+2018-05-02 Tom de Vries <tom@codesourcery.com>
+
+ PR libgomp/82428
+ * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
+ * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
+ (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
+ * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
+ (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
+ * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
+ __builtin_goacc_parlevel_size.
+
2018-05-02 Richard Biener <rguenther@suse.de>
PR tree-optimization/85597
#include "gimple-fold.h"
#include "intl.h"
#include "file-prefix-map.h" /* remap_macro_filename() */
+#include "gomp-constants.h"
+#include "omp-general.h"
struct target_builtins default_target_builtins;
#if SWITCHABLE_TARGET
return ret;
}
+/* Emit code to get the openacc gang, worker or vector id or size. */
+
+static rtx
+expand_builtin_goacc_parlevel_id_size (tree exp, rtx target, int ignore)
+{
+ const char *name;
+ rtx fallback_retval;
+ rtx_insn *(*gen_fn) (rtx, rtx);
+ switch (DECL_FUNCTION_CODE (get_callee_fndecl (exp)))
+ {
+ case BUILT_IN_GOACC_PARLEVEL_ID:
+ name = "__builtin_goacc_parlevel_id";
+ fallback_retval = const0_rtx;
+ gen_fn = targetm.gen_oacc_dim_pos;
+ break;
+ case BUILT_IN_GOACC_PARLEVEL_SIZE:
+ name = "__builtin_goacc_parlevel_size";
+ fallback_retval = const1_rtx;
+ gen_fn = targetm.gen_oacc_dim_size;
+ break;
+ default:
+ gcc_unreachable ();
+ }
+
+ if (oacc_get_fn_attrib (current_function_decl) == NULL_TREE)
+ {
+ error ("%qs only supported in OpenACC code", name);
+ return const0_rtx;
+ }
+
+ tree arg = CALL_EXPR_ARG (exp, 0);
+ if (TREE_CODE (arg) != INTEGER_CST)
+ {
+ error ("non-constant argument 0 to %qs", name);
+ return const0_rtx;
+ }
+
+ int dim = TREE_INT_CST_LOW (arg);
+ switch (dim)
+ {
+ case GOMP_DIM_GANG:
+ case GOMP_DIM_WORKER:
+ case GOMP_DIM_VECTOR:
+ break;
+ default:
+ error ("illegal argument 0 to %qs", name);
+ return const0_rtx;
+ }
+
+ if (ignore)
+ return target;
+
+ if (!targetm.have_oacc_dim_size ())
+ {
+ emit_move_insn (target, fallback_retval);
+ return target;
+ }
+
+ rtx reg = MEM_P (target) ? gen_reg_rtx (GET_MODE (target)) : target;
+ emit_insn (gen_fn (reg, GEN_INT (dim)));
+ if (reg != target)
+ emit_move_insn (target, reg);
+
+ return target;
+}
/* Expand an expression EXP that calls a built-in function,
with result going to TARGET if that's convenient
folding. */
break;
+ case BUILT_IN_GOACC_PARLEVEL_ID:
+ case BUILT_IN_GOACC_PARLEVEL_SIZE:
+ return expand_builtin_goacc_parlevel_id_size (exp, target, ignore);
+
default: /* just do library call, if unknown builtin */
break;
}
#define DEF_GOACC_BUILTIN_COMPILER(ENUM, NAME, TYPE, ATTRS) \
DEF_BUILTIN (ENUM, "__builtin_" NAME, BUILT_IN_NORMAL, TYPE, TYPE, \
flag_openacc, true, true, ATTRS, false, true)
+#undef DEF_GOACC_BUILTIN_ONLY
+#define DEF_GOACC_BUILTIN_ONLY(ENUM, NAME, TYPE, ATTRS) \
+ DEF_BUILTIN (ENUM, "__builtin_" NAME, BUILT_IN_NORMAL, TYPE, BT_LAST, \
+ false, false, true, ATTRS, false, flag_openacc)
#undef DEF_GOMP_BUILTIN
#define DEF_GOMP_BUILTIN(ENUM, NAME, TYPE, ATTRS) \
DEF_BUILTIN (ENUM, "__builtin_" NAME, BUILT_IN_NORMAL, TYPE, TYPE, \
code.
@end deftypefn
+@deftypefn {Built-in Function} int __builtin_goacc_parlevel_id (int x)
+Returns the openacc gang, worker or vector id depending on whether @var{x} is
+0, 1 or 2.
+@end deftypefn
+
+@deftypefn {Built-in Function} int __builtin_goacc_parlevel_size (int x)
+Returns the openacc gang, worker or vector size depending on whether @var{x} is
+0, 1 or 2.
+@end deftypefn
+
@node Target Builtins
@section Built-in Functions Specific to Particular Target Machines
+2018-05-02 Tom de Vries <tom@codesourcery.com>
+
+ PR libgomp/82428
+ * f95-lang.c (DEF_GOACC_BUILTIN_ONLY): Define.
