Use ./src instead of ./cpp

Create subdir for each cpp-based implementation

.github/workflows/main.yaml

@@ -7,7 +7,7 @@ jobs:
     runs-on: ubuntu-18.04
     defaults:
       run:
-        working-directory: ./cpp
+        working-directory: ./src
     steps:
       - uses: actions/checkout@v2
 
@@ -16,7 +16,7 @@ jobs:
         uses: actions/cache@v2
         with:
           path: compilers
-          key: ${{ runner.os }}-${{ hashFiles('./cpp/ci-prepare-bionic.sh') }}
+          key: ${{ runner.os }}-${{ hashFiles('./src/ci-prepare-bionic.sh') }}
 
       - name: Prepare compilers
         if: steps.prepare-compilers.outputs.cache-hit != 'true'

README.md

@@ -63,20 +63,22 @@ The project supports building with CMake >= 3.13.0, it can be installed without
 As with any CMake project, first configure the project:
 
 ```shell
-> cd babelstream/cpp
+> cd babelstream/src
 > cmake -Bbuild -H. -DMODEL=<model> <model specific flags prefixed with -D...> # configure the build, build type defaults to Release 
 > cmake --build build # compile it 
-> ./build/babelstream # executable available at ./build/
+> ./build/<model>-stream # executable available at ./build/
 ```
 
+Source for each model's implementations are located in `./src/<model>`.
+
 By default, we have defined a set of optimal flags for known HPC compilers.
 There are assigned those to `RELEASE_FLAGS`, and you can override them if required.
 
 To find out what flag each model supports or requires, simply configure while only specifying the model.
 For example:
 ```shell
-> cd babelstream/cpp
+> cd babelstream/src
-> cmake -Bbuild -H. -DMODEL=OCL 
+> cmake -Bbuild -H. -DMODEL=ocl 
 ...
 - Common Release flags are `-O3`, set RELEASE_FLAGS to override
 -- CXX_EXTRA_FLAGS: 
@@ -90,8 +92,8 @@ For example:
         Use this for linking extra libraries (e.g `-lmylib`, or simply `mylib`) 
 -- CXX_EXTRA_LINKER_FLAGS: 
         Append to linker flags (i.e GCC's `-Wl` or equivalent)
--- Available models:  OMP;OCL;STD;STD20;HIP;CUDA;KOKKOS;SYCL;ACC;RAJA
+-- Available models:  omp;ocl;std;std20;hip;cuda;kokkos;sycl;acc;raja
--- Selected model  :  OCL
+-- Selected model  :  ocl
 -- Supported flags:
 
    CMAKE_CXX_COMPILER (optional, default=c++): Any CXX compiler that is supported by CMake detection
@@ -107,7 +109,7 @@ Alternatively, refer to the [CI script](./ci-test-compile.sh), which test-compil
 We have supplied a series of Makefiles, one for each programming model, to assist with building.
 The Makefiles contain common build options, and should be simple to customise for your needs too.
 
-General usage is `make -f <Model>.make`
+General usage is `make -C src/<model>`
 Common compiler flags and names can be set by passing a `COMPILER` option to Make, e.g. `make COMPILER=GNU`.
 Some models allow specifying a CPU or GPU style target, and this can be set by passing a `TARGET` option to Make, e.g. `make TARGET=GPU`.
 
@@ -125,7 +127,7 @@ cd
 wget https://github.com/kokkos/kokkos/archive/3.1.01.tar.gz
 tar -xvf 3.1.01.tar.gz # should end up with ~/kokkos-3.1.01
 cd BabelStream
-make -f Kokkos.make KOKKOS_PATH=~/kokkos-3.1.01 
+make -C src/kokkos KOKKOS_PATH=~/kokkos-3.1.01 
 ```
 See make output for more information on supported flags.
 

cpp/.gitignore

@@ -1,30 +0,0 @@
-
-cuda-stream
-ocl-stream
-omp-stream
-acc-stream
-raja-stream
-kokkos-stream
-std-stream
-sycl-stream
-hip-stream
-
-*.o
-*.bc
-*.sycl
-*.tar
-*.gz
-*.a
-
-KokkosCore_config.*
-
-.DS_Store
-
-Makefile
-
-build/
-cmake-build-*/
-CMakeFiles/
-.idea/
-.vscode/
-.directory

legacy/Makefile

@@ -14,7 +14,8 @@ endif
 
 
 hc-stream: ../main.cpp HCStream.cpp
-	$(HCC) $(CXXFLAGS) -DHC  $^  $(LDFLAGS) $(EXTRA_FLAGS) -o $@
+	$(HCC) $(CXXFLAGS) -DHC  $^  $(LDFLAGS) $(EXTRA_FLAGS) -o $@ -I. -I..
+
 
 .PHONY: clean
 clean:

src/.gitignore

@@ -0,0 +1,29 @@
+
+**/cuda-stream
+**/ocl-stream
+**/omp-stream
+**/acc-stream
+**/raja-stream
+**/kokkos-stream
+**/std-stream
+**/sycl-stream
+**/hip-stream
+
+**/*.o
+**/*.bc
+**/*.sycl
+**/*.tar
+**/*.gz
+**/*.a
+
+**/KokkosCore_Config_*
+
+**/.DS_Store
+
+
+build/
+cmake-build-*/
+CMakeFiles/
+.idea/
+.vscode/
+.directory

src/CMakeLists.txt

@@ -8,46 +8,7 @@ project(BabelStream VERSION 3.5 LANGUAGES CXX)
 set(CMAKE_CXX_EXTENSIONS OFF)
 set(CMAKE_CXX_STANDARD_REQUIRED ON)
 
