Getting an account

more to come...

ssh configuration

You can add the following lines to ~/.ssh/config on your local machine:

Host frontera.tacc.utexas.edu frontera
     HostName frontera.tacc.utexas.edu
     User YOURUSERNAME

and replace YOURUSERNAME by your TACC username.

Make sure to also include the following in your ~/.ssh/config:

Host *
   ControlMaster auto
   ControlPath ~/.ssh/sockets/%r@%h-%p
   ControlPersist yes

It will allow you to use existing ssh connection for multiple sessions. As long as you have an active connection, ssh will not need a password or token to create a new session. You may need to mkdir ~/.ssh/sockets if this directory does not exist.

Once this is done, you can ssh frontera by simply doing:

ssh frontera

Environment

On frontera, add the following lines to ~/.bash_login:

export ISSM_DIR=PATHTOTRUNK
source $ISSM_DIR/etc/environment.sh
module load petsc/3.15

Log out and log back in to apply this change.

Installing ISSM on frontera

frontera will only be used to run the code, you will use your local machine for pre and post processing, you will never use frontera's matlab. You can check out ISSM and install the following packages:

• m1qn3

Use the following configuration script (adapt to your needs):

./configure \
   --prefix=$ISSM_DIR \
   --with-wrappers=no \
   --with-mpi-include="$TACC_IMPI_INC" \
   --with-mpi-libflags="-L$TACC_IMPI_LIB/release_mt -lmpi" \
   --with-petsc-dir="$TACC_PETSC_DIR" \
   --with-petsc-arch=$ISSM_ARCH \
   --with-metis-dir="$TACC_PETSC_DIR/$PETSC_ARCH" \
   --with-mkl-libflags="-L$TACC_MKL_LIB -mkl=parallel" \
   --with-mumps-dir="$TACC_PETSC_DIR/$PETSC_ARCH" \
   --with-scalapack-dir="$TACC_PETSC_DIR/$PETSC_ARCH" \
   --with-m1qn3-dir="$ISSM_DIR/externalpackages/m1qn3/install" \
   --enable-debugging \
   --enable-development

Installing ISSM with Matlab on frontera

If you want to use frontera to process the model after running it and keep the data on frontera, you can install ISSM with Matlab interface. Before doing the following steps, you should check with TACC to make sure you have the access to Matlab on frontera, see: ​https://docs.tacc.utexas.edu/software/matlab/

You can check out ISSM and install the following packages:

• triangle
• m1qn3

You will need to use interactive mode to compile ISSM with Matlab: ​https://docs.tacc.utexas.edu/software/idev/

Use the following configuration script (adapt to your needs):

./configure \
   --prefix=$ISSM_DIR \
   --with-matlab-dir="/home1/apps/matlab/2023a/" \
   --with-triangle-dir=$ISSM_DIR/externalpackages/triangle/install \
   --with-mpi-include="$TACC_IMPI_INC" \
   --with-mpi-libflags="-L$TACC_IMPI_LIB/release_mt -lmpi" \
   --with-petsc-dir="$TACC_PETSC_DIR" \
   --with-petsc-arch=$ISSM_ARCH \
   --with-metis-dir="$TACC_PETSC_DIR/$PETSC_ARCH" \
   --with-mkl-libflags="-L$TACC_MKL_LIB -mkl=parallel" \
   --with-mumps-dir="$TACC_PETSC_DIR/$PETSC_ARCH" \
   --with-scalapack-dir="$TACC_PETSC_DIR/$PETSC_ARCH" \
   --with-m1qn3-dir="$ISSM_DIR/externalpackages/m1qn3/install" \
   --enable-debugging \
   --enable-development

You need to compile ISSM serially with make install.

Remember frontera is a remote cluster, use matlab -nodesktop -nosplash -r "addpath $ISSM_DIR/src/m/dev; devpath; when running Matlab.

Before downloading the .outbin, you will need to set ` md.cluster.name = oshostname() md.miscellaneous.name = the_file_name_of_outbin ` Then, run md=loadresultsfromcluster(md, 'runtimename', the_folder_name_in_execution)

frontera_settings.m

You have to add a file in $ISSM_DIR/src/m entitled frontera_settings.m with your personal settings on your local issm install:

cluster.login='seroussi';
cluster.codepath='/home1/03729/seroussi/trunk-jpl/bin/';
cluster.executionpath='/work/03729/seroussi/trunk-jpl/execution/';

use your username for the login and enter your code path and execution path. These settings will be picked up automatically by matlab when you do md.cluster=frontera()

Note that the `executionpath' creates temporary binary files that can be removed once the job is complete. For this reason, you can set the path to be somewhere on the $SCRATCH filesystem, which is unlimited temporary storage on frontera.

Running jobs on frontera

On frontera, each node has 56 cores and you can use any multiple of 56 for the total number of processors. The more nodes and the longer the requested time, the more you will have to wait in the queue. So choose your settings wisely:

md.cluster=frontera('numnodes',2);

Before you run your job, make sure to have an active open ssh connection to frontera so that you don't need to enter your password.

To manually submit a job on frontera, do:

sbatch job.queue

Now if you want to check the status of your job and the queue you are using, type in the bash with the frontera session:

showq -u USERNAME

You can delete your job manually by typing:

scancel JOBID

where JOBID is the ID of your job (indicated in the Matlab session). Matlab indicates too the directory of your job where you can find the files JOBNAME.outlog and JOBNAME.errlog. The outlog file contains the informations that would appear if you were running your job on your local machine and the errlog file contains the error information in case the job encounters an error.

Use DeepXDE

Fontera supports container by a software called apptainer ​https://apptainer.org. A precompiled DeepXDE image with Tensorflow v.2 backend is available at docker://chenggongdartmouth/deepxde:v1.2 or at docker://mkrish234/deepxde:v0.3 .

You might need to build an appraiser image from the Docker image in Frontera. First find the path to your login node on Frontera with pwd. Then, allocate a compute node with the following

idev -m 60 -p rtx -N 1 -n 8

You will need to load the apptainer module as follows

module load tacc-apptainer

Build the apptainer image from the Docker image as follows

apptainer build <PATH_TO_LOGIN_NODE>/deepxde docker://<DOCKER_IMAGE>

The following is an example of using DeepXDE to run a script ./test.py on Frontera (with GPU node):

#!/bin/bash

#SBATCH -J job_name           # job name
#SBATCH -o output.%j          # output file named, output.jobID
#SBATCH -e error.%j           # error file named, error.jobID
#SBATCH -p rtx                # queue name
#SBATCH -N 1                  # number of nodes requested
#SBATCH --ntasks-per-node 4   # tasks per node
#SBATCH -t 10:00:00           # time, hh:mm:ss
#SBATCH --mail-user=<EMAIL_ADDRESS>
#SBATCH --mail-type=all

module load tacc-apptainer
stdbuf -i0 -o0 -e0 apptainer exec --nv --bind ~/:/mnt ~/deepxde python -u /mnt/test.py