+
2018-04-24 Steven G. Kargl <kargl@gcc.gnu.org>
PR fortran/85520
#undef DEF_GOACC_BUILTIN_COMPILER
#define DEF_GOACC_BUILTIN_COMPILER(code, name, type, attr) \
gfc_define_builtin (name, builtin_types[type], code, name, attr);
+#undef DEF_GOACC_BUILTIN_ONLY
+#define DEF_GOACC_BUILTIN_ONLY(code, name, type, attr) \
+ gfc_define_builtin ("__builtin_" name, builtin_types[type], code, NULL, \
+ attr);
#undef DEF_GOMP_BUILTIN
#define DEF_GOMP_BUILTIN(code, name, type, attr) /* ignore */
#include "../omp-builtins.def"
DEF_GOACC_BUILTIN_COMPILER (BUILT_IN_ACC_ON_DEVICE, "acc_on_device",
BT_FN_INT_INT, ATTR_CONST_NOTHROW_LEAF_LIST)
+DEF_GOACC_BUILTIN_ONLY (BUILT_IN_GOACC_PARLEVEL_ID, "goacc_parlevel_id",
+ BT_FN_INT_INT, ATTR_NOTHROW_LEAF_LIST)
+DEF_GOACC_BUILTIN_ONLY (BUILT_IN_GOACC_PARLEVEL_SIZE, "goacc_parlevel_size",
+ BT_FN_INT_INT, ATTR_NOTHROW_LEAF_LIST)
+
DEF_GOMP_BUILTIN (BUILT_IN_OMP_GET_THREAD_NUM, "omp_get_thread_num",
BT_FN_INT, ATTR_CONST_NOTHROW_LEAF_LIST)
DEF_GOMP_BUILTIN (BUILT_IN_OMP_GET_NUM_THREADS, "omp_get_num_threads",
+2018-05-02 Tom de Vries <tom@codesourcery.com>
+
+ PR libgomp/82428
+ * c-c++-common/goacc/builtin-goacc-parlevel-id-size-2.c: New test.
+ * c-c++-common/goacc/builtin-goacc-parlevel-id-size.c: New test.
+
2018-05-02 David Pagan <dave.pagan@oracle.com>
PR c/30552
--- /dev/null
+/* { dg-do compile } */
+/* { dg-additional-options "-O2" } */
+
+#include "../../../../include/gomp-constants.h"
+
+void
+foo (void)
+{
+ __builtin_goacc_parlevel_id (GOMP_DIM_GANG);
+ /* { dg-error "'__builtin_goacc_parlevel_id' only supported in OpenACC code" "" { target *-*-* } .-1 } */
+
+ __builtin_goacc_parlevel_size (GOMP_DIM_GANG);
+ /* { dg-error "'__builtin_goacc_parlevel_size' only supported in OpenACC code" "" { target *-*-* } .-1 } */
+}
+
+#pragma acc routine
+void
+foo2 (int arg)
+{
+ __builtin_goacc_parlevel_id (arg);
+ /* { dg-error "non-constant argument 0 to '__builtin_goacc_parlevel_id'" "" { target *-*-* } .-1 } */
+
+ __builtin_goacc_parlevel_size (arg);
+ /* { dg-error "non-constant argument 0 to '__builtin_goacc_parlevel_size'" "" { target *-*-* } .-1 } */
+
+ __builtin_goacc_parlevel_id (-1);
+ /* { dg-error "illegal argument 0 to '__builtin_goacc_parlevel_id'" "" { target *-*-* } .-1 } */
+
+ __builtin_goacc_parlevel_id (-1);
+ /* { dg-error "illegal argument 0 to '__builtin_goacc_parlevel_id'" "" { target *-*-* } .-1 } */
+
+ __builtin_goacc_parlevel_size (-1);
+ /* { dg-error "illegal argument 0 to '__builtin_goacc_parlevel_size'" "" { target *-*-* } .-1 } */
+
+ __builtin_goacc_parlevel_size (3);
+ /* { dg-error "illegal argument 0 to '__builtin_goacc_parlevel_size'" "" { target *-*-* } .-1 } */
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-additional-options "-O2" } */
+
+#include "../../../../include/gomp-constants.h"
+
+#pragma acc routine
+int
+foo (void)
+{
+ int res;
+
+ __builtin_goacc_parlevel_id (GOMP_DIM_GANG);
+ __builtin_goacc_parlevel_id (GOMP_DIM_WORKER);
+ __builtin_goacc_parlevel_id (GOMP_DIM_VECTOR);
+
+ __builtin_goacc_parlevel_size (GOMP_DIM_GANG);
+ __builtin_goacc_parlevel_size (GOMP_DIM_WORKER);
+ __builtin_goacc_parlevel_size (GOMP_DIM_VECTOR);
+
+ res += __builtin_goacc_parlevel_id (GOMP_DIM_GANG);