-
+ 
-#set(MODEL SYCL)
-#set(SYCL_COMPILER COMPUTECPP)
-#set(SYCL_COMPILER_DIR /home/tom/Desktop/computecpp_archive/ComputeCpp-CE-2.3.0-x86_64-linux-gnu)
-#set(MODEL RAJA)
-#set(RAJA_IN_TREE /home/tom/Downloads/RAJA-v0.13.0/)
-#set(ENABLE_CUDA ON)
-#set(TARGET NVIDIA)
-#set(CUDA_TOOLKIT_ROOT_DIR /opt/cuda-11.2)
-#set(CUDA_ARCH sm_70)
-#set(BLT_DIR /home/tom/Downloads/blt-0.3.6/)
-
-#set(MODEL STD)
-#set(ARCH  cc70)
-#set(CXX_EXTRA_FLAGS -v)
-
-#set(MODEL CUDA)
-#set(ARCH  sm_70)
-#set(CMAKE_CUDA_COMPILER /opt/cuda-11.2/bin/nvcc)
-
-#set(MODEL OCL)
-#set(OpenCL_LIBRARY /opt/rocm-4.0.0/opencl/lib/libOpenCL.so)
-#set(OpenCL_INCLUDE_DIR /opt/rocm-4.0.0/opencl/lib)
-#set(RELEASE_FLAGS -Ofast)
-#set(CXX_EXTRA_FLAGS -O2)
-
-#set(CMAKE_CXX_COMPILER /usr/lib/aomp/bin/clang++)
-#set(MODEL OMP)
-##set(OFFLOAD "AMD:gfx803")
-#set(OFFLOAD "NVIDIA:sm_35")
-#set(CXX_EXTRA_FLAGS --cuda-path=/opt/cuda-10.2/)
-
-#set(OFFLOAD "AMD:_70")
-#set(CXX_EXTRA_FLAGS --cuda-path=/opt/cuda-10.2/ --gcc-toolchain=/home/tom/spack/opt/spack/linux-fedora33-zen2/gcc-10.2.1/gcc-8.3.0-latmjo2hl2yv53255xkwko7k3y7bx2vv)
-#set(CXX_EXTRA_LINKER_FLAGS  )
-#set(MODEL HIP)
-
-#set(MODEL KOKKOS)
-#set(KOKKOS_IN_TREE /home/tom/Downloads/kokkos-3.3.00/)
-
 # the final executable name
 set(EXE_NAME babelstream)
 
@@ -101,17 +62,17 @@ endif ()
 include(register_models.cmake)
 
 # register out models <model_name> <preprocessor_def_name> <source files...>
-register_model(OMP OMP OMPStream.cpp)
+register_model(omp OMP OMPStream.cpp)
-register_model(OCL OCL OCLStream.cpp)
+register_model(ocl OCL OCLStream.cpp)
-register_model(STD STD STDStream.cpp)
+register_model(std STD STDStream.cpp)
-register_model(STD20 STD20 STD20Stream.cpp)
+register_model(std20 STD20 STD20Stream.cpp)
-register_model(HIP HIP HIPStream.cpp)
+register_model(hip HIP HIPStream.cpp)
-register_model(CUDA CUDA CUDAStream.cu)
+register_model(cuda CUDA CUDAStream.cu)
-register_model(KOKKOS KOKKOS KokkosStream.cpp)
+register_model(kokkos KOKKOS KokkosStream.cpp)
-register_model(SYCL SYCL SYCLStream.cpp)
+register_model(sycl SYCL SYCLStream.cpp)
-register_model(ACC ACC ACCStream.cpp)
+register_model(acc ACC ACCStream.cpp)
 # defining RAJA collides with the RAJA namespace so USE_RAJA
-register_model(RAJA USE_RAJA RAJAStream.cpp)
+register_model(raja USE_RAJA RAJAStream.cpp)
 
 
 set(USAGE ON CACHE BOOL "Whether to print all custom flags for the selected model")
@@ -169,6 +130,7 @@ message(STATUS "Executable  : ${EXE_NAME}")
 
 # below we have all the usual CMake target setup steps
 
+include_directories(.)
 add_executable(${EXE_NAME} ${IMPL_SOURCES} main.cpp)
 target_link_libraries(${EXE_NAME} PUBLIC ${LINK_LIBRARIES})
 target_compile_definitions(${EXE_NAME} PUBLIC ${IMPL_DEFINITIONS})
@@ -184,7 +146,7 @@ target_link_options(${EXE_NAME} PUBLIC LINKER:${CXX_EXTRA_LINKER_FLAGS})
 target_link_options(${EXE_NAME} PUBLIC ${LINK_FLAGS} ${CXX_EXTRA_LINK_FLAGS})
 
 # some models require the target to be already specified so they can finish their setup here
-# this only happens if the MODEL.cmake definition contains the `setup_target` macro
+# this only happens if the model.cmake definition contains the `setup_target` macro
 if (COMMAND setup_target)
     setup_target(${EXE_NAME})
 endif ()

src/acc/Makefile

@@ -50,8 +50,8 @@ endif
 FLAGS_GNU = -O3 -std=c++11 -Drestrict=__restrict -fopenacc
 CXXFLAGS = $(FLAGS_$(COMPILER))
 
-acc-stream: main.cpp ACCStream.cpp
+acc-stream: ../main.cpp ACCStream.cpp
-	$(COMPILER_$(COMPILER)) $(CXXFLAGS) -DACC $^ $(EXTRA_FLAGS) -o $@
+	$(COMPILER_$(COMPILER)) $(CXXFLAGS) -DACC $^ $(EXTRA_FLAGS) -o $@ -I. -I..
 
