Add remaining <compiler>-<backend>-env recipes currently used

.github/workflows/ghci-snl.yaml

@@ -72,6 +72,7 @@ jobs:
           labels: ${{ steps.meta.outputs.labels }}
           cache-from: type=gha
           cache-to: type=gha,mode=max
+
   compiler:
     runs-on: ubuntu-latest
     needs: base
@@ -136,13 +137,27 @@ jobs:
             BASE_TAG=${{ github.event_name == 'pull_request' && format('base-pr-{0}', github.event.pull_request.number) || 'base' }}
           cache-from: type=gha,scope=${{ matrix.compiler }}
           cache-to: type=gha,scope=${{ matrix.compiler }},mode=max
-  gnu-cpu-env:
+
+  final-env:
     runs-on: ubuntu-latest
     needs: compiler
     permissions:
       contents: read
       packages: write
       id-token: write
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - compiler: gnu
+            tag: gnu-cpu-env
+            version: 13.3.0
+          - compiler: gnu
+            tag: gnu-cuda-env
+            version: 13.3.0
+          - compiler: intel
+            tag: intel-cpu-env
+            version: 2024.1.0
 
     steps:
       - name: Checkout
@@ -168,22 +183,22 @@ jobs:
         with:
           images: ghcr.io/${{ github.repository_owner }}/e3sm-ghci-snl
           tags: |
-            type=raw,value=gnu-cpu-env,enable=${{ github.ref == 'refs/heads/main' }}
-            type=ref,event=pr,prefix=gnu-cpu-env-pr-
-            type=ref,event=branch,prefix=gnu-cpu-env-
-            type=match,pattern=ghci-snl-(.*),group=1,prefix=gnu-cpu-env
+            type=raw,value=${{ matrix.tag }},enable=${{ github.ref == 'refs/heads/main' }}
+            type=ref,event=pr,prefix=${{ matrix.tag }}-pr-
+            type=ref,event=branch,prefix=${{ matrix.tag }}-
+            type=match,pattern=ghci-snl-(.*),group=1,prefix=${{ matrix.tag }}-
           labels: |
-            org.opencontainers.image.description=GNU env image for E3SM ghci-snl testing on CPU on GitHub actions
+            org.opencontainers.image.description=${{ matrix.tag }} image for E3SM ghci-snl testing on CPU on GitHub actions
 
       - name: Push
         uses: docker/build-push-action@v6
         with:
-          context: ghci-snl/env/gnu-cpu
-          file: ghci-snl/env/gnu-cpu/Dockerfile
+          context: ghci-snl/final/${{ matrix.tag }}
+          file: ghci-snl/final/${{ matrix.tag }}/Dockerfile
           push: ${{ github.event_name != 'pull_request' || github.event.pull_request.head.repo.full_name == github.repository }}
           tags: ${{ steps.meta.outputs.tags }}
           labels: ${{ steps.meta.outputs.labels }}
           build-args: |
-            BASE_TAG=${{ github.event_name == 'pull_request' && format('gnu-13.3.0-pr-{0}', github.event.pull_request.number) || 'gnu-13.3.0' }}
-          cache-from: type=gha
-          cache-to: type=gha,mode=max
+            BASE_TAG=${{ matrix.compiler }}-${{ matrix.version }}${{ github.event_name == 'pull_request' && format('-pr-{0}', github.event.pull_request.number) || '' }}
+          cache-from: type=gha,scope=${{ matrix.tag }}
+          cache-to: type=gha,scope=${{ matrix.tag }},mode=max

ghci-snl/base/Dockerfile

@@ -1,18 +1,22 @@
 FROM registry.access.redhat.com/ubi9-minimal:latest
 
-# 1. Setup EPEL repository
+# 1. Setup certificates
 RUN microdnf update -y && \
     microdnf install -y ca-certificates && \
     update-ca-trust && \
     microdnf clean all && \
     rm -rf /var/cache/dnf
 