+ res += __builtin_goacc_parlevel_id (GOMP_DIM_WORKER);
+ res += __builtin_goacc_parlevel_id (GOMP_DIM_VECTOR);
+
+ res += __builtin_goacc_parlevel_size (GOMP_DIM_GANG);
+ res += __builtin_goacc_parlevel_size (GOMP_DIM_WORKER);
+ res += __builtin_goacc_parlevel_size (GOMP_DIM_VECTOR);
+
+ return res;
+}
+
+void
+foo2 (void)
+{
+ int res;
+
+#pragma acc parallel
+ {
+ __builtin_goacc_parlevel_id (GOMP_DIM_GANG);
+ __builtin_goacc_parlevel_id (GOMP_DIM_WORKER);
+ __builtin_goacc_parlevel_id (GOMP_DIM_VECTOR);
+
+ __builtin_goacc_parlevel_size (GOMP_DIM_GANG);
+ __builtin_goacc_parlevel_size (GOMP_DIM_WORKER);
+ __builtin_goacc_parlevel_size (GOMP_DIM_VECTOR);
+
+ res += __builtin_goacc_parlevel_id (GOMP_DIM_GANG);
+ res += __builtin_goacc_parlevel_id (GOMP_DIM_WORKER);
+ res += __builtin_goacc_parlevel_id (GOMP_DIM_VECTOR);
+
+ res += __builtin_goacc_parlevel_size (GOMP_DIM_GANG);
+ res += __builtin_goacc_parlevel_size (GOMP_DIM_WORKER);
+ res += __builtin_goacc_parlevel_size (GOMP_DIM_VECTOR);
+ }
+}
+
+void
+foo3 (void)
+{
+ int res;
+
+#pragma acc kernels
+ {
+ __builtin_goacc_parlevel_id (GOMP_DIM_GANG);
+ __builtin_goacc_parlevel_id (GOMP_DIM_WORKER);
+ __builtin_goacc_parlevel_id (GOMP_DIM_VECTOR);
+
+ __builtin_goacc_parlevel_size (GOMP_DIM_GANG);
+ __builtin_goacc_parlevel_size (GOMP_DIM_WORKER);
+ __builtin_goacc_parlevel_size (GOMP_DIM_VECTOR);
+
+ res += __builtin_goacc_parlevel_id (GOMP_DIM_GANG);
+ res += __builtin_goacc_parlevel_id (GOMP_DIM_WORKER);
+ res += __builtin_goacc_parlevel_id (GOMP_DIM_VECTOR);
+
+ res += __builtin_goacc_parlevel_size (GOMP_DIM_GANG);
+ res += __builtin_goacc_parlevel_size (GOMP_DIM_WORKER);
+ res += __builtin_goacc_parlevel_size (GOMP_DIM_VECTOR);
+ }
+}
+2018-05-02 Tom de Vries <tom@codesourcery.com>
+
+ PR libgomp/82428
+ * testsuite/libgomp.oacc-c-c++-common/gang-static-2.c: Use
+ __builtin_goacc_parlevel_{id,size}.
+ * testsuite/libgomp.oacc-c-c++-common/loop-auto-1.c: Same.
+ * testsuite/libgomp.oacc-c-c++-common/loop-dim-default.c: Same.
+ * testsuite/libgomp.oacc-c-c++-common/loop-g-1.c: Same.
+ * testsuite/libgomp.oacc-c-c++-common/loop-g-2.c: Same.
+ * testsuite/libgomp.oacc-c-c++-common/loop-gwv-1.c: Same.
+ * testsuite/libgomp.oacc-c-c++-common/loop-red-g-1.c: Same.
+ * testsuite/libgomp.oacc-c-c++-common/loop-red-gwv-1.c: Same.
+ * testsuite/libgomp.oacc-c-c++-common/loop-red-v-1.c: Same.
+ * testsuite/libgomp.oacc-c-c++-common/loop-red-v-2.c: Same.
+ * testsuite/libgomp.oacc-c-c++-common/loop-red-w-1.c: Same.
+ * testsuite/libgomp.oacc-c-c++-common/loop-red-w-2.c: Same.
+ * testsuite/libgomp.oacc-c-c++-common/loop-red-wv-1.c: Same.
+ * testsuite/libgomp.oacc-c-c++-common/loop-v-1.c: Same.
+ * testsuite/libgomp.oacc-c-c++-common/loop-w-1.c: Same.
+ * testsuite/libgomp.oacc-c-c++-common/loop-wv-1.c: Same.
+ * testsuite/libgomp.oacc-c-c++-common/parallel-dims.c: Same.
+ * testsuite/libgomp.oacc-c-c++-common/routine-g-1.c: Same.
+ * testsuite/libgomp.oacc-c-c++-common/routine-gwv-1.c: Same.
+ * testsuite/libgomp.oacc-c-c++-common/routine-v-1.c: Same.
+ * testsuite/libgomp.oacc-c-c++-common/routine-w-1.c: Same.
+ * testsuite/libgomp.oacc-c-c++-common/routine-wv-1.c: Same.
+ * testsuite/libgomp.oacc-c-c++-common/routine-wv-2.c: Same.