 .PHONY: clean
 clean:

src/ci-test-compile.sh

@@ -86,38 +86,38 @@ run_build() {
 }
 
 ###
-#KOKKOS_SRC="/home/tom/Downloads/kokkos-3.3.00"
+# KOKKOS_SRC="/home/tom/Downloads/kokkos-3.3.00"
-#RAJA_SRC="/home/tom/Downloads/RAJA-v0.13.0"
+# RAJA_SRC="/home/tom/Downloads/RAJA-v0.13.0"
-#
+
-#GCC_CXX="/usr/bin/g++"
+# GCC_CXX="/usr/bin/g++"
-#CLANG_CXX="/usr/bin/clang++"
+# CLANG_CXX="/usr/bin/clang++"
-#
+
-#NVSDK="/home/tom/Downloads/nvhpc_2021_212_Linux_x86_64_cuda_11.2/install_components/Linux_x86_64/21.2/"
+# NVSDK="/home/tom/Downloads/nvhpc_2021_212_Linux_x86_64_cuda_11.2/install_components/Linux_x86_64/21.2/"
-#NVHPC_NVCXX="$NVSDK/compilers/bin/nvc++"
+# NVHPC_NVCXX="$NVSDK/compilers/bin/nvc++"
-#NVHPC_NVCC="$NVSDK/cuda/11.2/bin/nvcc"
+# NVHPC_NVCC="$NVSDK/cuda/11.2/bin/nvcc"
-#NVHPC_CUDA_DIR="$NVSDK/cuda/11.2"
+# NVHPC_CUDA_DIR="$NVSDK/cuda/11.2"
-#"$NVSDK/compilers/bin/makelocalrc" "$NVSDK/compilers/bin/" -x
+# "$NVSDK/compilers/bin/makelocalrc" "$NVSDK/compilers/bin/" -x
-#
+
-#AOCC_CXX="/opt/AMD/aocc-compiler-2.3.0/bin/clang++"
+# AOCC_CXX="/opt/AMD/aocc-compiler-2.3.0/bin/clang++"
-#AOMP_CXX="/usr/lib/aomp/bin/clang++"
+# AOMP_CXX="/usr/lib/aomp/bin/clang++"
-#OCL_LIB="/home/tom/Downloads/oclcpuexp-2020.11.11.0.04_rel/x64/libOpenCL.so"
+# OCL_LIB="/home/tom/Downloads/oclcpuexp-2020.11.11.0.04_rel/x64/libOpenCL.so"
-#
+
-## AMD needs this rocm_path thing exported...
+# # AMD needs this rocm_path thing exported...
-#export ROCM_PATH="/opt/rocm-4.0.0"
+# export ROCM_PATH="/opt/rocm-4.0.0"
-#HIP_CXX="/opt/rocm-4.0.0/bin/hipcc"
+# HIP_CXX="/opt/rocm-4.0.0/bin/hipcc"
-#COMPUTECPP_DIR="/home/tom/Desktop/computecpp_archive/ComputeCpp-CE-2.3.0-x86_64-linux-gnu"
+# COMPUTECPP_DIR="/home/tom/Desktop/computecpp_archive/ComputeCpp-CE-2.3.0-x86_64-linux-gnu"
-#DPCPP_DIR="/home/tom/Downloads/dpcpp_compiler"
+# DPCPP_DIR="/home/tom/Downloads/dpcpp_compiler"
-#HIPSYCL_DIR="/opt/hipsycl/cff515c/"
+# HIPSYCL_DIR="/opt/hipsycl/cff515c/"
-#
+
-#ICPX_CXX="/opt/intel/oneapi/compiler/2021.1.2/linux/bin/icpx"
+# ICPX_CXX="/opt/intel/oneapi/compiler/2021.1.2/linux/bin/icpx"
-#ICPC_CXX="/opt/intel/oneapi/compiler/2021.1.2/linux/bin/intel64/icpc"
+# ICPC_CXX="/opt/intel/oneapi/compiler/2021.1.2/linux/bin/intel64/icpc"
-#
+
-#GCC_STD_PAR_LIB="tbb"
+# GCC_STD_PAR_LIB="tbb"
-#CLANG_STD_PAR_LIB="tbb"
+# CLANG_STD_PAR_LIB="tbb"
-#GCC_OMP_OFFLOAD_AMD=false
+# GCC_OMP_OFFLOAD_AMD=false
-#GCC_OMP_OFFLOAD_NVIDIA=true
+# GCC_OMP_OFFLOAD_NVIDIA=true
-#CLANG_OMP_OFFLOAD_AMD=false
+# CLANG_OMP_OFFLOAD_AMD=false
-#CLANG_OMP_OFFLOAD_NVIDIA=false
+# CLANG_OMP_OFFLOAD_NVIDIA=false
 ###
 
