merge main into amd-staging

Author: ronlieb

offload/DeviceRTL/src/Kernel.cpp

@@ -30,7 +30,8 @@ enum OMPTgtExecModeFlags : unsigned char {
   OMP_TGT_EXEC_MODE_GENERIC = 1 << 0,
   OMP_TGT_EXEC_MODE_SPMD = 1 << 1,
   OMP_TGT_EXEC_MODE_GENERIC_SPMD =
-      OMP_TGT_EXEC_MODE_GENERIC | OMP_TGT_EXEC_MODE_SPMD
+      OMP_TGT_EXEC_MODE_GENERIC | OMP_TGT_EXEC_MODE_SPMD,
+  OMP_TGT_EXEC_MODE_SPMD_NO_LOOP = 1 << 2 | OMP_TGT_EXEC_MODE_SPMD
 };
 
 static void

offload/plugins-nextgen/common/include/PluginInterface.h

@@ -551,10 +551,10 @@ struct GenericKernelTy {
   /// The number of threads \p NumThreads can be adjusted by this method.
   /// \p IsNumThreadsFromUser is true is \p NumThreads is defined by user via
   /// thread_limit clause.
-  virtual uint32_t getNumBlocks(GenericDeviceTy &GenericDevice,
-                                uint32_t BlockLimitClause[3],
-                                uint64_t LoopTripCount, uint32_t &NumThreads,
-                                bool IsNumThreadsFromUser) const;
+  virtual
+  uint32_t getNumBlocks(GenericDeviceTy &GenericDevice,
+                        uint32_t BlockLimitClause[3], uint64_t LoopTripCount,
+                        uint32_t &NumThreads, bool IsNumThreadsFromUser) const;
 
   /// The kernel name.
   std::string Name;

offload/plugins-nextgen/common/src/PluginInterface.cpp

@@ -858,6 +858,10 @@ uint32_t GenericKernelTy::getNumBlocks(GenericDeviceTy &GenericDevice,
     return std::min(NumTeamsClause[0], GenericDevice.getBlockLimit());
   }
 
+  // Return the number of teams required to cover the loop iterations.
+  if (isNoLoopMode())
+    return LoopTripCount > 0 ? (((LoopTripCount - 1) / NumThreads) + 1) : 1;
+
   uint64_t DefaultNumBlocks = GenericDevice.getDefaultNumBlocks();
   uint64_t TripCountNumBlocks = std::numeric_limits<uint64_t>::max();
   if (LoopTripCount > 0) {