open TELEMAC-MASCARET The mathematically superior suite of solvers

Apologies Jean-Michel. They are the TPXO harmonic databases, they will be too large to post on here but if you like I can send you a dropbox link to you?

I have done some further investigation and have found the following things:

1: I can now run it in scalar after increasing the number of boundary points keyword

2: it works well for short duration's in scalar but not long runs with result files in excess of 4GB, the result file is corrupted. Similarly to this post:

www.opentelemac.org/index.php/kunena/21-...s-corrupted?start=10

3: The model is running in parallel and if you stop it part way through, the partitioned results file is correct with a correct time stamp. However if allowed to finish, the recombined file has the error.

I need to find the command that allows me to run the model without merging the results file at the end and do it myself afterwards to investigate further.

Regards

David

Hi David

About the last point, you could try to run telemac in 3 steps:
Split the model in the right number of subdomain
Run it
Merge the result

There is option for these tasks:
--split --run --merge
have a look in the help of scripts (--help)

Regards

I forgot to mention the option to fix the name of the directory where the run occurs -wdir

Regards

Thank you Christophe

I'm still having trouble with a time stamp error. Whenever any model is run in parallel (using v7p1r0) I get a time stamp of -001/-1/-1 -1:-1:-1.000.

I have the separate results files before they are merged and the time stamps are correct. When they are merged, either automatically or separately afterwards the time stamp changes.

Is the bug in gretel that is referred to in: www.opentelemac.org/index.php/kunena/16-...llel-results-command

Are there any solutions?

Regards

David