Hi,
this patch adds the following builtins in C/C++:
- __builtin_goacc_gang_id
- __builtin_goacc_worker_id
- __builtin_goacc_vector_id
- __builtin_goacc_gang_size
- __builtin_goacc_worker_size
- __builtin_goacc_vector_size
We have openacc C/C++ test-cases using the following nvptx idiom:
...
__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));
...
Typically these asm insns are guarded with acc_on_device
(acc_device_nvidia), and skipping -O0:
...
/* 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" } { "" } } */
...
This patch replaces those asm statements with the corresponding new
builtins, allowing the test-cases to work for all offloading targets,
not just nvptx, and also at O0. [ Note that the proposed fix for PR82391
at https://gcc.gnu.org/ml/gcc-patches/2017-12/msg01529.html also
addresses the O0 problem. ]
Bootstrapped and reg-tested on x86_64.
Build and reg-tested on x86_64 with nvptx accelerator.
OK for trunk?
Thanks,
- Tom
Add __builtin_goacc_{gang,worker,vector}_{id,size}
2017-12-30 Tom de Vries <[email protected]>
PR libgomp/82428
* builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
* omp-builtins.def (BUILT_IN_GOACC_GANG_ID, BUILT_IN_GOACC_GANG_SIZE)
(BUILT_IN_GOACC_WORKER_ID, BUILT_IN_GOACC_WORKER_SIZE
(BUILT_IN_GOACC_VECTOR_ID, BUILT_IN_GOACC_VECTOR_SIZE): New builtin.
* omp-offload.c (oacc_dim_call_1): Factor out of ...
(oacc_dim_call): ... here.
(fold_goacc_dim_id_size): New function.
(execute_oacc_device_lower): Use fold_goacc_dim_id_size.
* internal-fn.c (expand_GOACC_DIM_SIZE, expand_GOACC_DIM_POS): Handle
MEM_P lhs.
* omp-general.c (oacc_get_fn_dim_size): Handle routines.
* f95-lang.c (DEF_GOACC_BUILTIN_ONLY): Define.
* testsuite/libgomp.oacc-c-c++-common/gang-static-2.c: Use
__builtin_goacc_{gang,worker,vector}_{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.
---
gcc/builtins.def | 4 +
gcc/fortran/f95-lang.c | 4 +
gcc/internal-fn.c | 10 ++-
gcc/omp-builtins.def | 13 ++++
gcc/omp-general.c | 3 +
gcc/omp-offload.c | 90 ++++++++++++++++++++--
.../libgomp.oacc-c-c++-common/gang-static-2.c | 20 ++---
.../libgomp.oacc-c-c++-common/loop-auto-1.c | 17 ++--
.../libgomp.oacc-c-c++-common/loop-dim-default.c | 13 ++--
.../testsuite/libgomp.oacc-c-c++-common/loop-g-1.c | 16 ++--
.../testsuite/libgomp.oacc-c-c++-common/loop-g-2.c | 15 ++--
.../libgomp.oacc-c-c++-common/loop-gwv-1.c | 16 ++--
.../libgomp.oacc-c-c++-common/loop-red-g-1.c | 15 ++--
.../libgomp.oacc-c-c++-common/loop-red-gwv-1.c | 15 ++--
.../libgomp.oacc-c-c++-common/loop-red-v-1.c | 15 ++--
.../libgomp.oacc-c-c++-common/loop-red-v-2.c | 15 ++--
.../libgomp.oacc-c-c++-common/loop-red-w-1.c | 15 ++--
.../libgomp.oacc-c-c++-common/loop-red-w-2.c | 15 ++--
.../libgomp.oacc-c-c++-common/loop-red-wv-1.c | 11 +--
.../testsuite/libgomp.oacc-c-c++-common/loop-v-1.c | 15 ++--
.../testsuite/libgomp.oacc-c-c++-common/loop-w-1.c | 15 ++--
.../libgomp.oacc-c-c++-common/loop-wv-1.c | 15 ++--
.../libgomp.oacc-c-c++-common/parallel-dims.c | 18 +----
.../libgomp.oacc-c-c++-common/routine-g-1.c | 17 ++--
.../libgomp.oacc-c-c++-common/routine-gwv-1.c | 17 ++--
.../libgomp.oacc-c-c++-common/routine-v-1.c | 17 ++--
.../libgomp.oacc-c-c++-common/routine-w-1.c | 17 ++--
.../libgomp.oacc-c-c++-common/routine-wv-1.c | 17 ++--
.../libgomp.oacc-c-c++-common/routine-wv-2.c | 18 ++---
.../testsuite/libgomp.oacc-c-c++-common/tile-1.c | 14 ++--
30 files changed, 263 insertions(+), 239 deletions(-)
diff --git a/gcc/builtins.def b/gcc/builtins.def
index 671097e..263dfc7ea 100644
--- a/gcc/builtins.def
+++ b/gcc/builtins.def
@@ -214,6 +214,10 @@ along with GCC; see the file COPYING3. If not see
#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, \
diff --git a/gcc/fortran/f95-lang.c b/gcc/fortran/f95-lang.c
index dc9a1ae..60a28b8 100644
--- a/gcc/fortran/f95-lang.c
+++ b/gcc/fortran/f95-lang.c
@@ -1202,6 +1202,10 @@ gfc_init_builtin_functions (void)
#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"
diff --git a/gcc/internal-fn.c b/gcc/internal-fn.c
index 21e7b10..abb6780 100644
