WIP Move CUDA CI to Hamilton

.github/workflows/ci-short.yml

@@ -24,7 +24,7 @@ jobs:
       # map to local user id on CI  machine to allow writing to build cache
       options: --user 1001
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
         with:
           submodules: 'true'
       - name: cpplint
@@ -33,7 +33,7 @@ jobs:
         run: |
           cmake -DCMAKE_CXX_FLAGS=-Werror -Bbuild-copyright-check
           cmake --build build-copyright-check -t check-copyright
-      - uses: actions/upload-artifact@v3
+      - uses: actions/upload-artifact@v4
         with:
           name: configure-log-style
           path: build-copyright-check/CMakeFiles/CMakeOutput.log
@@ -49,7 +49,7 @@ jobs:
       # map to local user id on CI  machine to allow writing to build cache
       options: --user 1001
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
         with:
           submodules: 'true'
       - name: Configure
@@ -65,7 +65,7 @@ jobs:
           # Pick GPU with most available memory
           export CUDA_VISIBLE_DEVICES=$(nvidia-smi --query-gpu=memory.free,index --format=csv,nounits,noheader | sort -nr | head -1 | awk '{ print $NF }')
           ctest -LE 'performance|regression'
-      - uses: actions/upload-artifact@v3
+      - uses: actions/upload-artifact@v4
         with:
           name: configure-log-unit-${{ matrix.device }}
           path: build/CMakeFiles/CMakeOutput.log
@@ -81,7 +81,7 @@ jobs:
       # map to local user id on CI  machine to allow writing to build cache
       options: --user 1001
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
         with:
           submodules: 'true'
       - name: Configure
@@ -124,7 +124,7 @@ jobs:
           if grep HtoD profile.txt; then exit 1; fi
           if grep DtoH profile.txt; then exit 1; fi
 
-      - uses: actions/upload-artifact@v3
+      - uses: actions/upload-artifact@v4
         with:
           name: configure-log-integration-${{ matrix.device }}
           path: |
@@ -143,7 +143,7 @@ jobs:
       CMAKE_GENERATOR: Ninja
       CMAKE_BUILD_PARALLEL_LEVEL: 8 # num threads for build
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
         with:
           submodules: 'true'
       - name: Configure
@@ -166,10 +166,50 @@ jobs:
           cd build
           ctest -R regression_mpi_test:particle_leapfrog
 
-      - uses: actions/upload-artifact@v3
+      - uses: actions/upload-artifact@v4
         with:
           name: configure-log-integration-amdgpu
           path: |
             build/CMakeFiles/CMakeOutput.log
           retention-days: 3
 
+  integration-nvidia-hamilton:
+    runs-on: [self-hosted, navi1030]
+    container:
+      image: ghcr.io/parthenon-hpc-lab/cuda11.6-mpi-hdf5-ascent
+      # Map to local user id on CI  machine to allow writing to build cache
+      options: --user 1000 --gpus all
+    env:
+      CMAKE_GENERATOR: Ninja
+      CMAKE_BUILD_PARALLEL_LEVEL: 8 # num threads for build
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          submodules: 'true'
+      - name: Configure
+        run: |
+          cmake -B build \
+            -DMACHINE_CFG=${PWD}/cmake/machinecfg/GitHubActions.cmake \
+            -DCMAKE_BUILD_TYPE=Release \
+            -DMACHINE_VARIANT=cuda-mpi \
+            -DCMAKE_CXX_COMPILER=${PWD}/external/Kokkos/bin/nvcc_wrapper
+      # Test example with "variables" and output
+      - name: advection
+        run: |
+          nvidia-smi
+          cmake --build build -t advection-example
+          cd build
+          ctest -R regression_mpi_test:output_hdf5
+      # Test example with swarms
+      - name: particle-leapfrog
+        run: |
+          cmake --build build -t particle-leapfrog
+          cd build
+          ctest -R regression_mpi_test:particle_leapfrog
+
+      - uses: actions/upload-artifact@v4
+        with:
+          name: configure-log-integration-amdgpu
+          path: |
+            build/CMakeFiles/CMakeOutput.log
+          retention-days: 3

cmake/machinecfg/GitHubActions.cmake

@@ -19,11 +19,12 @@ message(STATUS "Loading machine configuration for GitHub Actions CI. ")
 
 # common options
 set(NUM_MPI_PROC_TESTING "2" CACHE STRING "CI runs tests with 2 MPI ranks")
+set(Kokkos_ARCH_ZEN2 ON CACHE BOOL "Enable Zen2")
 
 set(MACHINE_CXX_FLAGS "")
 if (${MACHINE_VARIANT} MATCHES "cuda")
-  # using an arbitrary arch as GitHub Action runners don't have GPUs
-  set(Kokkos_ARCH_VOLTA70 ON CACHE BOOL "GPU architecture")
+  # using an arch that matches Hamilton at Hamburg Obs
+  set(Kokkos_ARCH_AMPERE86 ON CACHE BOOL "GPU architecture")
   set(Kokkos_ENABLE_CUDA ON CACHE BOOL "Enable Cuda")
   if (${CMAKE_CXX_COMPILER} MATCHES "clang")
     set(MACHINE_CXX_FLAGS "${MACHINE_CXX_FLAGS} -Wno-unknown-cuda-version")
@@ -32,7 +33,6 @@ elseif (${MACHINE_VARIANT} MATCHES "hip")
   # using an arch that matches Hamilton at Hamburg Obs
   set(Kokkos_ARCH_NAVI1030 ON CACHE BOOL "GPU architecture")
   set(Kokkos_ENABLE_HIP ON CACHE BOOL "Enable HIP")
-  set(Kokkos_ENABLE_ZEN3 ON CACHE BOOL "Enable Zen3")
 else()
   set(MACHINE_CXX_FLAGS "${MACHINE_CXX_FLAGS} -fopenmp-simd")
 endif()