end program main
@@ -1,5 +1,5 @@
! { dg-do run }
-! { dg-additional-options "-cpp -w" }
+! { dg-additional-options "-cpp" }
#define M 8
#define N 32
@@ -99,7 +99,7 @@ subroutine gang (a)
integer, intent (inout) :: a(N)
integer :: i
- !$acc loop gang
+ !$acc loop gang worker vector
do i = 1, N
a(i) = a(i) - i
end do
deleted file mode 100644
@@ -1,239 +0,0 @@
-! Copy of update-1.f90 with self exchanged with host for !$acc update
-
-! { dg-do run }
-! { dg-skip-if "" { *-*-* } { "*" } { "-DACC_MEM_SHARED=0" } }
-
-program update
- use openacc
- implicit none
- integer, parameter :: N = 8
- real :: a(N), b(N)
- integer i
-
- do i = 1, N
- a(i) = 3.0
- b(i) = 0.0
- end do
-
- !$acc enter data copyin (a, b)
-
- !$acc parallel present (a, b)
- do i = 1, N
- b(i) = a(i)
- end do
- !$acc end parallel
-
- !$acc update self (a, b)
-
- do i = 1, N
- if (a(i) .ne. 3.0) call abort
- if (b(i) .ne. 3.0) call abort
- end do
-
- if (acc_is_present (a) .neqv. .TRUE.) call abort
- if (acc_is_present (b) .neqv. .TRUE.) call abort
-
- do i = 1, N
- a(i) = 5.0
- b(i) = 1.0
- end do
-
- !$acc update device (a, b)
-
- !$acc parallel present (a, b)
- do i = 1, N
- b(i) = a(i)
- end do
- !$acc end parallel
-
- !$acc update self (a, b)
-
- do i = 1, N
- if (a(i) .ne. 5.0) call abort
- if (b(i) .ne. 5.0) call abort
- end do
-
- if (acc_is_present (a) .neqv. .TRUE.) call abort
- if (acc_is_present (b) .neqv. .TRUE.) call abort
-
- !$acc parallel present (a, b)
- do i = 1, N
- b(i) = a(i)
- end do
- !$acc end parallel
-
- !$acc update self (a, b)
-
- do i = 1, N
- if (a(i) .ne. 5.0) call abort
- if (b(i) .ne. 5.0) call abort
- end do
-
- do i = 1, N
- a(i) = 6.0
- b(i) = 0.0
- end do
-
- !$acc update device (a, b)
-
- do i = 1, N
- a(i) = 9.0
- end do
-
- !$acc parallel present (a, b)
- do i = 1, N
- b(i) = a(i)
- end do
- !$acc end parallel
-
- !$acc update self (a, b)
-
- do i = 1, N
- if (a(i) .ne. 6.0) call abort
- if (b(i) .ne. 6.0) call abort
- end do
-
- if (acc_is_present (a) .neqv. .TRUE.) call abort
- if (acc_is_present (b) .neqv. .TRUE.) call abort
-
- do i = 1, N
- a(i) = 7.0
- b(i) = 2.0
- end do
-
- !$acc update device (a, b)
-
- do i = 1, N
- a(i) = 9.0
- end do
-
- !$acc parallel present (a, b)
- do i = 1, N
- b(i) = a(i)
- end do
- !$acc end parallel
-
- !$acc update self (a, b)
-
- do i = 1, N
- if (a(i) .ne. 7.0) call abort
- if (b(i) .ne. 7.0) call abort
- end do
-
- do i = 1, N
- a(i) = 9.0
- end do
-
- !$acc update device (a)
-
- !$acc parallel present (a, b)
- do i = 1, N
- b(i) = a(i)
- end do
- !$acc end parallel
-
- !$acc update self (a, b)
-
- do i = 1, N
- if (a(i) .ne. 9.0) call abort
- if (b(i) .ne. 9.0) call abort
- end do
-
- if (acc_is_present (a) .neqv. .TRUE.) call abort
- if (acc_is_present (b) .neqv. .TRUE.) call abort
-
- do i = 1, N
- a(i) = 5.0
- end do
-
- !$acc update device (a)
-
- do i = 1, N
- a(i) = 6.0
- end do
-
- !$acc update device (a(1:rshift (N, 1)))
-
- !$acc parallel present (a, b)
- do i = 1, N
- b(i) = a(i)
- end do
- !$acc end parallel
-
- !$acc update self (a, b)
-
- do i = 1, rshift (N, 1)
- if (a(i) .ne. 6.0) call abort
- if (b(i) .ne. 6.0) call abort
- end do
-
- do i = rshift (N, 1) + 1, N
- if (a(i) .ne. 5.0) call abort
- if (b(i) .ne. 5.0) call abort
- end do
-
- if (acc_is_present (a) .neqv. .TRUE.) call abort
- if (acc_is_present (b) .neqv. .TRUE.) call abort
-
- do i = 1, N
- a(i) = 0.0
- end do
-
- !$acc update device (a(1:4))
-
- !$acc parallel present (a)
- do i = 1, N
- a(i) = a(i) + 1.0
- end do
- !$acc end parallel
-
- !$acc update self (a(5:N))
-
- do i = 1, rshift (N, 1)
- if (a(i) .ne. 0.0) call abort
- end do
-
- do i = rshift (N, 1) + 1, N
- if (a(i) .ne. 6.0) call abort
- end do
-
- !$acc update self (a(1:4))
-
- do i = 1, rshift (N, 1)
- if (a(i) .ne. 1.0) call abort
- end do
-
- do i = rshift (N, 1) + 1, N
- if (a(i) .ne. 6.0) call abort
- end do
-
- a(3) = 9
- a(4) = 9
- a(5) = 9
- a(6) = 9
-
- !$acc update device (a(3:6))
-
- !$acc parallel present (a(1:N))
- do i = 1, N
- a(i) = a(i) + 1.0
- end do
- !$acc end parallel
-
- !$acc update self (a(3:6))
-
- do i = 1, 2
- if (a(i) .ne. 1.0) call abort
- end do
-
- do i = 3, 6
- if (a(i) .ne. 10.0) call abort
- end do
-
- do i = 7, N
- if (a(i) .ne. 6.0) call abort
- end do
- !$acc exit data delete (a, b)
-
-end program
-