--- a/gcc/internal-fn.c
+++ b/gcc/internal-fn.c
@@ -2561,7 +2561,10 @@ expand_GOACC_DIM_SIZE (internal_fn, gcall *stmt)
{
rtx dim = expand_expr (gimple_call_arg (stmt, 0), NULL_RTX,
VOIDmode, EXPAND_NORMAL);
- emit_insn (targetm.gen_oacc_dim_size (target, dim));
+ rtx reg = MEM_P (target) ? gen_reg_rtx (GET_MODE (target)) : target;
+ emit_insn (targetm.gen_oacc_dim_size (reg, dim));
+ if (reg != target)
+ emit_move_insn (target, reg);
}
else
emit_move_insn (target, GEN_INT (1));
@@ -2582,7 +2585,10 @@ expand_GOACC_DIM_POS (internal_fn, gcall *stmt)
{
rtx dim = expand_expr (gimple_call_arg (stmt, 0), NULL_RTX,
VOIDmode, EXPAND_NORMAL);
- emit_insn (targetm.gen_oacc_dim_pos (target, dim));
+ rtx reg = MEM_P (target) ? gen_reg_rtx (GET_MODE (target)) : target;
+ emit_insn (targetm.gen_oacc_dim_pos (reg, dim));
+ if (reg != target)
+ emit_move_insn (target, reg);
}
else
emit_move_insn (target, const0_rtx);
diff --git a/gcc/omp-builtins.def b/gcc/omp-builtins.def
index 69b73f4..6675f82 100644
--- a/gcc/omp-builtins.def
+++ b/gcc/omp-builtins.def
@@ -51,6 +51,19 @@ DEF_GOACC_BUILTIN (BUILT_IN_GOACC_WAIT, "GOACC_wait",
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_GANG_ID, "goacc_gang_id",
+ BT_FN_INT, ATTR_NOTHROW_LEAF_LIST)
+DEF_GOACC_BUILTIN_ONLY (BUILT_IN_GOACC_GANG_SIZE, "goacc_gang_size",
+ BT_FN_INT, ATTR_NOTHROW_LEAF_LIST)
+DEF_GOACC_BUILTIN_ONLY (BUILT_IN_GOACC_WORKER_ID, "goacc_worker_id",
+ BT_FN_INT, ATTR_NOTHROW_LEAF_LIST)
+DEF_GOACC_BUILTIN_ONLY (BUILT_IN_GOACC_WORKER_SIZE, "goacc_worker_size",
+ BT_FN_INT, ATTR_NOTHROW_LEAF_LIST)
+DEF_GOACC_BUILTIN_ONLY (BUILT_IN_GOACC_VECTOR_ID, "goacc_vector_id",
+ BT_FN_INT, ATTR_NOTHROW_LEAF_LIST)
+DEF_GOACC_BUILTIN_ONLY (BUILT_IN_GOACC_VECTOR_SIZE, "goacc_vector_size",
+ BT_FN_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",
diff --git a/gcc/omp-general.c b/gcc/omp-general.c
index 0f53661..37262ed 100644
--- a/gcc/omp-general.c
+++ b/gcc/omp-general.c
@@ -615,6 +615,9 @@ oacc_get_fn_attrib (tree fn)
int
oacc_get_fn_dim_size (tree fn, int axis)
{
+ if (lookup_attribute ("omp declare target", DECL_ATTRIBUTES (fn)))
+ return 0;
+
tree attrs = oacc_get_fn_attrib (fn);
gcc_assert (axis < GOMP_DIM_MAX);
diff --git a/gcc/omp-offload.c b/gcc/omp-offload.c
index 9d5b8be..8ef3c89 100644
--- a/gcc/omp-offload.c
+++ b/gcc/omp-offload.c
@@ -226,20 +226,28 @@ omp_finish_file (void)
}
/* Call dim_pos (POS == true) or dim_size (POS == false) builtins for
- axis DIM. Return a tmp var holding the result. */
+ axis DIM. Set the call result to lhs. */
-static tree
-oacc_dim_call (bool pos, int dim, gimple_seq *seq)
+static gimple *
+oacc_dim_call_1 (bool pos, int dim, tree lhs)
{
tree arg = build_int_cst (unsigned_type_node, dim);
- tree size = create_tmp_var (integer_type_node);
enum internal_fn fn = pos ? IFN_GOACC_DIM_POS : IFN_GOACC_DIM_SIZE;
gimple *call = gimple_build_call_internal (fn, 1, arg);
+ gimple_call_set_lhs (call, lhs);
+ return call;
+}
- gimple_call_set_lhs (call, size);
- gimple_seq_add_stmt (seq, call);
+/* Call dim_pos (POS == true) or dim_size (POS == false) builtins for
+ axis DIM. Return a tmp var holding the result. */
- return size;
+static tree
+oacc_dim_call (bool pos, int dim, gimple_seq *seq)
+{
+ tree lhs = create_tmp_var (integer_type_node);
+ gimple *call = oacc_dim_call_1 (pos, dim, lhs);
+ gimple_seq_add_stmt (seq, call);
+ return lhs;
}
/* Find the number of threads (POS = false), or thread number (POS =
@@ -1451,6 +1459,52 @@ default_goacc_reduction (gcall *call)
gsi_replace_with_seq (&gsi, seq, true);
}
+/* Fold __builtin_goacc_{gang,worker,vector}_{id,size}. */
+
+static gimple *
+fold_goacc_dim_id_size (gcall *call)
+{
+ tree fndecl = gimple_call_fndecl (call);
+
+ int dim;
+ switch (DECL_FUNCTION_CODE (fndecl))
+ {
+ case BUILT_IN_GOACC_GANG_ID:
+ case BUILT_IN_GOACC_GANG_SIZE:
+ dim = GOMP_DIM_GANG;
+ break;
+ case BUILT_IN_GOACC_WORKER_ID:
+ case BUILT_IN_GOACC_WORKER_SIZE:
+ dim = GOMP_DIM_WORKER;
+ break;
+ case BUILT_IN_GOACC_VECTOR_ID:
