Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

enhance archdetect to support detection of NVIDIA GPUs + using that in EESSI init script #767

Merged
merged 8 commits into from
Oct 9, 2024
Merged

enhance archdetect to support detection of NVIDIA GPUs + using that in EESSI init script #767

Show file tree
Hide file tree
Changes from 2 commits
Commits
Show all changes
8 commits
Select commit Hold shift + click to select a range
67bed25
enhance archdetect to support detection of NVIDIA GPUs + using that i…
boegel Sep 30, 2024
6571035
introduce $EESSI_GPU_SOFTWARE_PATH which can be overriden via $EESSI_…
boegel Oct 3, 2024
228eaa2
use non-zero exit code in archdetect if nvidia-smi was not found or f…
boegel Oct 3, 2024
f823375
sanity check value of $EESSI_ACCELERATOR_TARGET_OVERRIDE, if it is se…
boegel Oct 3, 2024
b04cc29
take into account that nvidia-smi command may be available, but fail …
boegel Oct 3, 2024
d9bca16
only set $EESSI_MODULEPATH_ACCEL if directory exists
boegel Oct 3, 2024
24f0620
add CI workflow for testing NVIDIA GPU accelerator detection
boegel Oct 6, 2024
3bf3b12
temporarily disable errexit shell option when calling out to archdete…
boegel Oct 6, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 5 additions & 0 deletions init/bash
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -29,6 +29,11 @@ if [ $? -eq 0 ]; then
show_msg "Prepending site path $EESSI_SITE_MODULEPATH to \$MODULEPATH..."
module use $EESSI_SITE_MODULEPATH

if [ ! -z ${EESSI_MODULEPATH_ACCEL} ]; then
ocaisa marked this conversation as resolved.
Show resolved Hide resolved
show_msg "Prepending $EESSI_MODULEPATH_ACCEL to \$MODULEPATH..."
module use $EESSI_MODULEPATH_ACCEL
fi

#show_msg ""
#show_msg "*** Known problems in the ${EESSI_VERSION} software stack ***"
#show_msg ""
Expand Down
22 changes: 20 additions & 2 deletions init/eessi_archdetect.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -150,8 +150,25 @@ cpupath(){
fi
}

accelpath() {
# If EESSI_ACCELERATOR_TARGET_OVERRIDE is set, use it
log "DEBUG" "accelpath: Override variable set as '$EESSI_ACCELERATOR_TARGET_OVERRIDE' "
ocaisa marked this conversation as resolved.
Show resolved Hide resolved
[ $EESSI_ACCELERATOR_TARGET_OVERRIDE ] && echo ${EESSI_ACCELERATOR_TARGET_OVERRIDE} && exit

# check for NVIDIA GPUs via nvidia-smi command
nvidia_smi=$(command -v nvidia-smi)
if [[ $? -eq 0 ]]; then
log "DEBUG" "accelpath: nvidia-smi command found @ ${nvidia_smi}"
gpu_info=$(nvidia-smi --query-gpu=gpu_name,count,driver_version,compute_cap --format=csv,noheader | head -1)
ocaisa marked this conversation as resolved.
Show resolved Hide resolved
cuda_cc=$(echo $gpu_info | sed 's/, /,/g' | cut -f4 -d, | sed 's/\.//g')
echo "accel/nvidia/cc${cuda_cc}"
else
log "DEBUG" "accelpath: nvidia-smi command not found"
ocaisa marked this conversation as resolved.
Show resolved Hide resolved
fi
}

# Parse command line arguments
USAGE="Usage: eessi_archdetect.sh [-h][-d][-a] <action>"
USAGE="Usage: eessi_archdetect.sh [-h][-d][-a] <action: cpupath or accelpath>"

while getopts 'hdva' OPTION; do
case "$OPTION" in
Expand All @@ -168,5 +185,6 @@ ARGUMENT=${1:-none}

case "$ARGUMENT" in
"cpupath") cpupath; exit;;
*) echo "$USAGE"; log "ERROR" "Missing <action> argument (possible actions: 'cpupath')";;
"accelpath") accelpath; exit;;
*) echo "$USAGE"; log "ERROR" "Missing <action> argument (possible actions: 'cpupath', 'accelpath')";;
esac
16 changes: 16 additions & 0 deletions init/eessi_environment_variables
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -38,6 +38,16 @@ if [ -d $EESSI_PREFIX ]; then
break
fi
done
export EESSI_ACCEL_PATH=$(${EESSI_INIT_DIR_PATH}/eessi_archdetect.sh accelpath)
if [ -z ${EESSI_ACCEL_PATH} ]; then
show_msg "archdetect could not find any accelerators"
else
EESSI_GPU_SOFTWARE_SUBDIR=${$EESSI_GPU_SOFTWARE_SUBDIR_OVERRIDE:-${EESSI_SOFTWARE_SUBDIR}}
EESSI_GPU_SOFTWARE_PATH=${EESSI_PREFIX}/software/${EESSI_OS_TYPE}/${EESSI_GPU_SOFTWARE_SUBDIR}
if [ -d ${EESSI_GPU_SOFTWARE_PATH}/${EESSI_ACCEL_PATH} ]; then
show_msg "archdetect found accelerator: ${EESSI_ACCEL_PATH}"
fi
fi
elif [ "$EESSI_USE_ARCHSPEC" == "1" ]; then
# note: eessi_software_subdir_for_host.py will pick up value from $EESSI_SOFTWARE_SUBDIR_OVERRIDE if it's defined!
export EESSI_EPREFIX_PYTHON=$EESSI_EPREFIX/usr/bin/python3
Expand Down Expand Up @@ -106,6 +116,12 @@ if [ -d $EESSI_PREFIX ]; then
false
fi

EESSI_MODULEPATH_ACCEL=${EESSI_GPU_SOFTWARE_PATH}/${EESSI_ACCEL_PATH}/${EESSI_MODULE_SUBDIR}
if [ -d ${EESSI_MODULEPATH_ACCEL} ]; then
export EESSI_MODULEPATH_ACCEL=${EESSI_MODULEPATH_ACCEL}
show_msg "Using ${EESSI_MODULEPATH_ACCEL} as additional directory (for accelerators) to be added to MODULEPATH."
fi

# Fix wrong path for RHEL >=8 libcurl
# This is required here because we ship curl in our compat layer. If we only provided
# curl as a module file we could instead do this via a `modluafooter` in an EasyBuild
Expand Down