 AMD_ARCH="gfx_903"
@@ -128,39 +128,39 @@ build_gcc() {
   local name="gcc_build"
   local cxx="-DCMAKE_CXX_COMPILER=${GCC_CXX:?}"
 
-  run_build $name "${GCC_CXX:?}" OMP "$cxx"
+  run_build $name "${GCC_CXX:?}" omp "$cxx"
   if [ "$MODEL" = "all" ] || [ "$MODEL" = "OMP" ]; then
     # sanity check that it at least runs
-    echo "Sanity checking GCC OMP build..."
+    echo "Sanity checking GCC omp build..."
-    "./$BUILD_DIR/OMP_$name/omp-stream" -s 1048576 -n 10
+    "./$BUILD_DIR/omp_$name/omp-stream" -s 1048576 -n 10
   fi
 
   # some distributions like Ubuntu bionic implements std par with TBB, so conditionally link it here
-  run_build $name "${GCC_CXX:?}" STD "$cxx -DCXX_EXTRA_LIBRARIES=${GCC_STD_PAR_LIB:-}"
+  run_build $name "${GCC_CXX:?}" std "$cxx -DCXX_EXTRA_LIBRARIES=${GCC_STD_PAR_LIB:-}"
-  run_build $name "${GCC_CXX:?}" STD20 "$cxx -DCXX_EXTRA_LIBRARIES=${GCC_STD_PAR_LIB:-}"
+  run_build $name "${GCC_CXX:?}" std20 "$cxx -DCXX_EXTRA_LIBRARIES=${GCC_STD_PAR_LIB:-}"
 
   if [ "${GCC_OMP_OFFLOAD_AMD:-false}" != "false" ]; then
-    run_build "amd_$name" "${GCC_CXX:?}" ACC "$cxx -DCXX_EXTRA_FLAGS=-foffload=amdgcn-amdhsa"
+    run_build "amd_$name" "${GCC_CXX:?}" acc "$cxx -DCXX_EXTRA_FLAGS=-foffload=amdgcn-amdhsa"
-    run_build "amd_$name" "${GCC_CXX:?}" OMP "$cxx -DOFFLOAD=AMD:$AMD_ARCH"
+    run_build "amd_$name" "${GCC_CXX:?}" omp "$cxx -DOFFLOAD=AMD:$AMD_ARCH"
   fi
 
   if [ "${GCC_OMP_OFFLOAD_NVIDIA:-false}" != "false" ]; then
-    run_build "nvidia_$name" "${GCC_CXX:?}" ACC "$cxx -DCXX_EXTRA_FLAGS=-foffload=nvptx-none"
+    run_build "nvidia_$name" "${GCC_CXX:?}" acc "$cxx -DCXX_EXTRA_FLAGS=-foffload=nvptx-none"
-    run_build "nvidia_$name" "${GCC_CXX:?}" OMP "$cxx -DOFFLOAD=NVIDIA:$NV_ARCH"
+    run_build "nvidia_$name" "${GCC_CXX:?}" omp "$cxx -DOFFLOAD=NVIDIA:$NV_ARCH"
   fi
 
-  run_build $name "${GCC_CXX:?}" CUDA "$cxx -DCMAKE_CUDA_COMPILER=${NVHPC_NVCC:?} -DCUDA_ARCH=$NV_ARCH"
+  run_build $name "${GCC_CXX:?}" cuda "$cxx -DCMAKE_CUDA_COMPILER=${NVHPC_NVCC:?} -DCUDA_ARCH=$NV_ARCH"
-  run_build $name "${GCC_CXX:?}" CUDA "$cxx -DCMAKE_CUDA_COMPILER=${NVHPC_NVCC:?} -DCUDA_ARCH=$NV_ARCH -DMEM=MANAGED"
+  run_build $name "${GCC_CXX:?}" cuda "$cxx -DCMAKE_CUDA_COMPILER=${NVHPC_NVCC:?} -DCUDA_ARCH=$NV_ARCH -DMEM=MANAGED"
-  run_build $name "${GCC_CXX:?}" CUDA "$cxx -DCMAKE_CUDA_COMPILER=${NVHPC_NVCC:?} -DCUDA_ARCH=$NV_ARCH -DMEM=PAGEFAULT"
+  run_build $name "${GCC_CXX:?}" cuda "$cxx -DCMAKE_CUDA_COMPILER=${NVHPC_NVCC:?} -DCUDA_ARCH=$NV_ARCH -DMEM=PAGEFAULT"
-  #  run_build $name "${CC_CXX:?}" KOKKOS "$cxx -DKOKKOS_IN_TREE=${KOKKOS_SRC:?} -DKokkos_ENABLE_CUDA=ON"
+  #  run_build $name "${CC_CXX:?}" kokkos "$cxx -DKOKKOS_IN_TREE=${KOKKOS_SRC:?} -DKokkos_ENABLE_CUDA=ON"
-  run_build "cuda_$name" "${GCC_CXX:?}" KOKKOS "$cxx -DKOKKOS_IN_TREE=${KOKKOS_SRC:?} -DKokkos_ENABLE_OPENMP=ON"
+  run_build "cuda_$name" "${GCC_CXX:?}" kokkos "$cxx -DKOKKOS_IN_TREE=${KOKKOS_SRC:?} -DKokkos_ENABLE_OPENMP=ON"
-  run_build $name "${GCC_CXX:?}" OCL "$cxx -DOpenCL_LIBRARY=${OCL_LIB:?}"
+  run_build $name "${GCC_CXX:?}" ocl "$cxx -DOpenCL_LIBRARY=${OCL_LIB:?}"
-  run_build $name "${GCC_CXX:?}" RAJA "$cxx -DRAJA_IN_TREE=${RAJA_SRC:?}"
+  run_build $name "${GCC_CXX:?}" raja "$cxx -DRAJA_IN_TREE=${RAJA_SRC:?}"
 