+ case BUILT_IN_GOACC_VECTOR_SIZE:
+ dim = GOMP_DIM_VECTOR;
+ break;
+ default:
+ gcc_unreachable ();
+ }
+
+ bool pos;
+ switch (DECL_FUNCTION_CODE (fndecl))
+ {
+ case BUILT_IN_GOACC_GANG_ID:
+ case BUILT_IN_GOACC_WORKER_ID:
+ case BUILT_IN_GOACC_VECTOR_ID:
+ pos = true;
+ break;
+ case BUILT_IN_GOACC_GANG_SIZE:
+ case BUILT_IN_GOACC_WORKER_SIZE:
+ case BUILT_IN_GOACC_VECTOR_SIZE:
+ pos = false;
+ break;
+ default:
+ gcc_unreachable ();
+ }
+
+ return oacc_dim_call_1 (pos, dim, gimple_call_lhs (call));
+}
+
/* Main entry point for oacc transformations which run on the device
compiler after LTO, so we know what the target device is at this
point (including the host fallback). */
@@ -1552,6 +1606,28 @@ execute_oacc_device_lower ()
}
gcall *call = as_a <gcall *> (stmt);
+ tree fndecl;
+ if (gimple_call_builtin_p (call)
+ && (fndecl = gimple_call_fndecl (stmt)) != NULL_TREE
+ && DECL_BUILT_IN_CLASS (fndecl) == BUILT_IN_NORMAL)
+ switch (DECL_FUNCTION_CODE (fndecl))
+ {
+ case BUILT_IN_GOACC_GANG_ID:
+ case BUILT_IN_GOACC_WORKER_ID:
+ case BUILT_IN_GOACC_VECTOR_ID:
+ case BUILT_IN_GOACC_GANG_SIZE:
+ case BUILT_IN_GOACC_WORKER_SIZE:
+ case BUILT_IN_GOACC_VECTOR_SIZE:
+ {
+ gimple *repl = fold_goacc_dim_id_size (call);
+ gsi_replace (&gsi, repl, false);
+ gsi_next (&gsi);
+ continue;
+ }
+ default:
+ break;
+ }
+
if (!gimple_call_internal_p (call))
{
gsi_next (&gsi);
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/gang-static-2.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/gang-static-2.c
index 6de739a..99a306a 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/gang-static-2.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/gang-static-2.c
@@ -1,25 +1,22 @@
-/* { 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>
#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_gang_id () \
+ : (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++)
@@ -32,6 +29,9 @@ test_nonstatic(int *a, int gangs)
{
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);
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-auto-1.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-auto-1.c
index 863b6b3..8920ec2 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-auto-1.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-auto-1.c
@@ -1,7 +1,3 @@
-/* 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>
@@ -79,15 +75,12 @@ static int __attribute__((noinline)) place ()
{
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_gang_id ();
+ w = __builtin_goacc_worker_id ();
+ v = __builtin_goacc_vector_id ();
+ 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;
}
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-dim-default.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-dim-default.c
index e2b08c3..565e1e4 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-dim-default.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-dim-default.c
@@ -1,6 +1,3 @@
-/* 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>
@@ -13,13 +10,13 @@ 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_gang_id ();
+ w = __builtin_goacc_worker_id ();
+ v = __builtin_goacc_vector_id ();
res = (1 << 24) | (g << 16) | (w << 8) | v;
}
return res;
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-g-1.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-g-1.c
index ae1d588..1ceee00 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-g-1.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-g-1.c
@@ -1,8 +1,5 @@
-/* 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 N (32*32*32+17)
int main ()
@@ -20,13 +17,12 @@ 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_gang_id ();
+ w = __builtin_goacc_worker_id ();
+ v = __builtin_goacc_vector_id ();
ary[ix] = (g << 16) | (w << 8) | v;
ondev = 1;
}
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-g-2.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-g-2.c
index c06d861..f23c88a 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-g-2.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-g-2.c