+ENV SSL_CERT_DIR=/etc/pki/tls/certs \
+    SSL_CERT_FILE=/etc/pki/tls/cert.pem
+
+# 2. Setup EPEL repository
 RUN rpm -ivh https://dl.fedoraproject.org/pub/epel/epel-release-latest-9.noarch.rpm && \
     sed -i 's/enabled=1/enabled=0/g' /etc/yum.repos.d/epel*.repo && \
     microdnf clean all && \
     rm -rf /var/cache/dnf
 
-# 2. Core build tools & compilers
+# 3. Core build tools & compilers
 RUN microdnf -y install \
         binutils \
         bzip2-devel \
@@ -29,15 +33,15 @@ RUN microdnf -y install \
     && microdnf clean all \
     && rm -rf /var/cache/dnf
 
-# 3. Scripting runtimes (python & perl)
+# 4. Scripting runtimes (python & perl)
 RUN microdnf -y install \
         perl-devel \
         python3 \
         python3-pip \
     && microdnf clean all \
     && rm -rf /var/cache/dnf
 
-# 4. Utilities, archiving & environment tools
+# 5. Utilities, archiving & environment tools
 RUN microdnf -y --enablerepo=epel install \
         bind-utils \
         bzip2 \
@@ -52,12 +56,13 @@ RUN microdnf -y --enablerepo=epel install \
         sudo \
         tar \
         unzip \
+        vi \
         wget \
         xz \
     && microdnf clean all \
     && rm -rf /var/cache/dnf
 
-# 5. Locales configuration
+# 6. Locales configuration
 RUN microdnf -y install \
         glibc-langpack-en \
     && microdnf clean all \
@@ -68,7 +73,7 @@ ENV LANG=en_US.UTF-8 \
     LC_CTYPE=en_US.UTF-8 \
     LANGUAGE=en_US.UTF-8
 
-# 6. Setup SPACK
+# 7. Setup SPACK
 ARG SPACK_VER=v0.23.1
 ENV SPACK_ROOT=/spack \
     SPACK_ARCH=linux-rhel9-x86_64 \
@@ -82,10 +87,11 @@ ADD spack-configuration/packages.yaml /spack-configuration/packages.yaml
 ADD spack-configuration/config.yaml /spack-configuration/config.yaml
 ADD spack-configuration/mirrors.yaml /spack-configuration/mirrors.yaml
 
-# 7. Setup project directory & user shell environment
-RUN mkdir -p /projects/e3sm/software /projects/e3sm/data /opt/share && \
-    echo 'bind "\"\\e[A\": history-search-backward"' > /opt/share/bash-goodies.sh && \
-    echo 'bind "\"\\e[B\": history-search-forward"' >> /opt/share/bash-goodies.sh && \
-    echo 'git config --global url."https://github.com/".insteadOf git@github.com:' >> /opt/share/bash-goodies.sh
+# 8. Setup project directory & user shell environment
+RUN mkdir -p /projects/e3sm/software /projects/e3sm/data
+
+RUN echo 'bind "\"\\e[A\": history-search-backward"' > /etc/profile.d/eamxx-env.sh && \
+    echo 'bind "\"\\e[B\": history-search-forward"' >> /etc/profile.d/eamxx-env.sh && \
+    echo 'git config --global url."https://github.com/".insteadOf git@github.com:' >> /etc/profile.d/eamxx-env.sh
 
 CMD [ "/bin/bash" ]

ghci-snl/compiler/Dockerfile

@@ -48,7 +48,7 @@ RUN source /spack/share/spack/setup-env.sh && source /etc/profile.d/compiler_nam
 #       to be re-installed with diff compiler, and lmod to reload the proper version.
 #       This is 100% fine, but would produce rather lengthy warnings.
 RUN source /etc/profile.d/compiler_name.sh && \
-    echo "source /etc/profile.d/modules.sh" > /etc/profile.d/eamxx-env.sh &&  \
+    echo "source /etc/profile.d/modules.sh" >> /etc/profile.d/eamxx-env.sh &&  \
     echo "MODULEPATH=/modulefiles/${SPACK_ARCH}/Core" >> /etc/profile.d/eamxx-env.sh && \
     echo "export LMOD_EXPERT=1" >> /etc/profile.d/eamxx-env.sh && \
     echo "export LMOD_DISABLE_WARNINGS=yes" >> /etc/profile.d/eamxx-env.sh && \

ghci-snl/env/gnu-cpu/Dockerfile → ghci-snl/final/gnu-cpu-env/Dockerfile

@@ -1,4 +1,4 @@
-ARG BASE_TAG=base
+ARG BASE_TAG=UNSET
 ARG BASE_IMAGE=ghcr.io/e3sm-project/e3sm-ghci-snl
 FROM ${BASE_IMAGE}:${BASE_TAG}
 
