Add template-based operation dispatch

CMakeLists.txt

@@ -203,16 +203,11 @@ if (UMPIRE_ENABLE_TESTS)
   add_subdirectory(tests)
 endif ()
 
-if (UMPIRE_ENABLE_DEVELOPER_BENCHMARKS)
+if (UMPIRE_ENABLE_BENCHMARKS)
   add_subdirectory(benchmarks)
   if ((NOT CMAKE_BUILD_TYPE) OR (NOT ${CMAKE_BUILD_TYPE} STREQUAL "Release"))
     message("-- Warning: CMAKE_BUILD_TYPE not set to Release, benchmark information will not be reliable for this build!")
   endif()
-else()
-  if (UMPIRE_ENABLE_BENCHMARKS)
-    message("-- Warning: Benchmarks will not be built. If you want to build with benchmarks,\n"
-          "     set UMPIRE_ENABLE_DEVELOPER_BENCHMARKS to On.")
-  endif()
 endif ()
 
 if (UMPIRE_ENABLE_EXAMPLES)

Dockerfile

@@ -38,8 +38,8 @@ ENV GTEST_COLOR=1
 COPY . /home/umpire/workspace
 WORKDIR /home/umpire/workspace/build
 RUN cmake -DUMPIRE_ENABLE_DEVELOPER_DEFAULTS=On -DCMAKE_CXX_COMPILER=clang++ -DCMAKE_C_COMPILER=clang \
-          -DUMPIRE_ENABLE_C=On -DCMAKE_CXX_FLAGS="-fsanitize=address" -DENABLE_TESTS=On -DUMPIRE_ENABLE_TOOLS=On \
-          -DUMPIRE_ENABLE_ASAN=On -DUMPIRE_ENABLE_SANITIZER_TESTS=On .. && \
+    -DUMPIRE_ENABLE_C=On -DCMAKE_CXX_FLAGS="-fsanitize=address" -DENABLE_TESTS=On -DUMPIRE_ENABLE_TOOLS=On \
+    -DUMPIRE_ENABLE_ASAN=On -DUMPIRE_ENABLE_SANITIZER_TESTS=On .. && \
     make -j 2 && \
     ctest -T test -E operation_tests --output-on-failure
 
@@ -48,7 +48,7 @@ ENV GTEST_COLOR=1
 COPY . /home/umpire/workspace
 WORKDIR /home/umpire/workspace/build
 RUN cmake -DUMPIRE_ENABLE_DEVELOPER_DEFAULTS=On -DCMAKE_CXX_COMPILER=g++ -DENABLE_CUDA=On -DCMAKE_CUDA_COMPILER=/usr/local/cuda/bin/nvcc -DCMAKE_CUDA_ARCHITECTURES=70 .. && \
-    make -j 16
+    make -j 8
 
 # TODO: switch to ROCM 6
 FROM ghcr.io/llnl/radiuss:hip-5.6.1-ubuntu-20.04 AS hip

Makefile

@@ -10,7 +10,7 @@ else
 	DebugArgs=
 endif
 
-targets = asan clang10 clang11 clang12 clang13 gcc11 gcc7 gcc8 gcc9 hip hip.debug nvcc10 sycl umap_build
+targets = gcc clang umap_build asan cuda hip sycl intel 
 
 $(targets):
 	DOCKER_BUILDKIT=1 docker build --target $@ --no-cache $(DebugArgs) .

benchmarks/CMakeLists.txt

@@ -88,64 +88,71 @@ blt_add_target_compile_flags(
   TO pool_stress_test
   FLAGS ${UMPIRE_DISABLE_DEPRECATED_WARNINGS_FLAG})
 
-if (UMPIRE_ENABLE_BENCHMARKS)
-  set (benchmark_depends gbenchmark umpire)
+set (benchmark_depends gbenchmark umpire)
 
