Merge -next with CT changes, basics working

.gitlab-ci.yml

@@ -1,76 +1,23 @@
-# Continuous Integration testing for KHARMA
-# a.k.a did we break the basics?
-# This version run on LANL Darwin
-# See .gitlab-ci-docker.yml for a generic version,
-# which can be run on any Docker runner w/GPUs
-
-variables:
-  GIT_SUBMODULE_STRATEGY: recursive
-  SCHEDULER_PARAMETERS: "-N 1 --qos=debug -p volta-x86"
-  HOST_ARCH: HSW
-  NPROC: ""
-  OMP_NUM_THREADS: 28
-  OMP_PROC_BIND: "false"
-  MPI_EXE: mpirun
-  MPI_NUM_PROCS: 2
-  HTTP_PROXY: http://proxyout.lanl.gov:8080
-  http_proxy: http://proxyout.lanl.gov:8080
-  HTTPS_PROXY: http://proxyout.lanl.gov:8080
-  https_proxy: http://proxyout.lanl.gov:8080
-  NO_PROXY: lanl.gov,localhost,127.0.0.1,0.0.0.0,::1
-  no_proxy: lanl.gov,localhost,127.0.0.1,0.0.0.0,::1
-
-### DEFAULT TEST BEHAVIOR ###
 default:
-  tags:
-    - darwin-slurm-shared
-  # Load Python
-  before_script:
-    - module load miniconda3
-
   # Always keep logs and plots.  Results should be printed to console!
   artifacts:
     when: always
     paths:
       - tests/*/*.png
       - tests/*/*.txt
 
-# Tests can be executed in parallel,
-# but be careful about GPU arch
-stages:
-  - build
-  - tests
+trigger_darwin:
+  trigger:
+    include: scripts/ci/darwin.yml
 
-# Default rules
-.default-rules:
-  rules:
-    - if: $CI_COMMIT_BRANCH == "dev"
-      when: always
-    - when: manual
-  allow_failure: false
-
-# Build, obviously overrides script/artifacts
-build:
-  extends: .default-rules
-  stage: build
-  before_script:
-    - echo "Skipping pyharm install in build."
-  script:
-    - export PREFIX_PATH=$PWD/external/hdf5
-    - ./make.sh clean cuda hdf5 volta
-  artifacts:
-    paths:
-      - kharma.*
-      - make_args
+trigger_nvhpc:
+  trigger:
+    include: scripts/ci/nvhpc.yml
 
-# Run all tests in parallel
-tests:
-  extends: .default-rules
-  stage: tests
-  script:
-    - cd tests/$TEST
-    - ./run.sh
-  parallel:
-    matrix:
-      - TEST: [bondi, bondi_viscous, bz_monopole, emhdmodes, mhdmodes, noh, regrid, reinit, restart, tilt_init, torus_sanity]
+trigger_cpu:
+  trigger:
+    include: scripts/ci/cpu.yml
+    strategy: depend
 
+# TODO trigger_cpu w/intel or similar container
+# TODO build containers here

.gitmodules

@@ -1,7 +1,7 @@
 [submodule "external/parthenon"]
 	path = external/parthenon
-	url = https://github.com/parthenon-hpc-lab/parthenon.git
-	branch = bprather/backport-bicgstab
+	url = https://github.com/AFD-Illinois/parthenon.git
+	branch = kharma
 [submodule "external/variant"]
 	path = external/variant
 	url = https://github.com/mpark/variant.git

CMakeLists.txt

@@ -7,11 +7,10 @@ cmake_minimum_required(VERSION 3.10)
 project(kharma LANGUAGES C CXX)
 
 # We follow Parthenon in requiring C++17 going forward
-#set(CMAKE_CXX_STANDARD 17)
-#set(CMAKE_CXX_STANDARD_REQUIRED ON)
-#set(CMAKE_CXX17_STANDARD_COMPILE_OPTION "-std=c++17")
-#set(PARTHENON_ENABLE_CPP17 ON CACHE BOOL "KHARMA Override")
-set(CMAKE_CXX_STANDARD 14)
+set(CMAKE_CXX_STANDARD 17)
+set(CMAKE_CXX_STANDARD_REQUIRED ON)
+set(CMAKE_CXX17_STANDARD_COMPILE_OPTION "-std=c++17")
+set(PARTHENON_ENABLE_CPP17 ON CACHE BOOL "KHARMA Override")
 
 # Set the path to include cmake/ dir
 list(APPEND CMAKE_MODULE_PATH "${CMAKE_CURRENT_SOURCE_DIR}/cmake")
@@ -27,8 +26,6 @@ set(PARTHENON_DISABLE_SPARSE ON CACHE BOOL "KHARMA Override")
 # Parthenon internal build options
 set(BUILD_TESTING OFF CACHE BOOL "KHARMA Override")
 set(ENABLE_COMPILER_WARNINGS OFF CACHE BOOL "KHARMA Override")
-# TODO upstream Parthenon needs support before we do this
-#set(COORDINATE_TYPE GRCoordinates)
 # Always use static HDF5
 set(HDF5_USE_STATIC_LIBRARIES ON CACHE BOOL "KHARMA Override")
 
@@ -39,9 +36,9 @@ set(Kokkos_ENABLE_CUDA_CONSTEXPR ON CACHE BOOL "KHARMA Override")
 set(Kokkos_ENABLE_HWLOC OFF CACHE BOOL "KHARMA Override") # Possible speed improvement?
 set(Kokkos_ENABLE_AGGRESSIVE_VECTORIZATION ON CACHE BOOL "KHARMA Override")
 
-# If we build KokkosKernels at all, disable the extras
-set(KokkosKernels_ENABLE_TPL_CUSPARSE OFF CACHE BOOL "KHARMA Override")
-set(KokkosKernels_ENABLE_TPL_CUBLAS OFF CACHE BOOL "KHARMA Override")
+include(GetGitRevisionDescription)
+get_git_head_revision(GIT_REFSPEC GIT_SHA1)
+git_describe_working_tree(GIT_VERSION --tags)
 
 # Offer a KHARMA option to disable the MPI requirement
 # The only difference from setting PARTHENON_DISABLE_MPI is that