+ * testsuite/libgomp.oacc-c-c++-common/tile-1.c: Same.
+
2018-05-02 Tom de Vries <tom@codesourcery.com>
PR testsuite/85106
-/* { dg-do run { target openacc_nvidia_accel_selected } } */
-/* 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 <assert.h>
#include <openacc.h>
+#include <gomp-constants.h>
#define N 100
#define GANG_ID(I) \
- (acc_on_device (acc_device_nvidia) \
- ? ({unsigned __r; \
- __asm__ volatile ("mov.u32 %0,%%ctaid.x;" : "=r" (__r)); \
- __r; }) : (I))
+ (acc_on_device (acc_device_not_host) \
+ ? __builtin_goacc_parlevel_id (GOMP_DIM_GANG) \
+ : (I))
void
test_static(int *a, int num_gangs, int sarg)
{
int i, j;
- if (sarg == 0)
+ if (acc_on_device (acc_device_host))
+ return;
+
+ if (sarg == 0)
sarg = 1;
for (i = 0; i < N / sarg; i++)
{
int i, j;
+ if (acc_on_device (acc_device_host))
+ return;
+
for (i = 0; i < N; i+=gangs)
for (j = 0; j < gangs; j++)
assert (a[i+j] == i/gangs);
-/* 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" } { "" } } */
-
/* { dg-additional-options "-fopenacc-dim=32" } */
#include <stdio.h>
#include <openacc.h>
+#include <gomp-constants.h>
int check (const int *ary, int size, int gp, int wp, int vp)
{
{
int r = 0;
- if (acc_on_device (acc_device_nvidia))
- {
- int g = 0, w = 0, v = 0;
+ int g = 0, w = 0, v = 0;
+ g = __builtin_goacc_parlevel_id (GOMP_DIM_GANG);
+ w = __builtin_goacc_parlevel_id (GOMP_DIM_WORKER);
+ v = __builtin_goacc_parlevel_id (GOMP_DIM_VECTOR);
+ r = (g << 16) | (w << 8) | v;
- __asm__ volatile ("mov.u32 %0,%%ctaid.x;" : "=r" (g));
- __asm__ volatile ("mov.u32 %0,%%tid.y;" : "=r" (w));
- __asm__ volatile ("mov.u32 %0,%%tid.x;" : "=r" (v));
- r = (g << 16) | (w << 8) | v;
- }
return r;
}
-/* 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" } { "" } } */
/* { dg-additional-options "-fopenacc-dim=16:16" } */
#include <openacc.h>
#include <alloca.h>
#include <string.h>
#include <stdio.h>
+#include <gomp-constants.h>
#pragma acc routine
static int __attribute__ ((noinline)) coord ()
{
int res = 0;
- if (acc_on_device (acc_device_nvidia))
+ if (acc_on_device (acc_device_not_host))
{
- int g = 0, w = 0, v = 0;
+ int g, w, v;
- __asm__ volatile ("mov.u32 %0,%%ctaid.x;" : "=r" (g));
- __asm__ volatile ("mov.u32 %0,%%tid.y;" : "=r" (w));
- __asm__ volatile ("mov.u32 %0,%%tid.x;" : "=r" (v));
+ g = __builtin_goacc_parlevel_id (GOMP_DIM_GANG);
+ w = __builtin_goacc_parlevel_id (GOMP_DIM_WORKER);
+ v = __builtin_goacc_parlevel_id (GOMP_DIM_VECTOR);
res = (1 << 24) | (g << 16) | (w << 8) | v;
}
return res;
-/* 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>
+#include <gomp-constants.h>
#define N (32*32*32+17)
int main ()
#pragma acc loop gang
for (unsigned ix = 0; ix < N; ix++)
{
- if (__builtin_acc_on_device (5))
+ if (acc_on_device (acc_device_not_host))
{
- int g = 0, w = 0, v = 0;
-
- __asm__ volatile ("mov.u32 %0,%%ctaid.x;" : "=r" (g));
- __asm__ volatile ("mov.u32 %0,%%tid.y;" : "=r" (w));
- __asm__ volatile ("mov.u32 %0,%%tid.x;" : "=r" (v));
+ int g, w, v;
+ g = __builtin_goacc_parlevel_id (GOMP_DIM_GANG);
+ w = __builtin_goacc_parlevel_id (GOMP_DIM_WORKER);
+ v = __builtin_goacc_parlevel_id (GOMP_DIM_VECTOR);
ary[ix] = (g << 16) | (w << 8) | v;
ondev = 1;
}
-/* 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>
+#include <gomp-constants.h>
#define N (32*32*32+17)
int main ()
#pragma acc loop gang (static:1)
for (unsigned ix = 0; ix < N; ix++)
{
- if (__builtin_acc_on_device (5))
+ if (acc_on_device (acc_device_not_host))
{
- int g = 0, w = 0, v = 0;
+ int g, w, v;
