DESCRIPTION

Use the tnffile command to define a target file for the trace data generated by TNF probes. By default, the tnfcollection on command creates a trace file with an internally generated file name and sets the trace collection data files to the minimum size (128 Kbytes). Use the tnffile command to override those defaults.

The filename argument refers to the permanent file that Prism fills with the merged data taken from each process's (temporary) output trace file. If you specify a file name that already exists, an error message "file already exists" is issued and the tnffile command is ignored.

When collecting TNF data, Prism creates a trace file for every process. Use the optional size argument to specify the size (in kilobytes) of the trace files. The default size is 128 Kbytes. The trace files are circular buffers--once a file has been filled, more recent trace events overwrite the oldest ones. Once the trace data collection process is complete, Prism merges all of the trace files into the output file filename, which can be as large as the number of processes * size. You can view filename with the tnfview command, or by selecting Display TNF Data from Prism's Performance menu.

Prism's MPI Performance Analysis generates large volumes of data, particularly for long-running programs or programs with high process counts. As a result the tnfview command may fail if insufficient disk space is available for storing TNF output data in /usr/tmp. To work around this restriction, limit the collection interval using the tnfcollection command or limit the types of events collected using the tnfenable command. See Chapter 6 in the Prism 6.0 User's Guide for more information about Prism's MPI Performance Analysis.