"JOB TIME SUMMARY" in .dat file

[VetleSPZ]

Hi! I have tried to figure out how to display the total CPU time of different analysis in a parametric study of different element types and mesh sizes. At the end of my .dat file, the following information is printed

P R O B L E M S I Z E

NUMBER OF ELEMENTS IS 5701
NUMBER OF NODES IS 5777
NUMBER OF NODES DEFINED BY THE USER 5777
TOTAL NUMBER OF VARIABLES IN THE MODEL 34662
(DEGREES OF FREEDOM PLUS MAX NO. OF ANY LAGRANGE MULTIPLIER
VARIABLES. INCLUDE *PRINT,SOLVE=YES TO GET THE ACTUAL NUMBER.)

END OF USER INPUT PROCESSING

JOB TIME SUMMARY
USER TIME (SEC) = 0.32000
SYSTEM TIME (SEC) = 2.00000E-02
TOTAL CPU TIME (SEC) = 0.34000
WALLCLOCK TIME (SEC) = 1


I suspect that this "JOB TIME SUMMARY" is not linked to the analyses itself, but to the input processing. Mainly because it is printed to the .dat file before the simulation is finished, and the fact that total CPU time is quite low. I am running a explicit tube crash simulation.

Is my suspicion right, and is there any way to get a time summary of the whole analysis?

I am currently using ABAQUS 2016.

 

[Mustaine3]

After an analysis you'll find this output also at the end of the .msg file when running A/Standard.
For A/Explicit you should look at the .sta file.

 

[Dave442]

The dates/times that an analysis was submitted and completed are also printed in the .log file. Might be useful.

 

[VetleSPZ]

Thanks for the replies!

I have looked into the CPU time printed in the increments information in the .sta file. If I am not wrong, this time is just the same as how long the simulation have been running (end time minus start time). So running the simulation with different number of CPUs would result in different "run time", right?

Also, I discovered that this "run time" is differing somewhat (15-20%) for the same input file with the same number of CPUs. This might have to due with the fact that I run these simulations on a cluster.

Despite this, I guess this method is sufficient for my usage. But if someone knows a better method to evaluate the computational cost of different simulations, I would be pleased to know. My understanding after searching the web is that this evaluation is not as straight forward as I first expected.

 

[VetleSPZ]

Update to my main question:

I just looked at the .dat file in an old implicit simulation, where the following information was printed

THE ANALYSIS HAS BEEN COMPLETED

ANALYSIS COMPLETE
WITH 5 WARNING MESSAGES ON THE DAT FILE
AND 39 WARNING MESSAGES ON THE MSG FILE
28 WARNINGS ARE FOR NEGATIVE EIGENVALUES

JOB TIME SUMMARY
USER TIME (SEC) = 14.400
SYSTEM TIME (SEC) = 1.3000
TOTAL CPU TIME (SEC) = 15.700
WALLCLOCK TIME (SEC) = 16

Which was the information i wanted to find in my explicit simulation .dat file. I guess ABAQUS/Explicit does not print this information after all...

My questions from my previous post still stands

 

[IceBreakerSours]

Wall-clock time is the standard term used to convey the time it took for the analysis to finish.
If there is confusion, you could write a simple job submission script with some sort of "stop-watch" function keeping track of time.

*********************************************************
Are you new to this forum? If so, please read these FAQs:

http://www.eng-tips.com/faqs.cfm?fid=376

http://www.eng-tips.com/faqs.cfm?fid=1083