- __asm__ volatile ("mov.u32 %0,%%ctaid.x;" : "=r" (g));
- __asm__ volatile ("mov.u32 %0,%%tid.y;" : "=r" (w));
- __asm__ volatile ("mov.u32 %0,%%tid.x;" : "=r" (v));
+ g = __builtin_goacc_parlevel_id (GOMP_DIM_GANG);
+ w = __builtin_goacc_parlevel_id (GOMP_DIM_WORKER);
+ v = __builtin_goacc_parlevel_id (GOMP_DIM_VECTOR);
ary[ix] = (g << 16) | (w << 8) | v;
ondev = 1;
}
-/* 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>
+#include <gomp-constants.h>
#define N (32*32*32+17)
int main ()
#pragma acc loop gang worker vector
for (unsigned ix = 0; ix < N; ix++)
{
- if (__builtin_acc_on_device (5))
+ if (acc_on_device (acc_device_not_host))
{
- int g = 0, w = 0, v = 0;
+ int g, w, v;
+
+ g = __builtin_goacc_parlevel_id (GOMP_DIM_GANG);
+ w = __builtin_goacc_parlevel_id (GOMP_DIM_WORKER);
+ v = __builtin_goacc_parlevel_id (GOMP_DIM_VECTOR);
- __asm__ volatile ("mov.u32 %0,%%ctaid.x;" : "=r" (g));
- __asm__ volatile ("mov.u32 %0,%%tid.y;" : "=r" (w));
- __asm__ volatile ("mov.u32 %0,%%tid.x;" : "=r" (v));
ary[ix] = (g << 16) | (w << 8) | v;
ondev = 1;
}
-/* 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>
+#include <gomp-constants.h>
#define N (32*32*32+17)
int main ()
{
int val = ix;
- if (__builtin_acc_on_device (5))
+ if (acc_on_device (acc_device_not_host))
{
- int g = 0, w = 0, v = 0;
+ int g, w, v;
- __asm__ volatile ("mov.u32 %0,%%ctaid.x;" : "=r" (g));
- __asm__ volatile ("mov.u32 %0,%%tid.y;" : "=r" (w));
- __asm__ volatile ("mov.u32 %0,%%tid.x;" : "=r" (v));
+ g = __builtin_goacc_parlevel_id (GOMP_DIM_GANG);
+ w = __builtin_goacc_parlevel_id (GOMP_DIM_WORKER);
+ v = __builtin_goacc_parlevel_id (GOMP_DIM_VECTOR);
val = (g << 16) | (w << 8) | v;
ondev = 1;
}
-/* 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>
+#include <gomp-constants.h>
#define N (32*32*32+17)
int main ()
{
int val = ix;
- if (__builtin_acc_on_device (5))
+ if (acc_on_device (acc_device_not_host))
{
- int g = 0, w = 0, v = 0;
+ int g, w, v;
- __asm__ volatile ("mov.u32 %0,%%ctaid.x;" : "=r" (g));
- __asm__ volatile ("mov.u32 %0,%%tid.y;" : "=r" (w));
- __asm__ volatile ("mov.u32 %0,%%tid.x;" : "=r" (v));
+ g = __builtin_goacc_parlevel_id (GOMP_DIM_GANG);
+ w = __builtin_goacc_parlevel_id (GOMP_DIM_WORKER);
+ v = __builtin_goacc_parlevel_id (GOMP_DIM_VECTOR);
val = (g << 16) | (w << 8) | v;
ondev = 1;
}
-/* 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>
+#include <gomp-constants.h>
#define N (32*32*32+17)
{
int val = ix;
- if (__builtin_acc_on_device (5))
+ if (acc_on_device (acc_device_not_host))
{
- int g = 0, w = 0, v = 0;
+ int g, w, v;
- __asm__ volatile ("mov.u32 %0,%%ctaid.x;" : "=r" (g));
- __asm__ volatile ("mov.u32 %0,%%tid.y;" : "=r" (w));
- __asm__ volatile ("mov.u32 %0,%%tid.x;" : "=r" (v));
+ g = __builtin_goacc_parlevel_id (GOMP_DIM_GANG);
+ w = __builtin_goacc_parlevel_id (GOMP_DIM_WORKER);
+ v = __builtin_goacc_parlevel_id (GOMP_DIM_VECTOR);
val = (g << 16) | (w << 8) | v;
ondev = 1;
}
-/* 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>
+#include <gomp-constants.h>
#define N (32*32*32+17)
{
int val = ix;
- if (__builtin_acc_on_device (5))
+ if (acc_on_device (acc_device_not_host))
{
- int g = 0, w = 0, v = 0;
+ int g, w, v;
- __asm__ volatile ("mov.u32 %0,%%ctaid.x;" : "=r" (g));
- __asm__ volatile ("mov.u32 %0,%%tid.y;" : "=r" (w));
- __asm__ volatile ("mov.u32 %0,%%tid.x;" : "=r" (v));
+ g = __builtin_goacc_parlevel_id (GOMP_DIM_GANG);
+ w = __builtin_goacc_parlevel_id (GOMP_DIM_WORKER);
+ v = __builtin_goacc_parlevel_id (GOMP_DIM_VECTOR);
