Frontier (OLCF)
The Frontier cluster is located at OLCF.
On Frontier, each compute node provides four AMD MI250X GPUs, each with two Graphics Compute Dies (GCDs) for a total of 8 GCDs per node. You can think of the 8 GCDs as 8 separate GPUs, each having 64 GB of high-bandwidth memory (HBM2E).
Introduction
If you are new to this system, please see the following resources:
Batch system: Slurm
-
$HOME: per-user directory, use only for inputs, source and scripts; backed up; mounted as read-only on compute nodes, that means you cannot run in it (50 GB quota)$PROJWORK/$proj/: shared with all members of a project, purged every 90 days, Lustre (recommended)$MEMBERWORK/$proj/: single user, purged every 90 days, Lustre (usually smaller quota, 50TB default quota)$WORLDWORK/$proj/: shared with all users, purged every 90 days, Lustre (50TB default quota)
Note: the Orion Lustre filesystem on Frontier and the older Alpine GPFS filesystem on Summit are not mounted on each others machines. Use Globus to transfer data between them if needed.
Preparation
Use the following commands to download the WarpX source code:
git clone https://github.com/BLAST-WarpX/warpx.git $HOME/src/warpx
We use system software modules, add environment hints and further dependencies via the file $HOME/frontier_warpx.profile.
Create it now:
cp $HOME/src/warpx/Tools/machines/frontier-olcf/frontier_warpx.profile.example $HOME/frontier_warpx.profile
Edit the 2nd line of this script, which sets the export proj="" variable.
For example, if you are member of the project aph114, then run vi $HOME/frontier_warpx.profile.
Enter the edit mode by typing i and edit line 2 to read:
export proj="aph114"
Exit the vi editor with Esc and then type :wq (write & quit).
Important
Now, and as the first step on future logins to Frontier, activate these environment settings:
source $HOME/frontier_warpx.profile
Finally, since Frontier does not yet provide software modules for some of our dependencies, install them once:
bash $HOME/src/warpx/Tools/machines/frontier-olcf/install_dependencies.sh
source $HOME/sw/frontier/gpu/venvs/warpx-frontier/bin/activate
Compilation
Use the following cmake commands to compile the application executable:
cd $HOME/src/warpx
rm -rf build_frontier
cmake -S . -B build_frontier -DWarpX_COMPUTE=HIP -DWarpX_FFT=ON -DWarpX_QED_TABLE_GEN=ON -DWarpX_DIMS="1;2;RZ;3"
cmake --build build_frontier -j 16
The WarpX application executables are now in $HOME/src/warpx/build_frontier/bin/.
Additionally, the following commands will install WarpX as a Python module:
rm -rf build_frontier_py
cmake -S . -B build_frontier_py -DWarpX_COMPUTE=HIP -DWarpX_FFT=ON -DWarpX_QED_TABLE_GEN=ON -DWarpX_APP=OFF -DWarpX_PYTHON=ON -DWarpX_DIMS="1;2;RZ;3"
cmake --build build_frontier_py -j 16 --target pip_install
Now, you can submit Frontier compute jobs for WarpX Python (PICMI) scripts (example scripts).
Or, you can use the WarpX executables to submit Frontier jobs (example inputs).
For executables, you can reference their location in your job script or copy them to a location in $PROJWORK/$proj/.
Update WarpX & Dependencies
If you already installed WarpX in the past and want to update it, start by getting the latest source code:
cd $HOME/src/warpx
# read the output of this command - does it look ok?
git status
# get the latest WarpX source code
git fetch
git pull
# read the output of these commands - do they look ok?
git status
git log # press q to exit
And, if needed,
log out and into the system, activate the now updated environment profile as usual,
As a last step, clean the build directory rm -rf $HOME/src/warpx/build_frontier and rebuild WarpX.
Running
MI250X GPUs (2x64 GB)
After requesting an interactive node with the getNode alias above, run a simulation like this, here using 8 MPI ranks and a single node:
runNode ./warpx inputs
Or in non-interactive runs:
