Merge branch 'task/rhornung67/split-azure-ghactions' of github.com:LL…

…NL/RAJA into task/rhornung67/split-azure-ghactions
LLNL · Aug 14, 2024 · 8e550ca · 8e550ca
2 parents c99b902 + 3e422bc
commit 8e550ca
Show file tree

Hide file tree

Showing 12 changed files with 366 additions and 6 deletions.
diff --git a/CMakeLists.txt b/CMakeLists.txt
@@ -122,7 +122,7 @@ if (RAJA_ENABLE_CUDA)
     message(STATUS "CUDA compute architecture set to RAJA default 35 since it was not specified")
     set(CMAKE_CUDA_ARCHITECTURES "35" CACHE STRING "Set CMAKE_CUDA_ARCHITECTURES to RAJA minimum supported" FORCE)
   endif()
-  message(STATUS "CMAKE_CUDA_ARCHITECTURES set to ${CMAKE_CUDA_ARCHITECTURES}") 
+  message(STATUS "CMAKE_CUDA_ARCHITECTURES set to ${CMAKE_CUDA_ARCHITECTURES}")
   if ( (CMAKE_CXX_COMPILER_ID MATCHES GNU) AND (CMAKE_SYSTEM_PROCESSOR MATCHES ppc64le) )
     if (CMAKE_CXX_COMPILER_VERSION VERSION_GREATER 8.0)
       set (CMAKE_CUDA_FLAGS "${CMAKE_CUDA_FLAGS} -Xcompiler -mno-float128")
@@ -221,6 +221,7 @@ if (RAJA_ENABLE_SYCL)
     sycl)
 endif ()
 
+
 message(STATUS "Desul Atomics support is ${RAJA_ENABLE_DESUL_ATOMICS}")
 if (RAJA_ENABLE_DESUL_ATOMICS)
   # NOTE: ENABLE_OPENMP in Desul implies OpenMP OFFLOAD

diff --git a/benchmark/CMakeLists.txt b/benchmark/CMakeLists.txt
@@ -11,6 +11,10 @@ if (RAJA_ENABLE_CUDA)
     SOURCES host-device-lambda-benchmark.cpp)
 endif()
 
+raja_add_benchmark(
+  NAME benchmark-atomic
+  SOURCES benchmark-atomic.cpp)
+
 raja_add_benchmark(
   NAME ltimes
   SOURCES ltimes.cpp)
diff --git a/benchmark/benchmark-atomic.cpp b/benchmark/benchmark-atomic.cpp
diff --git a/docs/sphinx/user_guide/feature/policies.rst b/docs/sphinx/user_guide/feature/policies.rst
@@ -593,7 +593,7 @@ GPU Policies for SYCL
 	  x dimension for CUDA and HIP.
 
           Similarly, ``RAJA::kernel`` uses a three-dimensional work-group
-          configuration. SYCL imension 2 always exists and should be used as
+          configuration. SYCL dimension 2 always exists and should be used as
           one would use the x dimension in CUDA and HIP.  
 
 ======================================== ============= ==============================

diff --git a/examples/forall_multi-reductions.cpp b/examples/forall_multi-reductions.cpp
@@ -11,6 +11,8 @@
 
 #include "RAJA/RAJA.hpp"
 
+#include "camp/tuple.hpp"
+
 /*
  *  MultiReduction Example using RAJA forall
  *

diff --git a/test/functional/forall/reduce-basic/tests/test-forall-basic-expt-ReduceBitAnd.hpp b/test/functional/forall/reduce-basic/tests/test-forall-basic-expt-ReduceBitAnd.hpp
@@ -49,10 +49,10 @@ void ForallReduceBitAndBasicTestImpl(const SEG_TYPE& seg,
 
   ASSERT_EQ(static_cast<DATA_TYPE>(simpand.get()), 5);
 
-  
-  // 
+
+  //
   // And now a randomized test that pushes zeros around
-  // 
+  //
 
   const int modval = 100;
 
@@ -72,6 +72,7 @@ void ForallReduceBitAndBasicTestImpl(const SEG_TYPE& seg,
   RAJA::forall<EXEC_POLICY>(seg,
     RAJA::expt::Reduce<RAJA::operators::bit_and>(&redand),
     RAJA::expt::Reduce<RAJA::operators::bit_and>(&redand2),
+    RAJA::expt::KernelName("RAJA Reduce BitAnd"),
     [=] RAJA_HOST_DEVICE(IDX_TYPE idx, DATA_TYPE &r1, DATA_TYPE &r2) {
       r1 &= working_array[idx];
       r2 &= working_array[idx];
@@ -92,7 +93,7 @@ void ForallReduceBitAndBasicTestImpl(const SEG_TYPE& seg,
   }
 