@@ -1,8 +1,5 @@
-/* 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 N (32*32*32+17)
int main ()
@@ -20,13 +17,13 @@ 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_gang_id ();
+ w = __builtin_goacc_worker_id ();
+ v = __builtin_goacc_vector_id ();
ary[ix] = (g << 16) | (w << 8) | v;
ondev = 1;
}
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-gwv-1.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-gwv-1.c
index 42b612a..b026c8f 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-gwv-1.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-gwv-1.c
@@ -1,8 +1,5 @@
-/* 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 N (32*32*32+17)
int main ()
@@ -20,13 +17,14 @@ 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_gang_id ();
+ w = __builtin_goacc_worker_id ();
+ v = __builtin_goacc_vector_id ();
- __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;
}
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-g-1.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-g-1.c
index 929e01c..70a0376 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-g-1.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-g-1.c
@@ -1,8 +1,5 @@
-/* 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 N (32*32*32+17)
int main ()
@@ -18,13 +15,13 @@ 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_gang_id ();
+ w = __builtin_goacc_worker_id ();
+ v = __builtin_goacc_vector_id ();
val = (g << 16) | (w << 8) | v;
ondev = 1;
}
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-gwv-1.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-gwv-1.c
index 4ae4b7c..3535a2a 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-gwv-1.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-gwv-1.c
@@ -1,8 +1,5 @@
-/* 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 N (32*32*32+17)
int main ()
@@ -18,13 +15,13 @@ 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_gang_id ();
+ w = __builtin_goacc_worker_id ();
+ v = __builtin_goacc_vector_id ();
val = (g << 16) | (w << 8) | v;
ondev = 1;
}
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-v-1.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-v-1.c
index 0556455..804a8ce 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-v-1.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-v-1.c
@@ -1,8 +1,5 @@
-/* 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 N (32*32*32+17)
@@ -19,13 +16,13 @@ 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_gang_id ();
+ w = __builtin_goacc_worker_id ();
+ v = __builtin_goacc_vector_id ();
val = (g << 16) | (w << 8) | v;
ondev = 1;
}
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-v-2.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-v-2.c
index 16d8f9f..56d8c8c 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-v-2.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-v-2.c
@@ -1,8 +1,5 @@
-/* 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 N (32*32*32+17)
@@ -21,13 +18,13 @@ 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_gang_id ();
+ w = __builtin_goacc_worker_id ();
+ v = __builtin_goacc_vector_id ();
val = (g << 16) | (w << 8) | v;
ondev = 1;
}
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-w-1.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-w-1.c
index 19021d9..bbf4110 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-w-1.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-w-1.c
@@ -1,8 +1,5 @@
-/* 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 N (32*32*32+17)
int main ()
@@ -18,13 +15,13 @@ 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_gang_id ();
+ w = __builtin_goacc_worker_id ();
+ v = __builtin_goacc_vector_id ();
val = (g << 16) | (w << 8) | v;
ondev = 1;
}
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-w-2.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-w-2.c
index f0c9d81..d06485b 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-w-2.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-w-2.c
@@ -1,8 +1,5 @@