#!/usr/bin/env bash
#SBATCH -A <project id>
#SBATCH -J warpx
#SBATCH -o %x-%j.out
#SBATCH -t 00:10:00
#SBATCH -p batch
#SBATCH --ntasks-per-node=8
# Due to Frontier's Low-Noise Mode Layout only 7 instead of 8 cores are available per process
# https://docs.olcf.ornl.gov/systems/frontier_user_guide.html#low-noise-mode-layout
#SBATCH --cpus-per-task=7
#SBATCH --gpus-per-task=1
#SBATCH --gpu-bind=closest
#SBATCH -N 20
# load cray libs and ROCm libs
#export LD_LIBRARY_PATH=${CRAY_LD_LIBRARY_PATH}:${LD_LIBRARY_PATH}
# From the documentation:
# Each Frontier compute node consists of [1x] 64-core AMD EPYC 7A53
# "Optimized 3rd Gen EPYC" CPU (with 2 hardware threads per physical core) with
# access to 512 GB of DDR4 memory.
# Each node also contains [4x] AMD MI250X, each with 2 Graphics Compute Dies
# (GCDs) for a total of 8 GCDs per node. The programmer can think of the 8 GCDs
# as 8 separate GPUs, each having 64 GB of high-bandwidth memory (HBM2E).
# note (5-16-22 and 7-12-22)
# this environment setting is currently needed on Frontier to work-around a
# known issue with Libfabric (both in the May and June PE)
#export FI_MR_CACHE_MAX_COUNT=0 # libfabric disable caching
# or, less invasive:
export FI_MR_CACHE_MONITOR=memhooks # alternative cache monitor
# Seen since August 2023
# OLCFDEV-1597: OFI Poll Failed UNDELIVERABLE Errors
# https://docs.olcf.ornl.gov/systems/frontier_user_guide.html#olcfdev-1597-ofi-poll-failed-undeliverable-errors
export MPICH_SMP_SINGLE_COPY_MODE=NONE
export FI_CXI_RX_MATCH_MODE=software
# note (9-2-22, OLCFDEV-1079)
# this environment setting is needed to avoid that rocFFT writes a cache in
# the home directory, which does not scale.
export ROCFFT_RTC_CACHE_PATH=/dev/null
export OMP_NUM_THREADS=1
export WARPX_NMPI_PER_NODE=8
export TOTAL_NMPI=$(( ${SLURM_JOB_NUM_NODES} * ${WARPX_NMPI_PER_NODE} ))
srun -N${SLURM_JOB_NUM_NODES} -n${TOTAL_NMPI} --ntasks-per-node=${WARPX_NMPI_PER_NODE} \
./warpx inputs > output.txt
Post-Processing
For post-processing, most users use Python via OLCFs’s Jupyter service (Docs).
Please follow the same guidance as for OLCF Summit post-processing.
Known System Issues
Warning
May 16th, 2022 (OLCFHELP-6888): There is a caching bug in Libfabric that causes WarpX simulations to occasionally hang on Frontier on more than 1 node.
As a work-around, please export the following environment variable in your job scripts until the issue is fixed:
#export FI_MR_CACHE_MAX_COUNT=0 # libfabric disable caching
# or, less invasive:
export FI_MR_CACHE_MONITOR=memhooks # alternative cache monitor
Warning
Sep 2nd, 2022 (OLCFDEV-1079): rocFFT in ROCm 5.1-5.3 tries to write to a cache in the home area by default. This does not scale, disable it via:
export ROCFFT_RTC_CACHE_PATH=/dev/null
Warning
January, 2023 (OLCFDEV-1284, AMD Ticket: ORNLA-130): We discovered a regression in AMD ROCm, leading to 2x slower current deposition (and other slowdowns) in ROCm 5.3 and 5.4.
June, 2023: Although a fix was planned for ROCm 5.5, we still see the same issue in this release and continue to exchange with AMD and HPE on the issue.
Stay with the ROCm 5.2 module to avoid a 2x slowdown.
Warning
August, 2023 (OLCFDEV-1597, OLCFHELP-12850, OLCFHELP-14253):
With runs above 500 nodes, we observed issues in MPI_Waitall calls of the kind OFI Poll Failed UNDELIVERABLE.
According to the system known issues entry OLCFDEV-1597, we work around this by setting this environment variable in job scripts:
export MPICH_SMP_SINGLE_COPY_MODE=NONE
export FI_CXI_RX_MATCH_MODE=software
Warning
Checkpoints and I/O at scale seem to be slow with the default Lustre filesystem configuration.
Please test checkpointing and I/O with short #SBATCH -q debug runs before running the full simulation.
Execute lfs getstripe -d <dir> to show the default progressive file layout.
Consider using lfs setstripe to change the striping for new files before you submit the run.
mkdir /lustre/orion/proj-shared/<your-project>/<path/to/new/sim/dir>
cd <new/sim/dir/above>
# create your diagnostics directory first
mkdir diags
# change striping for new files before you submit the simulation
# this is an example, striping 10 MB blocks onto 32 nodes
lfs setstripe -S 10M -c 32 diags
Additionally, other AMReX users reported good performance for plotfile checkpoint/restart when using
amr.plot_nfiles = -1
amr.checkpoint_nfiles = -1
amrex.async_out_nfiles = 4096 # set to number of GPUs used