-  if (UMPIRE_ENABLE_OPENMP_TARGET)
-    set (benchmark_depends
-      ${benchmark_depends}
-      openmp)
-  endif()
+if (UMPIRE_ENABLE_OPENMP_TARGET)
+  set (benchmark_depends
+    ${benchmark_depends}
+    openmp)
+endif()
 
-  blt_add_executable(
-    NAME allocator_benchmarks
-    SOURCES allocator_benchmarks.cpp
-    DEPENDS_ON ${benchmark_depends})
+blt_add_executable(
+  NAME allocator_benchmarks
+  SOURCES allocator_benchmarks.cpp
+  DEPENDS_ON ${benchmark_depends})
 
-  blt_add_benchmark(
-    NAME allocator_benchmarks
-    COMMAND allocator_benchmarks)
+blt_add_benchmark(
+  NAME allocator_benchmarks
+  COMMAND allocator_benchmarks)
 
-  blt_add_executable(
-    NAME vendor_allocator_benchmarks
-    SOURCES vendor_allocator_benchmarks.cpp
-    DEPENDS_ON ${benchmark_depends})
+blt_add_executable(
+  NAME vendor_allocator_benchmarks
+  SOURCES vendor_allocator_benchmarks.cpp
+  DEPENDS_ON ${benchmark_depends})
+
+blt_add_benchmark(
+  NAME vendor_allocator_benchmarks
+  COMMAND vendor_allocator_benchmarks)
+
+blt_add_executable(
+  NAME debuglog_benchmarks
+  SOURCES debuglog_benchmarks.cpp
+  DEPENDS_ON ${benchmark_depends})
 
-  blt_add_benchmark(
-    NAME vendor_allocator_benchmarks
-    COMMAND vendor_allocator_benchmarks)
+if (UMPIRE_ENABLE_OPENMP)
+  set (benchmark_depends
+    ${benchmark_depends}
+    openmp)
 
   blt_add_executable(
-    NAME debuglog_benchmarks
-    SOURCES debuglog_benchmarks.cpp
+    NAME file_resource_benchmarks
+    SOURCES file_resource_benchmarks.cpp
     DEPENDS_ON ${benchmark_depends})
+endif()
 
-  if (UMPIRE_ENABLE_OPENMP)
-    set (benchmark_depends
-      ${benchmark_depends}
-      openmp)
+blt_add_executable(
+  NAME copy_benchmarks
+  SOURCES copy_benchmarks.cpp
+  DEPENDS_ON ${benchmark_depends})
 
-    blt_add_executable(
-      NAME file_resource_benchmarks
-      SOURCES file_resource_benchmarks.cpp
-      DEPENDS_ON ${benchmark_depends})
-  endif()
+blt_add_benchmark(
+  NAME copy_benchmarks
+  COMMAND copy_benchmarks)
 
-  blt_add_executable(
-    NAME copy_benchmarks
-    SOURCES copy_benchmarks.cpp
-    DEPENDS_ON ${benchmark_depends})
+blt_add_executable(
+  NAME inspector_benchmarks
+  SOURCES inspector_benchmarks.cpp
+  DEPENDS_ON ${benchmark_depends})
 
-  blt_add_benchmark(
-    NAME copy_benchmarks
-    COMMAND copy_benchmarks)
+blt_add_benchmark(
+  NAME inspector_benchmarks
+  COMMAND inspector_benchmarks)
 
-  blt_add_executable(
-    NAME inspector_benchmarks
-    SOURCES inspector_benchmarks.cpp
-    DEPENDS_ON ${benchmark_depends})
+blt_add_executable(
+  NAME copy_performance_benchmark
+  SOURCES op/copy_performance_benchmark.cpp
+  DEPENDS_ON ${benchmark_depends})
 
-  blt_add_benchmark(
-    NAME inspector_benchmarks
-    COMMAND inspector_benchmarks)
-endif()
+blt_add_benchmark(
+  NAME copy_performance_benchmark
+  COMMAND copy_performance_benchmark)