-/* 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 N (32*32*32+17)
int main ()
@@ -20,13 +17,13 @@ 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_gang_id ();
+ w = __builtin_goacc_worker_id ();
+ v = __builtin_goacc_vector_id ();
val = (g << 16) | (w << 8) | v;
ondev = 1;
}
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-wv-1.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-wv-1.c
index 0fec2dc..1d9fc2d 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-wv-1.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-wv-1.c
@@ -2,6 +2,7 @@
/* { dg-additional-options "-O2" } */
#include <stdio.h>
+#include <openacc.h>
#define N (32*32*32+17)
int main ()
@@ -17,13 +18,13 @@ 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_gang_id ();
+ w = __builtin_goacc_worker_id ();
+ v = __builtin_goacc_vector_id ();
val = (g << 16) | (w << 8) | v;
ondev = 1;
}
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-v-1.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-v-1.c
index 2974807..d9262be 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-v-1.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-v-1.c
@@ -1,8 +1,5 @@
-/* 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 N (32*32*32+17)
int main ()
@@ -20,13 +17,13 @@ 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_gang_id ();
+ w = __builtin_goacc_worker_id ();
+ v = __builtin_goacc_vector_id ();
ary[ix] = (g << 16) | (w << 8) | v;
ondev = 1;
}
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-w-1.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-w-1.c
index 33b6eae..07b77eb 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-w-1.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-w-1.c
@@ -1,8 +1,5 @@
-/* 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 N (32*32*32+17)
int main ()
@@ -20,13 +17,13 @@ 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_gang_id ();
+ w = __builtin_goacc_worker_id ();
+ v = __builtin_goacc_vector_id ();
ary[ix] = (g << 16) | (w << 8) | v;
ondev = 1;
}
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-wv-1.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-wv-1.c
index 578cfad..d4e048d 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-wv-1.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-wv-1.c
@@ -1,8 +1,5 @@
-/* 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 N (32*32*32+17)
int main ()
@@ -20,13 +17,13 @@ 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_gang_id ();
+ w = __builtin_goacc_worker_id ();
+ v = __builtin_goacc_vector_id ();
ary[ix] = (g << 16) | (w << 8) | v;
ondev = 1;
}
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/parallel-dims.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/parallel-dims.c
index 8308f7c..0b19fde 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/parallel-dims.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/parallel-dims.c
@@ -12,11 +12,7 @@ static unsigned int __attribute__ ((optimize ("O2"))) acc_gang ()
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_gang_id ();
else
__builtin_abort ();
}
@@ -27,11 +23,7 @@ static unsigned int __attribute__ ((optimize ("O2"))) acc_worker ()
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_worker_id ();
else
__builtin_abort ();
}
@@ -42,11 +34,7 @@ static unsigned int __attribute__ ((optimize ("O2"))) acc_vector ()
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_vector_id ();
else
__builtin_abort ();
}
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-g-1.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-g-1.c
index b6ab713..7055cd3 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-g-1.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-g-1.c
@@ -1,8 +1,5 @@
-/* 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 N (32*32*32+17)
@@ -12,13 +9,13 @@ void __attribute__ ((noinline)) gang (int ary[N])
#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_gang_id ();
+ w = __builtin_goacc_worker_id ();
+ v = __builtin_goacc_vector_id ();
ary[ix] = (g << 16) | (w << 8) | v;
}
else
@@ -38,7 +35,7 @@ int main ()