 #  FIXME fails due to https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100102
 #  FIXME we also got https://github.com/NVIDIA/nccl/issues/494
 
-#  run_build "cuda_$name" "${GCC_CXX:?}" RAJA "$cxx -DRAJA_IN_TREE=${RAJA_SRC:?} \
+#  run_build "cuda_$name" "${GCC_CXX:?}" raja "$cxx -DRAJA_IN_TREE=${RAJA_SRC:?} \
 #  -DENABLE_CUDA=ON \
 #  -DTARGET=NVIDIA \
 #  -DCUDA_TOOLKIT_ROOT_DIR=${NVHPC_CUDA_DIR:?} \
@@ -171,46 +171,46 @@ build_gcc() {
 build_clang() {
   local name="clang_build"
   local cxx="-DCMAKE_CXX_COMPILER=${CLANG_CXX:?}"
-  run_build $name "${CLANG_CXX:?}" OMP "$cxx"
+  run_build $name "${CLANG_CXX:?}" omp "$cxx"
 
   if [ "${CLANG_OMP_OFFLOAD_AMD:-false}" != "false" ]; then
-    run_build "amd_$name" "${GCC_CXX:?}" OMP "$cxx -DOFFLOAD=AMD:$AMD_ARCH"
+    run_build "amd_$name" "${GCC_CXX:?}" omp "$cxx -DOFFLOAD=AMD:$AMD_ARCH"
   fi
 
   if [ "${CLANG_OMP_OFFLOAD_NVIDIA:-false}" != "false" ]; then
-    run_build "nvidia_$name" "${GCC_CXX:?}" OMP "$cxx -DOFFLOAD=NVIDIA:$NV_ARCH"
+    run_build "nvidia_$name" "${GCC_CXX:?}" omp "$cxx -DOFFLOAD=NVIDIA:$NV_ARCH"
   fi
 
-  run_build $name "${CLANG_CXX:?}" CUDA "$cxx -DCMAKE_CUDA_COMPILER=${NVHPC_NVCC:?} -DCUDA_ARCH=$NV_ARCH"
+  run_build $name "${CLANG_CXX:?}" cuda "$cxx -DCMAKE_CUDA_COMPILER=${NVHPC_NVCC:?} -DCUDA_ARCH=$NV_ARCH"
-  run_build $name "${CLANG_CXX:?}" CUDA "$cxx -DCMAKE_CUDA_COMPILER=${NVHPC_NVCC:?} -DCUDA_ARCH=$NV_ARCH -DMEM=MANAGED"
+  run_build $name "${CLANG_CXX:?}" cuda "$cxx -DCMAKE_CUDA_COMPILER=${NVHPC_NVCC:?} -DCUDA_ARCH=$NV_ARCH -DMEM=MANAGED"
-  run_build $name "${CLANG_CXX:?}" CUDA "$cxx -DCMAKE_CUDA_COMPILER=${NVHPC_NVCC:?} -DCUDA_ARCH=$NV_ARCH -DMEM=PAGEFAULT"
+  run_build $name "${CLANG_CXX:?}" cuda "$cxx -DCMAKE_CUDA_COMPILER=${NVHPC_NVCC:?} -DCUDA_ARCH=$NV_ARCH -DMEM=PAGEFAULT"
-  run_build $name "${CLANG_CXX:?}" KOKKOS "$cxx -DKOKKOS_IN_TREE=${KOKKOS_SRC:?} -DKokkos_ENABLE_OPENMP=ON"
+  run_build $name "${CLANG_CXX:?}" kokkos "$cxx -DKOKKOS_IN_TREE=${KOKKOS_SRC:?} -DKokkos_ENABLE_OPENMP=ON"
-  run_build $name "${CLANG_CXX:?}" OCL "$cxx -DOpenCL_LIBRARY=${OCL_LIB:?}"
+  run_build $name "${CLANG_CXX:?}" ocl "$cxx -DOpenCL_LIBRARY=${OCL_LIB:?}"
-  run_build $name "${CLANG_CXX:?}" STD "$cxx -DCXX_EXTRA_LIBRARIES=${CLANG_STD_PAR_LIB:-}"
+  run_build $name "${CLANG_CXX:?}" std "$cxx -DCXX_EXTRA_LIBRARIES=${CLANG_STD_PAR_LIB:-}"
-  # run_build $name "${LANG_CXX:?}" STD20 "$cxx -DCXX_EXTRA_LIBRARIES=${CLANG_STD_PAR_LIB:-}" # not yet supported
+  # run_build $name "${LANG_CXX:?}" std20 "$cxx -DCXX_EXTRA_LIBRARIES=${CLANG_STD_PAR_LIB:-}" # not yet supported
-  run_build $name "${CLANG_CXX:?}" RAJA "$cxx -DRAJA_IN_TREE=${RAJA_SRC:?}"
+  run_build $name "${CLANG_CXX:?}" raja "$cxx -DRAJA_IN_TREE=${RAJA_SRC:?}"
   # no clang /w RAJA+cuda because it needs nvcc which needs gcc
 }
 
