[Openmp-commits] [openmp] r347710 - [OPENMP][NVPTX]Basic support for reductions across the teams.

[Alexey Bataev via Openmp-commits]

Author: abataev
Date: Tue Nov 27 13:06:09 2018
New Revision: 347710

URL: http://llvm.org/viewvc/llvm-project?rev=347710&view=rev
Log:
[OPENMP][NVPTX]Basic support for reductions across the teams.

Summary:
Added functions __kmpc_nvptx_teams_reduce_nowait_simple and
__kmpc_nvptx_teams_end_reduce_nowait_simple to implement basic support
for reductions across the teams.

Reviewers: gtbercea, kkwli0

Subscribers: guansong, jfb, caomhin, openmp-commits

Differential Revision: https://reviews.llvm.org/D54967

Modified:
    openmp/trunk/libomptarget/deviceRTLs/nvptx/src/interface.h
    openmp/trunk/libomptarget/deviceRTLs/nvptx/src/reduction.cu

Modified: openmp/trunk/libomptarget/deviceRTLs/nvptx/src/interface.h
URL: http://llvm.org/viewvc/llvm-project/openmp/trunk/libomptarget/deviceRTLs/nvptx/src/interface.h?rev=347710&r1=347709&r2=347710&view=diff
==============================================================================
--- openmp/trunk/libomptarget/deviceRTLs/nvptx/src/interface.h (original)
+++ openmp/trunk/libomptarget/deviceRTLs/nvptx/src/interface.h Tue Nov 27 13:06:09 2018
@@ -419,6 +419,12 @@ EXTERN int32_t __kmpc_nvptx_teams_reduce
     int32_t global_tid, int32_t num_vars, size_t reduce_size, void *reduce_data,
     kmp_ShuffleReductFctPtr shflFct, kmp_InterWarpCopyFctPtr cpyFct,
     kmp_CopyToScratchpadFctPtr sratchFct, kmp_LoadReduceFctPtr ldFct);
+EXTERN int32_t __kmpc_nvptx_teams_reduce_nowait_simple(kmp_Ident *loc,
+                                                       int32_t global_tid,
+                                                       kmp_CriticalName *crit);
+EXTERN void __kmpc_nvptx_teams_end_reduce_nowait_simple(kmp_Ident *loc,
+                                                        int32_t global_tid,
+                                                        kmp_CriticalName *crit);
 EXTERN int32_t __kmpc_shuffle_int32(int32_t val, int16_t delta, int16_t size);
 EXTERN int64_t __kmpc_shuffle_int64(int64_t val, int16_t delta, int16_t size);
 

Modified: openmp/trunk/libomptarget/deviceRTLs/nvptx/src/reduction.cu
URL: http://llvm.org/viewvc/llvm-project/openmp/trunk/libomptarget/deviceRTLs/nvptx/src/reduction.cu?rev=347710&r1=347709&r2=347710&view=diff
==============================================================================
--- openmp/trunk/libomptarget/deviceRTLs/nvptx/src/reduction.cu (original)
+++ openmp/trunk/libomptarget/deviceRTLs/nvptx/src/reduction.cu Tue Nov 27 13:06:09 2018
@@ -430,3 +430,21 @@ int32_t __kmpc_nvptx_teams_reduce_nowait
                                    /*isSPMDExecutionMode=*/false,
                                    /*isRuntimeUninitialized=*/true);
 }
+
+EXTERN int32_t __kmpc_nvptx_teams_reduce_nowait_simple(kmp_Ident *loc,
+                                                       int32_t global_tid,
+                                                       kmp_CriticalName *crit) {
+  if (checkSPMDMode(loc) && GetThreadIdInBlock() != 0)
+    return 0;
+  // The master thread of the team actually does the reduction.
+  while (atomicCAS((uint32_t *)crit, 0, 1))
+    ;
+  return 1;
+}
+
+EXTERN void
+__kmpc_nvptx_teams_end_reduce_nowait_simple(kmp_Ident *loc, int32_t global_tid,
+                                            kmp_CriticalName *crit) {
+  (void)atomicExch((uint32_t *)crit, 0);
+}
+