Abaqus/en: Difference between revisions

[https://www.3ds.com/products-services/simulia/products/abaqus/ Abaqus FEA] is a software suite for finite element analysis and computer-aided engineering.

Abaqus software modules are available on our clusters; however, you must provide your own license. To configure your account on a cluster, log in and create a file named <code>$HOME/.licenses/abaqus.lic</code> containing the following two lines which support versions 202X and 6.14.1 respectively. Next, replace <code>port@server</code> with the flexlm port number and server IP address (or fully qualified hostname) of your Abaqus license server.

You will then be sent a list of cluster IP addresses so that your administrator can open the local server firewall to allow connections from the cluster on both ports. Please note that a special license agreement must generally be negotiated and signed by SIMULIA and your institution before a local  license may be used remotely on Alliance hardware.

Below are prototype Slurm scripts for submitting thread and mpi-based parallel simulations to single or multiple compute nodes.  Most users will find it sufficient to use one of the <b>project directory scripts</b> provided in the <i>Single node computing</i> sections. The optional <code>memory=</code> argument found in the last line of the scripts is intended for larger memory or problematic jobs where 3072MB offset value may require tuning.  A listing of all Abaqus command line arguments can be obtained by loading an Abaqus module and running: <code>abaqus -help | less</code>.

Jobs that require large memory or larger compute resources (beyond that which a single compute node can provide) should use the mpi scripts in the <b>multiple node sections</b> below to distribute computing over arbitrary node ranges determined automatically by the scheduler.  Short scaling test jobs should be run to determine wall-clock times (and memory requirements) as a function of the number of cores (2, 4, 8, etc.) to determine the optimal number before running any long jobs.  

Abaqus solvers support thread-based and mpi-based parallelization.  Scripts for each type are provided below for running Standard Analysis type jobs on Single or Multiple nodes respectively.  Scripts to perform multiple node job restarts are not currently provided.

<tabs>

<tab name="project directory script">

module load abaqus/2021

unset SLURM_GTIDS

export MPI_IC_ORDER='tcp'

</tabs>

Users with large memory or compute needs (and correspondingly large licenses) can use the following script to perform mpi-based computing over an arbitrary range of nodes ideally left to the scheduler to  automatically determine.  A companion template script to perform restart multi-node jobs is not currently provided due to additional limitations when they can be used.

}}

Abaqus solvers support thread-based and mpi-based parallelization.  Scripts for each type are provided below for running explicit analysis type jobs on single or multiple nodes respectively.  Template scripts to perform multi-node job restarts are not currently provided pending further testing.

<tabs>

</tabs>

{{File

   |name="scriptep1-mpi.txt"

== Node memory ==

An estimate for the total slurm node memory (--mem=) required for a simulation to run fully in ram (without being virtualized to scratch disk) can be obtained by examining the Abaqus output <code>test.dat</code> file.  For example, a simulation that requires a fairly large amount of memory might show:

To run your simulation interactively and monitor the memory consumption, do the following:

To completely satisfy the recommended "MEMORY TO OPERATIONS REQUIRED MINIMIZE I/O" (MRMIO) value, at least the same amount of non-swapped physical memory (RES) must be available to Abaqus.  Since the RES will in general be less than the virtual memory (VIRT) by some relatively constant amount for a given simulation, it is necessary to slightly over-allocate the requested Slurm node memory <code>-mem=</code>.  In the above sample Slurm script, this over-allocation has been hardcoded to a conservative value of 3072MB based on initial testing of the standard Abaqus solver.  To avoid long queue wait times associated with large values of MRMIO, it may be worth investigating the simulation performance impact associated with reducing the RES memory that is made available to Abaqus significantly below the MRMIO.  This can be done by lowering the <code>-mem=</code> value which in turn will set an artificially low value of <code>memory=</code> in the Abaqus command (found in the last line of the slurm script).  In doing this one should be careful the RES does not dip below the MINIMUM MEMORY REQUIRED (MMR) otherwise Abaqus will exit due to Out of Memory (OOM).  As an example, if your MRMIO is 96GB try running a series of short test jobs with <code>#SBATCH --mem=8G, 16G, 32G, 64G</code> until an acceptable minimal performance impact is found, noting that smaller values will result in increasingly larger scratch space used by temporary files.

= Graphical use =

Abaqus/2020 can be run interactively in graphical mode on a cluster or gra-vdi using VNC by following these steps:

# Open a new terminal window and enter one of the following:<br><code>module load StdEnv/2016 abaqus/6.14.1</code>, or,<br><code>module load StdEnv/2016 abaqus/2020</code>, or,<br><code>module load StdEnv/2020 abaqus/2021</code>

# <code>abaqus cae -mesa</code><br><br>

# Connect to gra-vdi (24hr abaqus runtime limit) with [https://docs.computecanada.ca/wiki/VNC#VDI_Nodes TigerVNC]

# Open a new terminal window and enter one of the following:<br><code>module load CcEnv StdEnv/2016 abaqus/6.14.1</code>, or,<br><code>module load CcEnv StdEnv/2016 abaqus/2020</code>, or,<br><code>module load CcEnv StdEnv/2020 abaqus/2021</code>

o <b>Checking license availability</b>

There must be at least 1 license free (not in use) for <code>abaqus cae</code> to start according to:

  abaqus licensing lmstat -c $ABAQUSLM_LICENSE_FILE -a | grep "Users of cae"

</source>

SHARCNET provides a small but free license consisting of 2 cae and 35 execute tokens where usage limits are imposed 10 tokens/user and 15 tokens/group.  For groups that have purchased dedicated tokens, the free token usage limits are added to their reservation.  The free tokens are available on a first come first serve basis and mainly intended for testing and light usage before deciding whether or not to purchase dedicated tokens.  The costs for dedicated tokens are approximately CAD$110 per compute token and CAD$400 per GUI token: submit a ticket to request an official quote.  The license can be used by any Alliance researcher, but only on SHARCNET hardware.  Groups that purchase dedicated tokens to run on the SHARCNET license server may likewise only use them on SHARCNET hardware including gra-vdi (for running Abaqus in full graphical mode) and Graham or Dusky clusters (for submitting compute batch jobs to the queue).  Before you can use the license you must contact [[Technical support]] and request access.  In your email 1) mention that it is for use on SHARCNET systems and 2) include a copy/paste of the following <code>License Agreement</code> statement with your full name and username entered in the indicated locations.  Please note that every user must do this it cannot be done one time only for a group; this includes PIs who have purchased their own dedicated tokens.

<b>o  License agreement</b>

<pre>----------------------------------------------------------------------------------

Subject: Abaqus SHARCNET Academic License User Agreement

<b>o Configure license file</b>

Configure your license file as follows, noting that it is only usable on SHARCNET systems: Graham, gra-vdi and Dusky.

<source lang="bash">

[gra-login1:~] cat ~/.licenses/abaqus.lic

If your Abaqus jobs fail with the error message [*** ABAQUS/eliT_CheckLicense rank 0 terminated by signal 11 (Segmentation fault)] in the slurm output file, verify if your <code>abaqus.lic</code> file contains ABAQUSLM_LICENSE_FILE to use abaqus/2020.  If your Abaqus jobs fail with an error message starting [License server machine is down or not responding, etc.] in the output file verify your <code>abaqus.lic</code> file contains LM_LICENSE_FILE to use abaqus/6.14.1 as shown.  The <code>abaqus.lic</code> file shown contains both so you should not see this problem.

<b>o Query license server</b>

I) To check the SHARCNET license server for started and queued jobs by username, run:

<source lang="bash">

ssh graham.computecanada.ca

abaqus licensing lmstat -c $LM_LICENSE_FILE -a | grep "Users\|start\|queued\|RESERVATIONs"

</source>

II) To check the SHARCNET license server for reservations of products by purchasing groups, run:

<source lang="bash">

ssh graham.computecanada.ca

abaqus licensing lmstat -c $LM_LICENSE_FILE -a | grep "Users\|RESERVATIONs"

</source>

III) To check the SHARCNET license server for license usage of the cae, standard and explicit products, run:

<source lang="bash">

ssh graham.computecanada.ca

abaqus licensing lmstat -c $LM_LICENSE_FILE -a | grep "Users of" | grep "cae\|standard\|explicit"

</source>

When the output of query I) above indicates that a job for a particular username is queued this means the job has entered the "R"unning state from the perspective of <code>squeue -j jobid</code> or <code>sacct -j jobid</code> and is therefore idle on a compute node waiting for a license.  This will have the same impact on your account priority as if the job were performing computations and consuming CPU time.  Eventually when sufficient licenses come available the queued job will start.  To demonstrate, the following shows the license server and queue output for the situation where a user submits two jobs, but only the first job acquires enough licenses to start:

           JOBID    USER      ACCOUNT          NAME  ST  TIME_LEFT  NODES  CPUS  MIN_MEM  NODELIST (REASON)  

           29801  roberpj  def-roberpj  scriptep1.txt  R    2:59:18      1    12      8G  dus47 (None)  

           29802  roberpj  def-roberpj  scriptsp1.txt  R    2:59:33      1    12      8G  dus28 (None)  

   Users of abaqus:  (Total of 78 licenses issued;  Total of 71 licenses in use)

       roberpj dus47 /dev/tty (v62.2) (license3.sharcnet.ca/27050 275), start Thu 8/27 5:45, 14 licenses

<b>o Specify job resources</b>

To ensure optimal usage of both your Abaqus tokens and our resources, it's important to carefully specify the required memory and ncpus in your Slurm script.  The values can be determined by submitting a few short test jobs to the queue then checking their utilization.  For <b>completed</b> jobs use <code>seff JobNumber</code> to show the total <i>Memory Utilized</i> and <i>Memory Efficiency</i>. If the <i>Memory Efficiency</i> is less than ~90%, decrease the value of the <code>#SBATCH --mem=</code> setting in your Slurm script accordingly.  Notice that the <code>seff JobNumber</code> command also shows the total <i>CPU (time) Utilized</i> and <i>CPU Efficiency</i>. If the <i>CPU Efficiency</i> is less than ~90%, perform scaling tests to determine the optimal number of CPUs for optimal performance and then update the value of <code>#SBATCH --cpus-per-task=</code> in your Slurm script.  For <b>running</b> jobs, use the <code>srun --jobid=29821580 --pty top -d 5 -u $USER</code> command to watch the %CPU, %MEM and RES for each Abaqus parent process on the compute node. The %CPU and %MEM columns display the percent usage relative to the total available on the node while the RES column shows the per process resident memory size (in human readable format for values over 1GB). Further information regarding how to [Running_jobs#Monitoring_jobs monitor jobs] is available on our documentation wiki

<b>o Core token mapping</b>

<pre>

TOKENS 5  6  7  8  10  12  14  16  19  21  25  28  34  38

CORES  1  2  3  4  6  8  12  16  24  32  48  64  96 128

</pre>

where TOKENS = floor[5 X CORES^0.422]

== Western license ==

The Western site license may only be used by Western researchers on hardware located at Western's campus.  Currently, the Dusky cluster is the only system that satisfies these conditions. Graham and gra-vdi are excluded since they are located on Waterloo's campus.  Contact the Western Abaqus license server administrator <jmilner@robarts.ca> to inquire about using the Western Abaqus license.  You will need to provide your username and possibly make arrangements to purchase tokens.  If you are granted access then you may proceed to configure your <code>abaqus.lic</code> file to point to the Western license server as follows:

<b>o Configure license file</b>

prepend_path("ABAQUSLM_LICENSE_FILE","27000@license4.sharcnet.ca")

</source>

Once configured, submit your job as described in the <i>Cluster job submission</i> section above.  If there are any problems submit a problem ticket to [[Technical support|technical support]].  Specify that you are using the Abaqus Western license on dusky and provide the failed job number along with a paste of any error messages as applicable.

The full Abaqus documentation (latest version) can be accessed on gra-vdi as shown in the following steps.