 build_nvhpc() {
   local name="nvhpc_build"
   local cxx="-DCMAKE_CXX_COMPILER=${NVHPC_NVCXX:?}"
-  run_build $name "${NVHPC_NVCXX:?}" STD "$cxx -DNVHPC_OFFLOAD=$NV_ARCH_CCXY"
+  run_build $name "${NVHPC_NVCXX:?}" std "$cxx -DNVHPC_OFFLOAD=$NV_ARCH_CCXY"
-  run_build $name "${NVHPC_NVCXX:?}" ACC "$cxx -DTARGET_DEVICE=gpu -DTARGET_PROCESSOR=px -DCUDA_ARCH=$NV_ARCH_CCXY"
+  run_build $name "${NVHPC_NVCXX:?}" acc "$cxx -DTARGET_DEVICE=gpu -DTARGET_PROCESSOR=px -DCUDA_ARCH=$NV_ARCH_CCXY"
-  run_build $name "${NVHPC_NVCXX:?}" ACC "$cxx -DTARGET_DEVICE=multicore -DTARGET_PROCESSOR=zen"
+  run_build $name "${NVHPC_NVCXX:?}" acc "$cxx -DTARGET_DEVICE=multicore -DTARGET_PROCESSOR=zen"
 }
 
 build_aocc() {
-  run_build aocc_build "${AOCC_CXX:?}" OMP "-DCMAKE_CXX_COMPILER=${AOCC_CXX:?}"
+  run_build aocc_build "${AOCC_CXX:?}" omp "-DCMAKE_CXX_COMPILER=${AOCC_CXX:?}"
 }
 
 build_aomp() {
-  run_build aomp_amd_build "${AOMP_CXX:?}" OMP "-DCMAKE_CXX_COMPILER=${AOMP_CXX:?} -DOFFLOAD=AMD:gfx906"
+  run_build aomp_amd_build "${AOMP_CXX:?}" omp "-DCMAKE_CXX_COMPILER=${AOMP_CXX:?} -DOFFLOAD=AMD:gfx906"
   #run_build aomp_nvidia_build "-DCMAKE_CXX_COMPILER=${AOMP_CXX:?} -DOFFLOAD=NVIDIA:$NV_ARCH"
 }
 
 build_hip() {
-  run_build hip_build "${HIP_CXX:?}" HIP "-DCMAKE_CXX_COMPILER=${HIP_CXX:?}"
+  run_build hip_build "${HIP_CXX:?}" hip "-DCMAKE_CXX_COMPILER=${HIP_CXX:?}"
 }
 
 build_icpx() {
@@ -218,7 +218,7 @@ build_icpx() {
   set +u
   source /opt/intel/oneapi/setvars.sh -force || true
   set -u
-  run_build intel_build "${ICPX_CXX:?}" OMP "-DCMAKE_CXX_COMPILER=${ICPX_CXX:?} -DOFFLOAD=INTEL"
+  run_build intel_build "${ICPX_CXX:?}" omp "-DCMAKE_CXX_COMPILER=${ICPX_CXX:?} -DOFFLOAD=INTEL"
 }
 
 build_icpc() {
@@ -228,31 +228,31 @@ build_icpc() {
   set -u
   local name="intel_build"
   local cxx="-DCMAKE_CXX_COMPILER=${ICPC_CXX:?}"
-  run_build $name "${ICPC_CXX:?}" OMP "$cxx"
+  run_build $name "${ICPC_CXX:?}" omp "$cxx"
-  run_build $name "${ICPC_CXX:?}" OCL "$cxx -DOpenCL_LIBRARY=${OCL_LIB:?}"
+  run_build $name "${ICPC_CXX:?}" ocl "$cxx -DOpenCL_LIBRARY=${OCL_LIB:?}"
-  run_build $name "${ICPC_CXX:?}" RAJA "$cxx -DRAJA_IN_TREE=${RAJA_SRC:?}"
+  run_build $name "${ICPC_CXX:?}" raja "$cxx -DRAJA_IN_TREE=${RAJA_SRC:?}"
-  run_build $name "${ICPC_CXX:?}" KOKKOS "$cxx -DKOKKOS_IN_TREE=${KOKKOS_SRC:?} -DKokkos_ENABLE_OPENMP=ON"
+  run_build $name "${ICPC_CXX:?}" kokkos "$cxx -DKOKKOS_IN_TREE=${KOKKOS_SRC:?} -DKokkos_ENABLE_OPENMP=ON"
 }
 
 build_computecpp() {
-  run_build computecpp_build "compute++" SYCL "-DCMAKE_CXX_COMPILER=${GCC_CXX:?} \
+  run_build computecpp_build "compute++" sycl "-DCMAKE_CXX_COMPILER=${GCC_CXX:?} \
   -DSYCL_COMPILER=COMPUTECPP \
   -DSYCL_COMPILER_DIR=${COMPUTECPP_DIR:?} \
   -DOpenCL_LIBRARY=${OCL_LIB:?}"
 }
 
 build_dpcpp() {
-  run_build intel_build "${DPCPP_DIR:?}" SYCL "-DCMAKE_CXX_COMPILER=${GCC_CXX:?} \
+  run_build intel_build "${DPCPP_DIR:?}" sycl "-DCMAKE_CXX_COMPILER=${GCC_CXX:?} \
   -DSYCL_COMPILER=DPCPP \
   -DSYCL_COMPILER_DIR=${DPCPP_DIR:?}"
 
