[MITgcm-support] I/O issue with CTRL_SET_PREC_32
Maëlle Nodet
maellenodet at mac.com
Tue Oct 30 16:45:28 EDT 2007
Patrick,
I rerun the unchanged verification/natl_box_adjoint (without control
precision set to 32): works fine.
Then I turned the control precision to 32: same problem as before
"attempt to access non-existent record..."
This was on dolphin (linux_amd64_ifort) and also on Tom's pc (pentium
proc with g77).
I think the MITgcm version we use might be 3 weeks old.
Thanks for your help,
Cheers,
Maelle
Le 29 oct. 07 à 16:56pm, Patrick Heimbach a écrit :
>
> Hi Maelle,
>
> this should not cause any problems.
> We are using this both with older and with the essentially most up-
> to-date version.
>
> Maybe one guess:
> Make sure you remove all fields xx_* etc. when
> doing a run with modified setup.
>
> Second, since your setup seems modified,
> rerun the un-changed natl_box_adjoint/ verification
> to convince yourself that it works,
> then check which step in your modifs cause the problem.
>
> Cheers
> -Patrick
>
> PS:
> Good to hear you're back on the case.
>
>
>
> On Oct 29, 2007, at 2:18 PM, Maëlle Nodet wrote:
>
>> Dear MITgcm developpers,
>>
>> I have an I/O issue with the code.
>>
>> I use it with ECCO options in order to do optimization. The
>> configuration used here is very close to the one in /verification/
>> natl_box_adjoint.
>>
>> When I set the control precision to 64, the code runs well, and
>> when I set it to 32 (#define CTRL_SET_PREC_32 in
>> ECCO_CPPOPTIONS.h) I get "forrtl: severe (36): attempt to access
>> non-existent record, file xx_theta.0000000000.001.001.data".
>>
>> Would you know how to fix this?
>>
>> Thanks,
>>
>> Maelle
>>
>>
>> _______________________________________________
>> MITgcm-support mailing list
>> MITgcm-support at mitgcm.org
>> http://mitgcm.org/mailman/listinfo/mitgcm-support
>
> ---
> Patrick Heimbach | heimbach at mit.edu | http://www.mit.edu/~heimbach
> MIT | EAPS 54-1518 | 77 Massachusetts Ave | Cambridge MA 02139 USA
> FON +1-617-253-5259 | FAX +1-617-253-4464 | SKYPE patrick.heimbach
>
>
More information about the MITgcm-support
mailing list