   ASSERT_EQ(static_cast<DATA_TYPE>(redand), ref_and);
-   
+
 
   deallocateForallTestData<DATA_TYPE>(working_res,
                                       working_array,

diff --git a/test/functional/forall/reduce-basic/tests/test-forall-basic-expt-ReduceBitOr.hpp b/test/functional/forall/reduce-basic/tests/test-forall-basic-expt-ReduceBitOr.hpp
@@ -72,6 +72,7 @@ void ForallReduceBitOrBasicTestImpl(const SEG_TYPE& seg,
   RAJA::forall<EXEC_POLICY>(seg,
     RAJA::expt::Reduce<RAJA::operators::bit_or>(&redor),
     RAJA::expt::Reduce<RAJA::operators::bit_or>(&redor2),
+    RAJA::expt::KernelName("RAJA Reduce BitOr"),
     [=] RAJA_HOST_DEVICE(IDX_TYPE idx, DATA_TYPE &r1, DATA_TYPE &r2) {
       r1 |= working_array[idx];
       r2 |= working_array[idx];

diff --git a/test/functional/forall/reduce-basic/tests/test-forall-basic-expt-ReduceMax.hpp b/test/functional/forall/reduce-basic/tests/test-forall-basic-expt-ReduceMax.hpp
@@ -54,6 +54,7 @@ void ForallReduceMaxBasicTestImpl(const SEG_TYPE& seg,
   RAJA::forall<EXEC_POLICY>(seg, 
     RAJA::expt::Reduce<RAJA::operators::maximum>(&maxinit),
     RAJA::expt::Reduce<RAJA::operators::maximum>(&max),
+    RAJA::expt::KernelName("RAJA Reduce Max"),
     [=] RAJA_HOST_DEVICE(IDX_TYPE idx, DATA_TYPE &mi, DATA_TYPE &m) {
       mi = RAJA_MAX(working_array[idx], mi);
       m  = RAJA_MAX(working_array[idx], m);

diff --git a/test/functional/forall/reduce-basic/tests/test-forall-basic-expt-ReduceMaxLoc.hpp b/test/functional/forall/reduce-basic/tests/test-forall-basic-expt-ReduceMaxLoc.hpp
@@ -64,6 +64,7 @@ void ForallReduceMaxLocBasicTestImpl(const SEG_TYPE& seg,
   RAJA::forall<EXEC_POLICY>(seg, 
     RAJA::expt::Reduce<RAJA::operators::maximum>(&maxinit),
     RAJA::expt::Reduce<RAJA::operators::maximum>(&max),
+    RAJA::expt::KernelName("RAJA Reduce MaxLoc"),
     [=] RAJA_HOST_DEVICE(IDX_TYPE idx, VL_TYPE &mi, VL_TYPE &m) {
       mi.max( working_array[idx], idx );
       m.max( working_array[idx], idx );

diff --git a/test/functional/forall/reduce-basic/tests/test-forall-basic-expt-ReduceMin.hpp b/test/functional/forall/reduce-basic/tests/test-forall-basic-expt-ReduceMin.hpp
@@ -54,6 +54,7 @@ void ForallReduceMinBasicTestImpl(const SEG_TYPE& seg,
   RAJA::forall<EXEC_POLICY>(seg, 
     RAJA::expt::Reduce<RAJA::operators::minimum>(&mininit),
     RAJA::expt::Reduce<RAJA::operators::minimum>(&min),
+    RAJA::expt::KernelName("RAJA Reduce Min"),
     [=] RAJA_HOST_DEVICE(IDX_TYPE idx, DATA_TYPE &mi, DATA_TYPE &m) {
       mi = RAJA_MIN(working_array[idx], mi);
       m  = RAJA_MIN(working_array[idx], m);

diff --git a/test/functional/forall/reduce-basic/tests/test-forall-basic-expt-ReduceMinLoc.hpp b/test/functional/forall/reduce-basic/tests/test-forall-basic-expt-ReduceMinLoc.hpp
@@ -64,6 +64,7 @@ void ForallReduceMinLocBasicTestImpl(const SEG_TYPE& seg,
   RAJA::forall<EXEC_POLICY>(seg, 
     RAJA::expt::Reduce<RAJA::operators::minimum>(&mininit),
     RAJA::expt::Reduce<RAJA::operators::minimum>(&min),
+    RAJA::expt::KernelName("RAJA Reduce MinLoc"),
     [=] RAJA_HOST_DEVICE(IDX_TYPE idx, VL_TYPE &mi, VL_TYPE &m) {
       mi.min( working_array[idx], idx );
       m.min( working_array[idx], idx );

diff --git a/test/functional/forall/reduce-basic/tests/test-forall-basic-expt-ReduceSum.hpp b/test/functional/forall/reduce-basic/tests/test-forall-basic-expt-ReduceSum.hpp
@@ -52,6 +52,7 @@ void ForallReduceSumBasicTestImpl(const SEG_TYPE& seg,
   RAJA::forall<EXEC_POLICY>(seg, 
     RAJA::expt::Reduce<RAJA::operators::plus>(&sum),
     RAJA::expt::Reduce<RAJA::operators::plus>(&sum2),
+    RAJA::expt::KernelName("RAJA Reduce Sum"),
     [=] RAJA_HOST_DEVICE(IDX_TYPE idx, DATA_TYPE &s1, DATA_TYPE &s2) {
       s1 += working_array[idx];
       s2 += working_array[idx];