#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);
}
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-gwv-1.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-gwv-1.c
index ace2f49..b9d36a8 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-gwv-1.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-gwv-1.c
@@ -1,8 +1,5 @@
-/* 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 N (32*32*32+17)
@@ -12,13 +9,13 @@ void __attribute__ ((noinline)) gang (int ary[N])
#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_gang_id ();
+ w = __builtin_goacc_worker_id ();
+ v = __builtin_goacc_vector_id ();
ary[ix] = (g << 16) | (w << 8) | v;
}
else
@@ -38,7 +35,7 @@ int main ()
#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);
}
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-v-1.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-v-1.c
index 2503e8d..8c553d5 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-v-1.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-v-1.c
@@ -1,8 +1,5 @@
-/* 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 N (32*32*32+17)
@@ -12,13 +9,13 @@ void __attribute__ ((noinline)) vector (int ary[N])
#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_gang_id ();
+ w = __builtin_goacc_worker_id ();
+ v = __builtin_goacc_vector_id ();
ary[ix] = (g << 16) | (w << 8) | v;
}
else
@@ -38,7 +35,7 @@ int main ()
#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);
}
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-w-1.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-w-1.c
index 80cd462..7847336 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-w-1.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-w-1.c
@@ -1,8 +1,5 @@
-/* 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 N (32*32*32+17)
@@ -12,13 +9,13 @@ void __attribute__ ((noinline)) worker (int ary[N])
#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_gang_id ();
+ w = __builtin_goacc_worker_id ();
+ v = __builtin_goacc_vector_id ();
ary[ix] = (g << 16) | (w << 8) | v;
}
else
@@ -38,7 +35,7 @@ int main ()
#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);
}
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-wv-1.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-wv-1.c
index 5e45fad..748978d 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-wv-1.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-wv-1.c
@@ -1,8 +1,5 @@
-/* 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 N (32*32*32+17)
@@ -12,13 +9,13 @@ void __attribute__ ((noinline)) worker (int ary[N])
#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_gang_id ();
+ w = __builtin_goacc_worker_id ();
+ v = __builtin_goacc_vector_id ();
ary[ix] = (g << 16) | (w << 8) | v;
}
else
@@ -38,7 +35,7 @@ int main ()
#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);
}
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-wv-2.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-wv-2.c
index b5cbc90..8a1f60f 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-wv-2.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-wv-2.c
@@ -1,7 +1,3 @@
-/* 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>
@@ -11,15 +7,13 @@
#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_worker_id () \
+ : (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_vector_id () \
+ : (I % N))
#pragma acc routine worker
void __attribute__ ((noinline))
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/tile-1.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/tile-1.c
index 8dcb956..4669291 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/tile-1.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/tile-1.c
@@ -1,7 +1,3 @@
-/* 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>
@@ -79,13 +75,13 @@ static int __attribute__((noinline)) place ()
{
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_gang_id ();
+ w = __builtin_goacc_worker_id ();
+ v = __builtin_goacc_vector_id ();
r = (g << 16) | (w << 8) | v;
}
return r;
