[Openmp-commits] [openmp] r343476 - [libomptarget-nvptx] reduction: Determine if runtime uninitialized

[Jonas Hahnfeld via Openmp-commits]

Author: hahnfeld
Date: Mon Oct  1 07:14:26 2018
New Revision: 343476

URL: http://llvm.org/viewvc/llvm-project?rev=343476&view=rev
Log:
[libomptarget-nvptx] reduction: Determine if runtime uninitialized

Pass in the correct value of isRuntimeUninitialized() which solves
parallel reductions as reported on the mailing list.
For reference: r333285 did the same for loop scheduling.

Differential Revision: https://reviews.llvm.org/D52725

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

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=343476&r1=343475&r2=343476&view=diff
==============================================================================
--- openmp/trunk/libomptarget/deviceRTLs/nvptx/src/reduction.cu (original)
+++ openmp/trunk/libomptarget/deviceRTLs/nvptx/src/reduction.cu Mon Oct  1 07:14:26 2018
@@ -148,7 +148,7 @@ int32_t nvptx_parallel_reduce_nowait(int
                                      kmp_ShuffleReductFctPtr shflFct,
                                      kmp_InterWarpCopyFctPtr cpyFct,
                                      bool isSPMDExecutionMode,
-                                     bool isRuntimeUninitialized = false) {
+                                     bool isRuntimeUninitialized) {
   uint32_t BlockThreadId = GetLogicalThreadIdInBlock();
   uint32_t NumThreads = GetNumberOfOmpThreads(
       BlockThreadId, isSPMDExecutionMode, isRuntimeUninitialized);
@@ -240,9 +240,10 @@ EXTERN
 int32_t __kmpc_nvptx_parallel_reduce_nowait(
     int32_t global_tid, int32_t num_vars, size_t reduce_size, void *reduce_data,
     kmp_ShuffleReductFctPtr shflFct, kmp_InterWarpCopyFctPtr cpyFct) {
-  return nvptx_parallel_reduce_nowait(global_tid, num_vars, reduce_size,
-                                      reduce_data, shflFct, cpyFct,
-                                      /*isSPMDExecutionMode=*/isSPMDMode());
+  return nvptx_parallel_reduce_nowait(
+      global_tid, num_vars, reduce_size, reduce_data, shflFct, cpyFct,
+      /*isSPMDExecutionMode=*/isSPMDMode(),
+      /*isRuntimeUninitialized=*/isRuntimeUninitialized());
 }
 
 EXTERN
@@ -270,7 +271,7 @@ int32_t nvptx_teams_reduce_nowait(
     int32_t global_tid, int32_t num_vars, size_t reduce_size, void *reduce_data,
     kmp_ShuffleReductFctPtr shflFct, kmp_InterWarpCopyFctPtr cpyFct,
     kmp_CopyToScratchpadFctPtr scratchFct, kmp_LoadReduceFctPtr ldFct,
-    bool isSPMDExecutionMode, bool isRuntimeUninitialized = false) {
+    bool isSPMDExecutionMode, bool isRuntimeUninitialized) {
   uint32_t ThreadId = GetLogicalThreadIdInBlock();
   // In non-generic mode all workers participate in the teams reduction.
   // In generic mode only the team master participates in the teams
@@ -399,9 +400,10 @@ int32_t __kmpc_nvptx_teams_reduce_nowait
                                          kmp_InterWarpCopyFctPtr cpyFct,
                                          kmp_CopyToScratchpadFctPtr scratchFct,
                                          kmp_LoadReduceFctPtr ldFct) {
-  return nvptx_teams_reduce_nowait(global_tid, num_vars, reduce_size,
-                                   reduce_data, shflFct, cpyFct, scratchFct,
-                                   ldFct, /*isSPMDExecutionMode=*/isSPMDMode());
+  return nvptx_teams_reduce_nowait(
+      global_tid, num_vars, reduce_size, reduce_data, shflFct, cpyFct,
+      scratchFct, ldFct, /*isSPMDExecutionMode=*/isSPMDMode(),
+      /*isRuntimeUninitialized=*/isRuntimeUninitialized());
 }
 
 EXTERN