val = (g << 16) | (w << 8) | v;
ondev = 1;
}
-/* 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>
+#include <gomp-constants.h>
#define N (32*32*32+17)
int main ()
{
int val = ix;
- if (__builtin_acc_on_device (5))
+ if (acc_on_device (acc_device_not_host))
{
- int g = 0, w = 0, v = 0;
+ int g, w, v;
- __asm__ volatile ("mov.u32 %0,%%ctaid.x;" : "=r" (g));
- __asm__ volatile ("mov.u32 %0,%%tid.y;" : "=r" (w));
- __asm__ volatile ("mov.u32 %0,%%tid.x;" : "=r" (v));
+ g = __builtin_goacc_parlevel_id (GOMP_DIM_GANG);
+ w = __builtin_goacc_parlevel_id (GOMP_DIM_WORKER);
+ v = __builtin_goacc_parlevel_id (GOMP_DIM_VECTOR);
val = (g << 16) | (w << 8) | v;
ondev = 1;
}
-/* 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>
+#include <gomp-constants.h>
#define N (32*32*32+17)
int main ()
{
int val = ix;
- if (__builtin_acc_on_device (5))
+ if (acc_on_device (acc_device_not_host))
{
- int g = 0, w = 0, v = 0;
+ int g, w, v;
- __asm__ volatile ("mov.u32 %0,%%ctaid.x;" : "=r" (g));
- __asm__ volatile ("mov.u32 %0,%%tid.y;" : "=r" (w));
- __asm__ volatile ("mov.u32 %0,%%tid.x;" : "=r" (v));
+ g = __builtin_goacc_parlevel_id (GOMP_DIM_GANG);
+ w = __builtin_goacc_parlevel_id (GOMP_DIM_WORKER);
+ v = __builtin_goacc_parlevel_id (GOMP_DIM_VECTOR);
val = (g << 16) | (w << 8) | v;
ondev = 1;
}
/* { dg-additional-options "-O2" } */
#include <stdio.h>
+#include <openacc.h>
+#include <gomp-constants.h>
#define N (32*32*32+17)
int main ()
{
int val = ix;
- if (__builtin_acc_on_device (5))
+ if (acc_on_device (acc_device_not_host))
{
- int g = 0, w = 0, v = 0;
+ int g, w, v;
- __asm__ volatile ("mov.u32 %0,%%ctaid.x;" : "=r" (g));
- __asm__ volatile ("mov.u32 %0,%%tid.y;" : "=r" (w));
- __asm__ volatile ("mov.u32 %0,%%tid.x;" : "=r" (v));
+ g = __builtin_goacc_parlevel_id (GOMP_DIM_GANG);
+ w = __builtin_goacc_parlevel_id (GOMP_DIM_WORKER);
+ v = __builtin_goacc_parlevel_id (GOMP_DIM_VECTOR);
val = (g << 16) | (w << 8) | v;
ondev = 1;
}
-/* 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>
+#include <gomp-constants.h>
#define N (32*32*32+17)
int main ()
#pragma acc loop vector
for (unsigned ix = 0; ix < N; ix++)
{
- if (__builtin_acc_on_device (5))
+ if (acc_on_device (acc_device_not_host))
{
- int g = 0, w = 0, v = 0;
+ int g, w, v;
- __asm__ volatile ("mov.u32 %0,%%ctaid.x;" : "=r" (g));
- __asm__ volatile ("mov.u32 %0,%%tid.y;" : "=r" (w));
- __asm__ volatile ("mov.u32 %0,%%tid.x;" : "=r" (v));
+ g = __builtin_goacc_parlevel_id (GOMP_DIM_GANG);
+ w = __builtin_goacc_parlevel_id (GOMP_DIM_WORKER);
+ v = __builtin_goacc_parlevel_id (GOMP_DIM_VECTOR);
ary[ix] = (g << 16) | (w << 8) | v;
ondev = 1;
}
-/* 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>
+#include <gomp-constants.h>
#define N (32*32*32+17)
int main ()
#pragma acc loop worker
for (unsigned ix = 0; ix < N; ix++)
{
- if (__builtin_acc_on_device (5))
+ if (acc_on_device (acc_device_not_host))
{
- int g = 0, w = 0, v = 0;
+ int g, w, v;
- __asm__ volatile ("mov.u32 %0,%%ctaid.x;" : "=r" (g));
- __asm__ volatile ("mov.u32 %0,%%tid.y;" : "=r" (w));
- __asm__ volatile ("mov.u32 %0,%%tid.x;" : "=r" (v));
+ g = __builtin_goacc_parlevel_id (GOMP_DIM_GANG);
+ w = __builtin_goacc_parlevel_id (GOMP_DIM_WORKER);
+ v = __builtin_goacc_parlevel_id (GOMP_DIM_VECTOR);
ary[ix] = (g << 16) | (w << 8) | v;
ondev = 1;
}
-/* 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>
+#include <gomp-constants.h>
#define N (32*32*32+17)
int main ()
#pragma acc loop worker vector
for (unsigned ix = 0; ix < N; ix++)
{
