Hi Kázmér,
If you got all the output files (in dti3.bedpostX) you should not care
about this message.
The monitor script had some problems but in the last version (5.0.7), we
have improved it.
I think 2 hours is too much, what kind of dataset are you processing
(dimensions-size)?
and which GPU model ?
Maybe bumblebee is slowing down the process.
Best,
Moises.
El viernes, 30 de mayo de 2014, Kazmer Kovacs <[log in to unmask]>
escribió:
> Dear FSL Experts!
>
> I've tried to set up a system on my laptop to run the bedpostx_gpu (Centos
> 6.5, Cuda 5.0). It has intel and nvidia graphic processors too. To use my
> nvidia card in the laptop I 've installed the bumblebee. After running the
> bedpostx_gpu (2 hours) I have got the following error. Should I care about
> this message?
> The intermediate files are still in the subjectdir. The probtracx runs (it
> seems it is OK).
>
> [kkazmer@localhost data]$ optirun --no-xorg bedpostx_gpu dti3/
> subjectdir is /home/kkazmer/Documents/data/dti3
> Making bedpostx directory structure
> Copying files to bedpost directory
> Queuing parallel processing stage
> 0 parts processed of 27
> .
> .
> .
> 26 parts processed of 27
> Queuing post processing stage
> 0 parts processed of 27
> 0 parts processed of 27
> /usr/local/fsl/bin/bedpostx_gpu: line 302: 9898 User defined signal 1
> ${subjdir}.bedpostX/monitor
>
> Thank You
> Kázmér
>