   #  for oneAPI BaseKit:
   #  source /opt/intel/oneapi/setvars.sh -force
-  #  run_build intel_build "dpcpp" SYCL "-DCMAKE_CXX_COMPILER=${GCC_CXX:?} -DSYCL_COMPILER=ONEAPI-DPCPP"
+  #  run_build intel_build "dpcpp" sycl "-DCMAKE_CXX_COMPILER=${GCC_CXX:?} -DSYCL_COMPILER=ONEAPI-DPCPP"
 }
 
 build_hipsycl() {
-  run_build hipsycl_build "syclcc" SYCL "
+  run_build hipsycl_build "syclcc" sycl "
   -DSYCL_COMPILER=HIPSYCL \
   -DSYCL_COMPILER_DIR=${HIPSYCL_DIR:?}"
 }

src/cuda/Makefile

@@ -31,8 +31,8 @@ MEM_PAGEFAULT= -DPAGEFAULT
 MEM_MODE = $(MEM_$(MEM))
 
 
-cuda-stream: main.cpp CUDAStream.cu
+cuda-stream: ../main.cpp CUDAStream.cu
-	$(CUDA_CXX) -std=c++11 $(CXXFLAGS) -arch=$(NVARCH) $(MEM_MODE) -DCUDA $^ $(EXTRA_FLAGS) -o $@
+	$(CUDA_CXX) -std=c++11 $(CXXFLAGS) -arch=$(NVARCH) $(MEM_MODE) -DCUDA $^ $(EXTRA_FLAGS) -o $@ -I. -I..
 
 .PHONY: clean
 clean:

src/hip/Makefile

@@ -2,8 +2,8 @@
 HIP_PATH?= /opt/rocm/hip
 HIPCC=$(HIP_PATH)/bin/hipcc
 
-hip-stream: main.cpp HIPStream.cpp
+hip-stream: ../main.cpp HIPStream.cpp
-	$(HIPCC) $(CXXFLAGS) -O3 -std=c++11 -DHIP $^ $(EXTRA_FLAGS) -o $@
+	$(HIPCC) $(CXXFLAGS) -O3 -std=c++11 -DHIP $^ $(EXTRA_FLAGS) -o $@ -I. -I..
 
 .PHONY: clean
 clean:

src/kokkos/Makefile

@@ -71,7 +71,7 @@ CXX = $(NVCC_WRAPPER)
 endif
 endif
 
-OBJ = main.o KokkosStream.o
+OBJ = KokkosStream.o
 CXXFLAGS = -O3 
 LINKFLAGS = # empty for now
 
@@ -85,14 +85,15 @@ endif
 endif
 
 include $(KOKKOS_PATH)/Makefile.kokkos
+HEADERS = $(wildcard $(MAKEFILE_PATH)*.hpp)
 
-kokkos-stream: $(OBJ) $(KOKKOS_LINK_DEPENDS)
+kokkos-stream: ../main.cpp $(OBJ) $(KOKKOS_LINK_DEPENDS)
-	$(CXX) $(KOKKOS_LDFLAGS) $(LINKFLAGS) $(EXTRA_PATH) $(OBJ) $(KOKKOS_LIBS) $(LIB) -DKOKKOS -o $@
+	$(CXX) $(KOKKOS_LDFLAGS) $(LINKFLAGS) $(EXTRA_PATH) $(OBJ) $(KOKKOS_LIBS) $(LIB) -DKOKKOS -o $@ -I. -I..
 
-%.o: %.cpp $(KOKKOS_CPP_DEPENDS)
+%.o: %.cpp $(KOKKOS_CPP_DEPENDS) $(HEADERS)
-	$(CXX) $(KOKKOS_CPPFLAGS) $(KOKKOS_CXXFLAGS) $(CXXFLAGS) $(EXTRA_INC) -DKOKKOS -c $<
+	$(CXX) $(KOKKOS_CPPFLAGS) $(KOKKOS_CXXFLAGS) $(CXXFLAGS) $(EXTRA_INC) -c $< -o $(notdir $@)
 
 .PHONY: clean
 clean:
-	rm -f kokkos-stream main.o KokkosStream.o Kokkos_*.o
+	rm -f kokkos-stream main.o KokkosStream.o Kokkos_*.o KokkosCore_*
 

src/ocl/Makefile

@@ -30,8 +30,8 @@ else
   LIBS = -lOpenCL
 endif
 
-ocl-stream: main.cpp OCLStream.cpp
+ocl-stream: ../main.cpp OCLStream.cpp
-	$(CXX) $(CXXFLAGS) -DOCL $^ $(EXTRA_FLAGS) $(LIBS) -o $@
+	$(CXX) $(CXXFLAGS) -DOCL $^ $(EXTRA_FLAGS) $(LIBS) -o $@ -I. -I..
 
 .PHONY: clean
 clean:

src/omp/Makefile

@@ -91,8 +91,8 @@ endif
 
 OMP = $(OMP_$(COMPILER)_$(TARGET))
 
-omp-stream: main.cpp OMPStream.cpp
+omp-stream: ../main.cpp OMPStream.cpp
-	$(CXX) $(CXXFLAGS) -DOMP $^ $(OMP) $(EXTRA_FLAGS) -o $@
+	$(CXX) $(CXXFLAGS) -DOMP  $^ $(OMP) $(EXTRA_FLAGS) -o $@ -I. -I..
 