- if (__builtin_acc_on_device (5))
+ if (acc_on_device (acc_device_not_host))
{
- int g = 0, w = 0, v = 0;
+ int g, w, v;
- __asm__ volatile ("mov.u32 %0,%%ctaid.x;" : "=r" (g));
- __asm__ volatile ("mov.u32 %0,%%tid.y;" : "=r" (w));
- __asm__ volatile ("mov.u32 %0,%%tid.x;" : "=r" (v));
+ g = __builtin_goacc_parlevel_id (GOMP_DIM_GANG);
+ w = __builtin_goacc_parlevel_id (GOMP_DIM_WORKER);
+ v = __builtin_goacc_parlevel_id (GOMP_DIM_VECTOR);
ary[ix] = (g << 16) | (w << 8) | v;
ondev = 1;
}
#include <limits.h>
#include <openacc.h>
+#include <gomp-constants.h>
/* TODO: "(int) acc_device_*" casts because of the C++ acc_on_device wrapper
not behaving as expected for -O0. */
if (acc_on_device ((int) acc_device_host))
return 0;
else if (acc_on_device ((int) acc_device_nvidia))
- {
- unsigned int r;
- asm volatile ("mov.u32 %0,%%ctaid.x;" : "=r" (r));
- return r;
- }
+ return __builtin_goacc_parlevel_id (GOMP_DIM_GANG);
else
__builtin_abort ();
}
if (acc_on_device ((int) acc_device_host))
return 0;
else if (acc_on_device ((int) acc_device_nvidia))
- {
- unsigned int r;
- asm volatile ("mov.u32 %0,%%tid.y;" : "=r" (r));
- return r;
- }
+ return __builtin_goacc_parlevel_id (GOMP_DIM_WORKER);
else
__builtin_abort ();
}
if (acc_on_device ((int) acc_device_host))
return 0;
else if (acc_on_device ((int) acc_device_nvidia))
- {
- unsigned int r;
- asm volatile ("mov.u32 %0,%%tid.x;" : "=r" (r));
- return r;
- }
+ return __builtin_goacc_parlevel_id (GOMP_DIM_VECTOR);
else
__builtin_abort ();
}
-/* 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>
+#include <gomp-constants.h>
#define N (32*32*32+17)
#pragma acc loop gang
for (unsigned ix = 0; ix < N; ix++)
{
- if (__builtin_acc_on_device (5))
+ if (acc_on_device (acc_device_not_host))
{
- int g = 0, w = 0, v = 0;
+ int g, w, v;
- __asm__ volatile ("mov.u32 %0,%%ctaid.x;" : "=r" (g));
- __asm__ volatile ("mov.u32 %0,%%tid.y;" : "=r" (w));
- __asm__ volatile ("mov.u32 %0,%%tid.x;" : "=r" (v));
+ g = __builtin_goacc_parlevel_id (GOMP_DIM_GANG);
+ w = __builtin_goacc_parlevel_id (GOMP_DIM_WORKER);
+ v = __builtin_goacc_parlevel_id (GOMP_DIM_VECTOR);
ary[ix] = (g << 16) | (w << 8) | v;
}
else
#pragma acc parallel num_gangs(32) copy(ary) copy(ondev)
{
- ondev = __builtin_acc_on_device (5);
+ ondev = acc_on_device (acc_device_not_host);
gang (ary);
}
-/* 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>
+#include <gomp-constants.h>
#define N (32*32*32+17)
#pragma acc loop gang worker vector
for (unsigned ix = 0; ix < N; ix++)
{
- if (__builtin_acc_on_device (5))
+ if (acc_on_device (acc_device_not_host))
{
- int g = 0, w = 0, v = 0;
+ int g, w, v;
- __asm__ volatile ("mov.u32 %0,%%ctaid.x;" : "=r" (g));
- __asm__ volatile ("mov.u32 %0,%%tid.y;" : "=r" (w));
- __asm__ volatile ("mov.u32 %0,%%tid.x;" : "=r" (v));
+ g = __builtin_goacc_parlevel_id (GOMP_DIM_GANG);
+ w = __builtin_goacc_parlevel_id (GOMP_DIM_WORKER);
+ v = __builtin_goacc_parlevel_id (GOMP_DIM_VECTOR);
ary[ix] = (g << 16) | (w << 8) | v;
}
else
#pragma acc parallel num_gangs(32) num_workers(32) vector_length(32) copy(ary) copy(ondev)
{
- ondev = __builtin_acc_on_device (5);
+ ondev = acc_on_device (acc_device_not_host);
gang (ary);
}
-/* 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>
+#include <gomp-constants.h>
#define N (32*32*32+17)
#pragma acc loop vector
for (unsigned ix = 0; ix < N; ix++)
{
- if (__builtin_acc_on_device (5))
+ if (acc_on_device (acc_device_not_host))
{
- int g = 0, w = 0, v = 0;
+ int g, w, v;
- __asm__ volatile ("mov.u32 %0,%%ctaid.x;" : "=r" (g));