@@ -67,7 +67,7 @@ RUN microdnf install -y autoconf automake libtool && \
 
 RUN . /etc/profile.d/eamxx-env.sh && \
     git clone https://github.com/e3sm-project/e3sm /tmp/e3sm && \
-    cd /tmp/e3sm && . /opt/share/bash-goodies.sh && git submodule update --init cime
+    cd /tmp/e3sm && git submodule update --init cime
 
 RUN . /etc/profile.d/eamxx-env.sh && \
     curl -fsSL "https://bitbucket.org/fathomteam/moab/raw/vijaysm/automated-build-scripts/install/install-bootstrap.sh?cb=$RANDOM" > /tmp/install-bootstrap.sh && \

ghci-snl/final/gnu-cuda-env/Dockerfile

@@ -0,0 +1,69 @@
+ARG BASE_TAG=UNSET
+ARG NVIDIA_DRIVER_VERSION=535
+ARG BASE_IMAGE=ghcr.io/e3sm-project/e3sm-ghci-snl
+FROM ${BASE_IMAGE}:${BASE_TAG}
+
+# Install packages in chunks, so we can tweak versions of some pkg without
+# rebuilding the whole image. Hence, put "stable" pkgs first to minimize rebuild time
+# NOTE: the env vars COMPILER, COMPILER_VERSION, and SPACK_ARCH are defined in the base image
+
+# Common utilities
+RUN source /spack/share/spack/setup-env.sh && \
+    spack install yaml-cpp cmake %${COMPILER}@${COMPILER_VERSION} arch=${SPACK_ARCH}
+
+# Common scientific libs (non-MPI dependent)
+RUN source /spack/share/spack/setup-env.sh && \
+    spack install boost netlib-lapack %${COMPILER}@${COMPILER_VERSION} arch=${SPACK_ARCH}
+
+# CUDA
+RUN source /spack/share/spack/setup-env.sh && \
+    spack install cuda@12.4.0 %${COMPILER}@${COMPILER_VERSION} arch=${SPACK_ARCH}
+
+# MPI
+RUN source /spack/share/spack/setup-env.sh && \
+    spack install mpich@4.1.1 %${COMPILER}@${COMPILER_VERSION} arch=${SPACK_ARCH}
+
+# Netcdf libraries
+RUN source /spack/share/spack/setup-env.sh && \
+    spack install netcdf-c+mpi netcdf-fortran parallel-netcdf %${COMPILER}@${COMPILER_VERSION} arch=${SPACK_ARCH}
+
+# Make sure the modules for the installed packages are all available
+RUN source /spack/share/spack/setup-env.sh && \
+    spack module lmod refresh -y --delete-tree
+
+# Generate the profile.d script to load them all at bash startup
+RUN echo "module load yaml-cpp cmake" >> /etc/profile.d/eamxx-env.sh
+RUN echo "module load boost netlib-lapack" >> /etc/profile.d/eamxx-env.sh
+RUN echo "module load cuda" >> /etc/profile.d/eamxx-env.sh
+RUN echo "module load mpich" >> /etc/profile.d/eamxx-env.sh
+RUN echo "module load netcdf-c netcdf-fortran parallel-netcdf" >> /etc/profile.d/eamxx-env.sh
+
+# Install numpy/cupy
+RUN source /etc/profile.d/modules.sh        &&  \
+    source /etc/profile.d/eamxx-env.sh &&  \
+    LIBRARY_PATH="$CUDA_HOME/lib64/stubs:$LIBRARY_PATH" \
+    python3 -m pip install numpy cupy
+
+# Install pytorch
+RUN source /etc/profile.d/modules.sh        &&  \
+    source /etc/profile.d/eamxx-env.sh &&  \
+    python3 -m pip install                      \
+            --extra-index-url https://download.pytorch.org/whl/cu124 \
+            torch==2.6.0+cu124
+
+# Whomever will run the container will be able to pip install pkgs in the venv
+# After all, the container is ephemeral/disposable, so there's no persistent state
+RUN chmod -R a+rwX /projects/e3sm/software/eamxx-venv
+
+# Clean up garbage packages that were only needed to BUILD other pkgs (not to USE them)
+RUN source /spack/share/spack/setup-env.sh && spack gc -y
+
+# Remove spack repo (except .git, so you can restore it) and spack buildcache (to keep image size under control)
+RUN rm -rf /spack-buildcache && \
+    rm -rf /spack/*
+
+# CIME tests need a bunch of perl
+RUN microdnf install -y perl
+
+ENTRYPOINT []
+CMD [ "/bin/bash", "--login" ]

