mampac — Sequentially execute multiple input files.
The mampac utility is useful for executing a group of
AMPAC input files or even an entire test suite of input files.
Mampac runs in the background and AMPAC jobs
are executed sequentially. A simple summary of the each AMPAC job is
written to the file MAMPAC.LOG
. To monitor the progress of
mampac while it is running, use tail -f MAMPAC.LOG in
the same directory in which mampac is running.
AMPAC 9 no longer supports version control. (In previous versions, AMPAC would avoid overwritting previous results files by appending a version number to each of the result files.) Care should be taken in running jobs to not overwrite previous output files associated with the input file.
Mampac can be run in three different modes.
Run all AMPAC input files (*.dat
) in
the current directory (mampac).
Recursively run all input files (*.dat
) in the current
directory and all sub-directories (mampac -r).
Run a select group of files. Wildcards are expanded allowing easy specification of groups of files. For example, to run all jobs in the current directory starting with “test”, use mampac test*.dat.
mampac [-r] [-mem] [-nproc=#|all|most|half] {filename(s)}
|
Recursively search directories and run AMPAC jobs [default current directory only]. |
|
Turn on memory profiling (see below for details). |
|
Specifies the number of processors to use--equivalent to NPROC. This is ignored if an NPROC command appears on the job's keyword line. |
|
Run each of the specified AMPAC jobs in the
background. When |
The memory profiling option is a new feature that allows users to determine the amount of memory used by a particular job. As AMPAC is being used to study larger and larger systems, it is important for users to be able to know how much memory a job requires so as to estimate how big of a system can be run on their computer. In many cases, top or ps is sufficient for this purpose, however, memory profiling provides a much more detailed report and so can provide information that these other commands cannot.
Using -mem will cause the AMPAC job to be run in the background
while the script repeatedly polls the job using the ps -el command once
per second until it finishes. The results for filename.dat
will appear
in a seperate file named filename.mem
in plain text format. The maximum
memory used by the job is also reported to the screen.
There are a few things to keep in mind in using this feature. The memory usage should be considered approximate and may vary slightly each time the job is run because we are only looking at snapshots. The results should only be compared to jobs running under the same operating system, since operating systems vary in how they report memory. Jobs that complete in a few seconds or less will likely produce no results and report zero memory consumption, since it may finish before there is a chance to test for memory usage.
When mampac is launched a small amount of text is written to the screen:
Process: 8668
Running AMPAC 9.0 for *.dat in directory:
/home/semo/john/Ampac_Geometries/Small_Molecules
To check the status of your job, type “cat MAMPAC.LOG”.
The text file named MAMPAC.LOG
is written to the current working
directory:
Running AMPAC 9.0 for *.dat in directory: /home/semo/john/Ampac_Geometries/Small_Molecules Tue Jun 10 15:15:49 CDT 2008 Submitting AMPAC 9.0 Data file: Ammonia.dat Running in Temporary Directory: amp8675 AMPAC: /home/semo/Ampac/program/ampac.exe Tue Jun 10 15:15:50 CDT 2008 Submitting AMPAC 9.0 Data file: Ammonium.dat Running in Temporary Directory: amp8675 AMPAC: /home/semo/Ampac/program/ampac.exe Tue Jun 10 15:15:51 CDT 2008 ... additional output ... Tue Jun 10 15:16:43 CDT 2008 Submitting AMPAC 9.0 Data file: Water.dat Running in Temporary Directory: amp8675 AMPAC: /home/semo/Ampac/program/ampac.exe Tue Jun 10 15:16:44 CDT 2008 16.4u 19.8s 1:04 56% THAT'S ALL FOLKS...
Copyright © 1992-2013 Semichem, Inc. All rights reserved. |