open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi,

I am running T2D coupled with SISYPH, in parallel, on v5p9.
I get errors reading FORMATTED DATA FILE 1.
My bord.f looks like this:

...
USE DECLARATIONS_TELEMAC2D, ONLY : T2D_FILES, T2DFO1
INTEGER INP
...
INP = T2D_FILES(T2DFO1)%LU
READ(INP,*) ...

I get this error:
PGFIO-F-214/list-directed read/unit=0/READ not allowed for write-only file.
File name = stderr formatted, sequential access record = 0
In source file t2dfort.f, at line number 200

Is there anything I've missed?
Thanks very much,
Peter

Hello,

Never seen this kind of error, what you do seems correct. You can try :

PRINT*,'ACTION=',T2D_FILES(T2DF01)%ACTION to see if it is READ, WRITE, or READWRITE, and T2DF01 is equal to 26. INP is equal to 26 without coupling and is different when coupling. It is not sure that this problem happens on this file, it seems to be unit 0 ?

But why 5.9 ?

Regards,

Jean-Michel Hervouet

Thanks for the very quick reply

Your suggestion gives:

ACTION =
INP = 0
T2DFO1 = 0
PGFIO/stdio: Illegal seek
PGFIO-F-/REWIND/unit=0/error code returned by host stdio - 29.
File name = stderr formatted, sequential access record = 0
In source file t2dfort.f, at line number 201

Stragely, there is nothing printed for T2D_FILES(T2DFO1)%ACTION
I tried INP=T2D_FILES(26)%LU (can you do this?), and got LU=0 as well

Anything else you can think of?
Thanks
Peter

...(sorry)I meant INP=0 (not LU=0)

So there is definitely a memory error somewhere, first to look for in your specific Fortran. I is unlikely that it is in version 5.9 because parallelism and coupling with Sisyphe have been tested. Anyway you could try to move to version 6.1, and also try Telemac-2D in parallel without Sisyphe, then T2DF01 is equal to 26.

JMH

OK, using T2D only (in parallel):
code works using unit=26, but not using T2D_FILES(T2DFO1)%LU

and using T2D+SISYPHE (in parallel):
code does not works using unit=26, or T2D_FILES(T2DFO1)%LU

So is the problem with the command 'T2D_FILES(T2DFO1)%LU' ?

I can't get 6.1 working at the moment

Thanks,
Peter

No, the fact that T2DF01 equals 0 shows tha there is a memory error. T2DF01 is computed by a procedure that distributes channels between Telemac-2D ans Sisyphe, but it starts by 1 and and goes up, so 0 means that the memory has been erased.

JMH

... I've also noticed the FORMATTED DATA FILE 1 is copied to each processor temp file in the following way. Should I use T2DFO100007 instead?


- FORMATTED DATA FILE 1 : TIDEBC.DAT
COPY TIDEBC.DAT /homedir/.../T2D/cas.txt15667_tmp/T2DFO100007-00000
COPY TIDEBC.DAT /homedir/.../T2D/cas.txt15667_tmp/T2DFO100007-00001
COPY TIDEBC.DAT /homedir/.../T2D/cas.txt15667_tmp/T2DFO100007-00002
COPY TIDEBC.DAT /homedir/.../T2D/cas.txt15667_tmp/T2DFO100007-00003
COPY TIDEBC.DAT /homedir/.../T2D/cas.txt15667_tmp/T2DFO100007-00004
COPY TIDEBC.DAT /homedir/.../T2D/cas.txt15667_tmp/T2DFO100007-00005
COPY TIDEBC.DAT /homedir/.../T2D/cas.txt15667_tmp/T2DFO100007-00006
COPY TIDEBC.DAT /homedir/.../T2D/cas.txt15667_tmp/T2DFO100007-00007

Yes, this is probably not necessary, and it is due to the last value of field SUBMIT for this file in the dictionary, which is PARAL, SCAL instead would avoid this, but actually we do not know what the user does with this file and it could depend on the sub-domain.

JMH

OK, all I need to do is open the FORMATTED DATA FILE (in bord.f) and read the data, then close it.

How do I do this using SCAL?

Can I just OPEN the file (I think not, because each parallel partition tries to open it together, then I get 'END OF FILE' type error messages?)

Thanks,
Peter