aboutsummaryrefslogtreecommitdiff
path: root/libgomp/plugin/plugin-nvptx.c
diff options
context:
space:
mode:
authorThomas Schwinge <thomas@codesourcery.com>2023-02-27 12:02:02 +0100
committerThomas Schwinge <thomas@codesourcery.com>2023-03-10 15:48:43 +0100
commit199867d07be65cb0227a318ebf42b8376ca09313 (patch)
tree9e1955c9f2beff346cf80dd2aec1daabebd406fd /libgomp/plugin/plugin-nvptx.c
parentb5037d4a073f2e4625afab5ec1f35624d9f9eba1 (diff)
downloadgcc-199867d07be65cb0227a318ebf42b8376ca09313.zip
gcc-199867d07be65cb0227a318ebf42b8376ca09313.tar.gz
gcc-199867d07be65cb0227a318ebf42b8376ca09313.tar.bz2
Simplify OpenACC 'no_create' clause implementation
For 'OFFSET_INLINED', 'gomp_map_val' does the right thing, and we may then simplify the device plugins accordingly. This is a follow-up to Subversion r279551 (Git commit a6163563f2ce502bd4ef444bd5de33570bb8eeb1) "Add OpenACC 2.6's no_create", Subversion r279622 (Git commit 5bcd470bf0749e1f56d05dd43aa9584ff2e3a090) "Use gomp_map_val for OpenACC host-to-device address translation". libgomp/ * target.c (gomp_map_vars_internal): Use 'OFFSET_INLINED' for 'GOMP_MAP_IF_PRESENT'. * plugin/plugin-gcn.c (gcn_exec, GOMP_OFFLOAD_openacc_exec) (GOMP_OFFLOAD_openacc_async_exec): Adjust. * plugin/plugin-nvptx.c (nvptx_exec, GOMP_OFFLOAD_openacc_exec) (GOMP_OFFLOAD_openacc_async_exec): Likewise. * testsuite/libgomp.oacc-c-c++-common/no_create-1.c: Add 'async' testing. * testsuite/libgomp.oacc-c-c++-common/no_create-2.c: Likewise.
Diffstat (limited to 'libgomp/plugin/plugin-nvptx.c')
-rw-r--r--libgomp/plugin/plugin-nvptx.c19
1 files changed, 9 insertions, 10 deletions
diff --git a/libgomp/plugin/plugin-nvptx.c b/libgomp/plugin/plugin-nvptx.c
index 1166807..13e3115 100644
--- a/libgomp/plugin/plugin-nvptx.c
+++ b/libgomp/plugin/plugin-nvptx.c
@@ -742,8 +742,7 @@ link_ptx (CUmodule *module, const struct targ_ptx_obj *ptx_objs,
}
static void
-nvptx_exec (void (*fn), size_t mapnum, void **hostaddrs, void **devaddrs,
- unsigned *dims, void *targ_mem_desc,
+nvptx_exec (void (*fn), size_t mapnum, unsigned *dims, void *targ_mem_desc,
CUdeviceptr dp, CUstream stream)
{
struct targ_fn_descriptor *targ_fn = (struct targ_fn_descriptor *) fn;
@@ -1530,7 +1529,8 @@ GOMP_OFFLOAD_free (int ord, void *ptr)
void
GOMP_OFFLOAD_openacc_exec (void (*fn) (void *), size_t mapnum,
- void **hostaddrs, void **devaddrs,
+ void **hostaddrs __attribute__((unused)),
+ void **devaddrs,
unsigned *dims, void *targ_mem_desc)
{
GOMP_PLUGIN_debug (0, " %s: prepare mappings\n", __FUNCTION__);
@@ -1549,7 +1549,7 @@ GOMP_OFFLOAD_openacc_exec (void (*fn) (void *), size_t mapnum,
size_t s = mapnum * sizeof (void *);
hp = alloca (s);
for (int i = 0; i < mapnum; i++)
- hp[i] = (devaddrs[i] ? devaddrs[i] : hostaddrs[i]);
+ hp[i] = devaddrs[i];
CUDA_CALL_ASSERT (cuMemAlloc, &dp, s);
if (profiling_p)
goacc_profiling_acc_ev_alloc (thr, (void *) dp, s);
@@ -1591,8 +1591,7 @@ GOMP_OFFLOAD_openacc_exec (void (*fn) (void *), size_t mapnum,
}
}
- nvptx_exec (fn, mapnum, hostaddrs, devaddrs, dims, targ_mem_desc,
- dp, NULL);
+ nvptx_exec (fn, mapnum, dims, targ_mem_desc, dp, NULL);
CUresult r = CUDA_CALL_NOCHECK (cuStreamSynchronize, NULL);
const char *maybe_abort_msg = "(perhaps abort was called)";
@@ -1617,7 +1616,8 @@ cuda_free_argmem (void *ptr)
void
GOMP_OFFLOAD_openacc_async_exec (void (*fn) (void *), size_t mapnum,
- void **hostaddrs, void **devaddrs,
+ void **hostaddrs __attribute__((unused)),
+ void **devaddrs,
unsigned *dims, void *targ_mem_desc,
struct goacc_asyncqueue *aq)
{
@@ -1639,7 +1639,7 @@ GOMP_OFFLOAD_openacc_async_exec (void (*fn) (void *), size_t mapnum,
block = (void **) GOMP_PLUGIN_malloc (2 * sizeof (void *) + s);
hp = block + 2;
for (int i = 0; i < mapnum; i++)
- hp[i] = (devaddrs[i] ? devaddrs[i] : hostaddrs[i]);
+ hp[i] = devaddrs[i];
CUDA_CALL_ASSERT (cuMemAlloc, &dp, s);
if (profiling_p)
goacc_profiling_acc_ev_alloc (thr, (void *) dp, s);
@@ -1688,8 +1688,7 @@ GOMP_OFFLOAD_openacc_async_exec (void (*fn) (void *), size_t mapnum,
}
}
- nvptx_exec (fn, mapnum, hostaddrs, devaddrs, dims, targ_mem_desc,
- dp, aq->cuda_stream);
+ nvptx_exec (fn, mapnum, dims, targ_mem_desc, dp, aq->cuda_stream);
if (mapnum > 0)
GOMP_OFFLOAD_openacc_async_queue_callback (aq, cuda_free_argmem, block);