add ci to gpu

Author: arciyer123

.github/workflows/phoenix/submit.sh

@@ -6,21 +6,23 @@ usage() {
     echo "Usage: $0 [script.sh] [cpu|gpu]"
 }
 
-if [ ! -z "$1" ]; then
-    sbatch_script_contents=`cat $1`
-else
+if [ -z "$1" ] || [ -z "$2" ]; then
     usage
     exit 1
 fi
 
+sbatch_script_contents=`cat $1`
+
+precision="${3:-}"
+
 sbatch_cpu_opts="\
 #SBATCH -p cpu-small               # partition
 #SBATCH --ntasks-per-node=24       # Number of cores per node required
 #SBATCH --mem-per-cpu=2G           # Memory per core\
 "
 
 sbatch_gpu_opts="\
-#SBATCH -CV100-16GB
+#SBATCH -C V100-16GB
 #SBATCH -G2\
 "
 
@@ -35,25 +37,30 @@ fi
 
 job_slug="`basename "$1" | sed 's/\.sh$//' | sed 's/[^a-zA-Z0-9]/-/g'`-$2"
 
+if [ -n "$precision" ]; then
+    job_slug="$job_slug-$precision"
+fi
+
 sbatch <<EOT
 #!/bin/bash
 #SBATCH -Jshb-$job_slug            # Job name
 #SBATCH --account=gts-sbryngelson3 # charge account
 #SBATCH -N1                        # Number of nodes required
 $sbatch_device_opts
-#SBATCH -t 02:00:00                # Duration of the job (Ex: 15 mins)
+#SBATCH -t 02:00:00                # Duration of the job (Ex: 2 hours)
 #SBATCH -q embers                  # QOS Name
-#SBATCH -o$job_slug.out            # Combined output and error messages file
+#SBATCH -o $job_slug.out           # Combined output and error messages file
 #SBATCH -W                         # Do not exit until the submitted job terminates.
 
 set -e
 set -x
 
 cd "\$SLURM_SUBMIT_DIR"
-echo "Running in $(pwd):"
+echo "Running in \$(pwd):"
 
 job_slug="$job_slug"
 job_device="$2"
+precision="$precision"
 
 . ./mfc.sh load -c p -m $2
 

.github/workflows/phoenix/test.sh

@@ -4,8 +4,12 @@ build_opts=""
 if [ "$job_device" == "gpu" ]; then
     build_opts="--gpu"
 fi
+precision_flag = ""
+if ["$precision" == "single"]; then
+    precision_flag = "--single"
+fi
 
-./mfc.sh build -j 8 $build_opts
+./mfc.sh build -j 8 $build_opts $precision_flag
 
 n_test_threads=8
 
@@ -16,6 +20,6 @@ if [ "$job_device" == "gpu" ]; then
     n_test_threads=`expr $gpu_count \* 2`
 fi
 
-./mfc.sh test --max-attempts 3 -a -j $n_test_threads $device_opts -- -c phoenix
+./mfc.sh test --max-attempts 3 -a -j $n_test_threads $device_opts $precision_flag -- -c phoenix
 
 

.github/workflows/test.yml

@@ -36,6 +36,13 @@ jobs:
         include:
           - os:    ubuntu
             mpi:   no-mpi
+            precision: 
+            debug: no-debug
+            intel: false
+
+        include:
+          - os:    ubuntu
+            mpi:   mpi
             precision: single
             debug: no-debug
             intel: false
@@ -106,9 +113,14 @@ jobs:
       matrix:
         device: ['cpu', 'gpu']
         lbl: ['gt', 'frontier']
+        precision: ['']
         exclude:
           - device:   cpu
             lbl: frontier
+        include:
+          - device: gpu
+            lbl: gt
+            precision: single
     runs-on:
       group:  phoenix
       labels: ${{ matrix.lbl }}
@@ -121,23 +133,23 @@ jobs:
 
       - name: Build & Test
         if:   matrix.lbl == 'gt'
-        run:  bash .github/workflows/phoenix/submit.sh .github/workflows/phoenix/test.sh ${{ matrix.device }}
+        run:  bash .github/workflows/phoenix/submit.sh .github/workflows/phoenix/test.sh ${{ matrix.device }} ${{ matrix.precision }}
 
       - name: Build
         if:   matrix.lbl == 'frontier'
         run:  bash .github/workflows/frontier/build.sh
 
       - name: Test
         if:   matrix.lbl == 'frontier'
-        run:  bash .github/workflows/frontier/submit.sh .github/workflows/frontier/test.sh ${{matrix.device}}
+        run:  bash .github/workflows/frontier/submit.sh .github/workflows/frontier/test.sh ${{matrix.device}} ${{ matrix.precision }}
 
       - name: Print Logs
         if:   always()
-        run:  cat test-${{ matrix.device }}.out
+        run:  cat test-${{ matrix.device }}-${{ matrix.precision }}.out
 
       - name: Archive Logs
         uses: actions/upload-artifact@v4
         if:   always()
         with:
-          name: logs-${{ strategy.job-index }}-${{ matrix.device }}
-          path: test-${{ matrix.device }}.out
+          name: logs-${{ strategy.job-index }}-${{ matrix.device }}-${{ matrix.precision }}
+          path: test-${{ matrix.device }}-${{ matrix.precision }}.out