ghci-snl/final/intel-cpu-env/Dockerfile

@@ -0,0 +1,58 @@
+ARG BASE_TAG=UNSET
+ARG BASE_IMAGE=ghcr.io/e3sm-project/e3sm-ghci-snl
+FROM ${BASE_IMAGE}:${BASE_TAG}
+
+# Install packages in chunks, so we can tweak versions of some pkg without
+# rebuilding the whole image. Hence, put "stable" pkgs first to minimize rebuild time
+# NOTE: the env vars COMPILER, COMPILER_VERSION, and SPACK_ARCH are defined in the base image
+
+# Common utilities
+RUN source /spack/share/spack/setup-env.sh && \
+    spack install yaml-cpp cmake %${COMPILER}@${COMPILER_VERSION} arch=${SPACK_ARCH}
+
+# Common scientific libs (non-MPI dependent)
+RUN source /spack/share/spack/setup-env.sh && \
+    spack install boost intel-oneapi-mkl %${COMPILER}@${COMPILER_VERSION} arch=${SPACK_ARCH}
+
+# MPI and Netcdf libraries
+RUN source /spack/share/spack/setup-env.sh && \
+    spack install openmpi@4.1.4 netcdf-c+mpi netcdf-fortran parallel-netcdf %${COMPILER}@${COMPILER_VERSION} arch=${SPACK_ARCH}
+
+# Debugging libs
+RUN source /spack/share/spack/setup-env.sh && \
+    spack install valgrind@3.20~mpi %${COMPILER}@${COMPILER_VERSION} arch=${SPACK_ARCH}
+
+# Make sure the modules for the installed packages are all available
+RUN source /spack/share/spack/setup-env.sh && \
+    spack module lmod refresh -y --delete-tree
+
+# Generate the profile.d script to load them all at bash startup
+RUN echo "module load yaml-cpp cmake" >> /etc/profile.d/eamxx-env.sh
+RUN echo "module load boost intel-oneapi-mkl" >> /etc/profile.d/eamxx-env.sh
+RUN echo "module load openmpi" >> /etc/profile.d/eamxx-env.sh
+RUN echo "module load netcdf-c netcdf-fortran parallel-netcdf" >> /etc/profile.d/eamxx-env.sh
+RUN echo "module load valgrind" >> /etc/profile.d/eamxx-env.sh
+
+# Install pytorch
+RUN source /etc/profile.d/modules.sh        &&  \
+    source /etc/profile.d/eamxx-env.sh &&  \
+    python3 -m pip install                      \
+            --extra-index-url https://download.pytorch.org/whl/cpu \
+            torch numpy
+
+# Whomever will run the container will be able to pip install pkgs in the venv
+# After all, the container is ephemeral/disposable, so there's no persistent state
+RUN chmod -R a+rwX /projects/e3sm/software/eamxx-venv
+
+# Clean up garbage packages that were only needed to BUILD other pkgs (not to USE them)
+RUN source /spack/share/spack/setup-env.sh && spack gc -y
+
+# Remove spack repo (except .git, so you can restore it) and spack buildcache (to keep image size under control)
+RUN rm -rf /spack-buildcache && \
+    rm -rf /spack/*
+
+# CIME tests need a bunch of perl
+RUN microdnf install -y perl
+
+ENTRYPOINT []
+CMD [ "/bin/bash", "--login" ]