[Openmp-commits] [PATCH] D119311: [OpenMP][CUDA] Refine the logic to determine grid size

[Shilei Tian via Phabricator via Openmp-commits]

tianshilei1992 created this revision.
tianshilei1992 added reviewers: jhuber6, JonChesterfield, jdoerfert.
Herald added subscribers: carlosgalvezp, guansong, yaxunl.
tianshilei1992 requested review of this revision.
Herald added subscribers: openmp-commits, sstefan1.
Herald added a project: OpenMP.

This patch refines the logic to determine grid size as previous method
can skip the check of whether `CudaBlocksPerGrid` could be greater than the actual
hardware limit.


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D119311

Files:
  openmp/libomptarget/plugins/cuda/src/rtl.cpp


Index: openmp/libomptarget/plugins/cuda/src/rtl.cpp
===================================================================
--- openmp/libomptarget/plugins/cuda/src/rtl.cpp
+++ openmp/libomptarget/plugins/cuda/src/rtl.cpp
@@ -1170,15 +1170,17 @@
         DP("Using default number of teams %d\n", DeviceData[DeviceId].NumTeams);
         CudaBlocksPerGrid = DeviceData[DeviceId].NumTeams;
       }
-    } else if (TeamNum > DeviceData[DeviceId].BlocksPerGrid) {
-      DP("Capping number of teams to team limit %d\n",
-         DeviceData[DeviceId].BlocksPerGrid);
-      CudaBlocksPerGrid = DeviceData[DeviceId].BlocksPerGrid;
     } else {
       DP("Using requested number of teams %d\n", TeamNum);
       CudaBlocksPerGrid = TeamNum;
     }
 
+    if (CudaBlocksPerGrid > DeviceData[DeviceId].BlocksPerGrid) {
+      DP("Capping number of teams to team limit %d\n",
+         DeviceData[DeviceId].BlocksPerGrid);
+      CudaBlocksPerGrid = DeviceData[DeviceId].BlocksPerGrid;
+    }
+
     INFO(OMP_INFOTYPE_PLUGIN_KERNEL, DeviceId,
          "Launching kernel %s with %d blocks and %d threads in %s mode\n",
          (getOffloadEntry(DeviceId, TgtEntryPtr))


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D119311.407046.patch
Type: text/x-patch
Size: 1159 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/openmp-commits/attachments/20220209/f71e8dc7/attachment.bin>