open TELEMAC-MASCARET The mathematically superior suite of solvers

hi all
When I use telemac with the two keywords “computation continued” and “previous computation file” the simulation stops after 30 days. Also it happens if I use “initial time reset to zero” keyword with them.
But the simulation continued normally if I did not use the above keywords and has the normal end.
I run the simulation in parallel mode using 8 processes.
thanks

Hello,

We would need some extra information. When you say "the computation stops after 30 days", is it a crash or a stop with some message ? Do you use files with a specific implementaion to read them ?

JMH

thanks
since the previous result file size is 43 Gb, i used the matlab telemac tools to extract only the last time step and save the serafin file again. i apply also the wind and presure data that change in space and time using meteo.f
the error message is access violation which not appear if i did not use the computation continued keywords
i will try to send the error message to you but it will take some time since i will run the simulation again now

Hello,

OK. A look at your file meteo.f would be also useful.

JMH

here is the princi.f file that has all of my new subroutines i used

the file with txt format
the file has several subroutines

Hello,

OK thanks. After reading meteo.f: I may be wrong, but I do not understand how you manage parallelism in this file. In commands like :

DO i=1,NPOIN
PATMOS (i)= pres((round*41891)+(i))
ENDDO

you seem to read pressure in a file on channel 98, that would be in global numbering, while you set it in the loop in local numbering, am I wrong ?

Anyway as you deal with months in this subroutine, and because your computation stops after 30 days, I would strongly suspect a problem there.

With best regards,

Jean-Michel Hervouet

first thanks for your interactive reply
the simulation run correctly without any probleme if i did not use continuation. i also checked it is transfer from one month to another correctly inside meteo.f
did you mean for parallisim i must use for example the following
DO i=1,NPOIN
PATMOS%R(i)= pres((round*41891)+(i))
END DO
please correct me if the above modification to meteo.f is not what you mean, and did you think it has relation to the continuation of the process
in any way i will tray to use continuation without parallism and see

best regards

Hello,

No, I would just like to know what is in file 98. If it is a file with all atmospheric pressures at every point of your original file, then it is in global numbering, where NPOIN is the total number of points. Then when you do the loop to fill PATMOS(I), NPOIN is the local number of points, so there would be a mismatch between numbering. However this would explain differences between scalar runs and parallel runs, not a problem of computation continued. Your subroutine Meteo is rather complicated and in case of continuation you will not go at the same address in the data files, this is why I suspect an error here. If it is possible, comparisons between scalar and parallel would be useful.

Jean-Michel Hervouet