- __asm__ volatile ("mov.u32 %0,%%tid.y;" : "=r" (w));
- __asm__ volatile ("mov.u32 %0,%%tid.x;" : "=r" (v));
+ g = __builtin_goacc_parlevel_id (GOMP_DIM_GANG);
+ w = __builtin_goacc_parlevel_id (GOMP_DIM_WORKER);
+ v = __builtin_goacc_parlevel_id (GOMP_DIM_VECTOR);
ary[ix] = (g << 16) | (w << 8) | v;
}
else
#pragma acc parallel vector_length(32) copy(ary) copy(ondev)
{
- ondev = __builtin_acc_on_device (5);
+ ondev = acc_on_device (acc_device_not_host);
vector (ary);
}
-/* 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>
+#include <gomp-constants.h>
#define N (32*32*32+17)
#pragma acc loop worker
for (unsigned ix = 0; ix < N; ix++)
{
- if (__builtin_acc_on_device (5))
+ if (acc_on_device (acc_device_not_host))
{
- int g = 0, w = 0, v = 0;
+ int g, w, v;
- __asm__ volatile ("mov.u32 %0,%%ctaid.x;" : "=r" (g));
- __asm__ volatile ("mov.u32 %0,%%tid.y;" : "=r" (w));
- __asm__ volatile ("mov.u32 %0,%%tid.x;" : "=r" (v));
+ g = __builtin_goacc_parlevel_id (GOMP_DIM_GANG);
+ w = __builtin_goacc_parlevel_id (GOMP_DIM_WORKER);
+ v = __builtin_goacc_parlevel_id (GOMP_DIM_VECTOR);
ary[ix] = (g << 16) | (w << 8) | v;
}
else
#pragma acc parallel num_workers(32) vector_length(32) copy(ary) copy(ondev)
{
- ondev = __builtin_acc_on_device (5);
+ ondev = acc_on_device (acc_device_not_host);
worker (ary);
}
-/* 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>
+#include <gomp-constants.h>
#define N (32*32*32+17)
#pragma acc loop worker vector
for (unsigned ix = 0; ix < N; ix++)
{
- if (__builtin_acc_on_device (5))
+ if (acc_on_device (acc_device_not_host))
{
- int g = 0, w = 0, v = 0;
+ int g, w, v;
- __asm__ volatile ("mov.u32 %0,%%ctaid.x;" : "=r" (g));
- __asm__ volatile ("mov.u32 %0,%%tid.y;" : "=r" (w));
- __asm__ volatile ("mov.u32 %0,%%tid.x;" : "=r" (v));
+ g = __builtin_goacc_parlevel_id (GOMP_DIM_GANG);
+ w = __builtin_goacc_parlevel_id (GOMP_DIM_WORKER);
+ v = __builtin_goacc_parlevel_id (GOMP_DIM_VECTOR);
ary[ix] = (g << 16) | (w << 8) | v;
}
else
#pragma acc parallel num_workers(32) vector_length(32) copy(ary) copy(ondev)
{
- ondev = __builtin_acc_on_device (5);
+ ondev = acc_on_device (acc_device_not_host);
worker (ary);
}
-/* 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>
+#include <gomp-constants.h>
#define NUM_WORKERS 16
#define NUM_VECTORS 32
#define HEIGHT 32
#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))
+ (acc_on_device (acc_device_not_host) \
+ ? __builtin_goacc_parlevel_id (GOMP_DIM_WORKER) \
+ : (I % N))
#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))
+ (acc_on_device (acc_device_not_host) \
+ ? __builtin_goacc_parlevel_id (GOMP_DIM_VECTOR) \
+ : (I % N))
#pragma acc routine worker
void __attribute__ ((noinline))
-/* 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" } { "" } } */
-
/* { dg-additional-options "-fopenacc-dim=32" } */
#include <stdio.h>
#include <openacc.h>
+#include <gomp-constants.h>
static int check (const int *ary, int size, int gp, int wp, int vp)
{
{
int r = 0;
- if (acc_on_device (acc_device_nvidia))
+ if (acc_on_device (acc_device_not_host))
{
- int g = 0, w = 0, v = 0;
+ int g, w, v;
- __asm__ volatile ("mov.u32 %0,%%ctaid.x;" : "=r" (g));
- __asm__ volatile ("mov.u32 %0,%%tid.y;" : "=r" (w));
- __asm__ volatile ("mov.u32 %0,%%tid.x;" : "=r" (v));
+ g = __builtin_goacc_parlevel_id (GOMP_DIM_GANG);
+ w = __builtin_goacc_parlevel_id (GOMP_DIM_WORKER);
+ v = __builtin_goacc_parlevel_id (GOMP_DIM_VECTOR);
r = (g << 16) | (w << 8) | v;
}
return r;
projects / gcc.git / commitdiff