 .PHONY: clean
 clean:

src/raja/Makefile

@@ -49,8 +49,8 @@ endif
 CXXFLAGS = --expt-extended-lambda -O3 -std=c++11 -x cu -Xcompiler -fopenmp -arch $(ARCH)
 endif
 
-raja-stream: main.cpp RAJAStream.cpp
+raja-stream: ../main.cpp RAJAStream.cpp
-	$(CXX) $(CXXFLAGS) -DUSE_RAJA -I$(RAJA_PATH)/include $^ $(EXTRA_FLAGS) -L$(RAJA_PATH)/lib -lRAJA -o $@
+	$(CXX) $(CXXFLAGS) -DUSE_RAJA -I$(RAJA_PATH)/include $^ $(EXTRA_FLAGS) -L$(RAJA_PATH)/lib -lRAJA -o $@ -I. -I..
 
 .PHONY: clean
 clean:

src/register_models.cmake

@@ -118,22 +118,24 @@ endfunction()
 
 
 macro(register_model NAME PREPROCESSOR_NAME)
-    string(TOUPPER ${NAME} MODEL_UPPER)
     list(APPEND REGISTERED_MODELS "${NAME}")
 
-    list(APPEND IMPL_${MODEL_UPPER}_SOURCES "${ARGN}")
+    string(TOUPPER ${NAME} MODEL_UPPER)
+    list(APPEND IMPL_${MODEL_UPPER}_SOURCES "${NAME}/${ARGN}")
     list(APPEND IMPL_${MODEL_UPPER}_DEFINITIONS "${PREPROCESSOR_NAME}")
 endmacro()
 
 
 macro(load_model MODEL)
-    string(TOUPPER "${MODEL}" MODEL_UPPER)
+    if ("${MODEL}" IN_LIST REGISTERED_MODELS)
-    if ("${MODEL_UPPER}" IN_LIST REGISTERED_MODELS)
+        string(TOLOWER "${MODEL}" MODEL_LOWER)
-        set(MODEL_FILE ${CMAKE_CURRENT_SOURCE_DIR}/${MODEL_UPPER}.cmake)
+        set(MODEL_FILE ${CMAKE_CURRENT_SOURCE_DIR}/${MODEL_LOWER}/model.cmake)
+        include_directories(${CMAKE_CURRENT_SOURCE_DIR}/${MODEL_LOWER})
         if (NOT EXISTS ${MODEL_FILE})
             message(FATAL_ERROR "${MODEL_FILE} not found, perhaps it needs to be implemented?")
         endif ()
         include(${MODEL_FILE})
+        string(TOUPPER "${MODEL}" MODEL_UPPER)
         list(APPEND IMPL_SOURCES ${IMPL_${MODEL_UPPER}_SOURCES})
         list(APPEND IMPL_DEFINITIONS ${IMPL_${MODEL_UPPER}_DEFINITIONS})
 

src/std/Makefile

@@ -6,8 +6,8 @@
 CXXFLAGS=-O3 -std=c++17 -stdpar -DSTD
 STD_CXX=nvc++
 
-std-stream: main.cpp STDStream.cpp
+std-stream: ../main.cpp STDStream.cpp
-	$(STD_CXX) $(CXXFLAGS) $^ $(EXTRA_FLAGS) -o $@
+	$(STD_CXX) $(CXXFLAGS) $^ $(EXTRA_FLAGS) -o $@ -I. -I..
 
 .PHONY: clean
 clean:

src/std20/Makefile

@@ -17,8 +17,8 @@ FLAGS_GNU = -O3 -std=c++2a -march=native
 CXXFLAGS = $(FLAGS_$(COMPILER))
 
 
-std20-stream: main.cpp STD20Stream.cpp
+std20-stream: ../main.cpp STD20Stream.cpp
-	$(CXX) -DSTD20 $(CXXFLAGS) $^ $(EXTRA_FLAGS) -o $@
+	$(CXX) -DSTD20 $(CXXFLAGS) $^ $(EXTRA_FLAGS) -o $@ -I. -I..
 
 .PHONY: clean
 clean:

src/sycl/Makefile

@@ -73,8 +73,8 @@ SYCL_LINK_FLAGS = $(SYCL_$(COMPILER)_LINK_FLAGS)
 SYCL_INCLUDE = $(SYCL_$(COMPILER)_INCLUDE)
 
 # only ComputeCpp generates .sycl files which is a bit odd to deal with so we opted to compile everything together
-sycl-stream: main.cpp SYCLStream.cpp
+sycl-stream: ../main.cpp SYCLStream.cpp
-	$(SYCL_SYCLCXX) $(SYCL_SYCLFLAGS) $(SYCL_FLAGS) $(SYCL_INCLUDE) -DSYCL $(EXTRA_FLAGS) $(SYCL_LINK_FLAGS) $^ -o $@
+	$(SYCL_SYCLCXX) $(SYCL_SYCLFLAGS) $(SYCL_FLAGS) $(SYCL_INCLUDE) -DSYCL $(EXTRA_FLAGS) $(SYCL_LINK_FLAGS) $^ -o $@ -I. -I..
 
 .PHONY: clean
 clean: