Abaqus Performance Summary Generation

The Abaqus performance summary generation procedure collects job performance-related data from execution output files and writes them to a single file in a format that is useful for evaluation by SIMULIA support staff. Several parameters can be set either on the command line or in the environment file (see Environment File Settings). The following file types are read when producing the summary:

  • the data (.dat) file, for model size and performance-related estimate data;

  • the message (.msg) file, for increment and iteration data; and

  • the use (.use) file, including the multiple files (.use.n) created for domain parallel Abaqus/Standard analyses for internal Abaqus procedure phase-by-phase details. The use file is written when the system environment variable ABA_RESOURCE_MONITOR is set to any value.

The results of the performance summary are written to a file with the extension xml, in the Extensible Markup Language (XML) format. The arrangement of data within this file (the file schema) is not documented.

This page discusses:

Alternative Use through the Abaqus Analysis Procedure

In addition to the use of the summarize procedure detailed in the following sections, you can direct the Abaqus analysis procedure (Abaqus/Standard and Abaqus/Explicit Execution) to automatically perform the summarize procedure when the job completes, by using the command line parameter performance_summary or by specifying performance_summary=ON.

Use the following command.

abaqus job=structure performance_summary

The procedure will then produce the file summary_summary.xml when the job is completed.

Command Summary

abaqus summarizejobjob-namerollup{allonexit}detail{basicintermediateadvanced}resourcemarks

Command Line Options

job

The value of this option specifies the name prefix of the files read for the summary generation.

If this option is omitted from the command line, you will be prompted for its value.

rollup

The value of this option specifies how the summary procedure handles use (.use) file entry and exit identifiers for program subphases that are entered multiple times during an analysis; for example, when an equilibrium iteration subphase is entered in an Abaqus/Standard analysis.

The default setting of rollup=all results in a single combined summary (a rollup) of performance measures for these repeated subphases; for example, the equilibrium iteration subphase rollup would consider all such iteration subphases in the analysis.

A setting of rollup=onexit results in a measures rollup for subphases each time a subphase parent phase is exited; for example, an equilibrium iteration subphase rollup would occur for each solution increment (the parent subphase).

detail

The value of this option specifies the amount of performance data to be extracted from the use file (.use). The default setting of detail=basic gathers data for wallclock time and peak memory usage. A setting of detail=intermediate gathers data for wallclock time, user CPU time, system CPU time, peak memory usage, and peak virtual memory usage. A setting of detail=advanced gathers data for all results available in the use file. Alternatively, you can set the environment file (Environment File Settings) parameter performance_summary_detail to a value of BASIC, INTERMEDIATE, or ADVANCED to obtain the respective summary detail level.

resourcemarks

When specified, individual subphase entry and exit records from the use file (.use) are converted to the summary file format and added to the summary file.

Examples

The following example illustrates the use of the summarize procedure following execution of an Abaqus/Standard job named “structure” where the system environment variable ABA_RESOURCE_MONITOR was set.

Running the Summarize Procedure

Use the following command:

abaqus summarize job=structure

The procedure will then read the files job.dat, job.msg, and job.use, producing the file job_summary.xml.