Dear Jesper,
Thank you for your answer! As I do not have a CUDA-gpu, I
used eddy_openmp. However I get the following error
message.
How would you proceed?
Thanks a lot!
Best,
Theodor.
Starting eddy...
Segmentation violation, Address not mapped, Offending
address = 0x402ee54d0
eddy_openmp ) [0x4e3db3] [_�
eddy_openmp ) [0x5255ce] [�Q@
eddy_openmp ) [0x525d20] [�Q@
eddy_openmp ) [0x451022] [�Q@
eddy_openmp ) [0x45279e] [�Q@
eddy_openmp ) [0x4479ac] [�Q@
eddy_openmp ) [0x458db6] [�Q@
./eddy_current_correction-1: Zeile 6 [i.e. line 6 (of the
batch)]: 23639 Speicherzugriffsfehler [i.e. "segmentation
fault"](Speicherabzug geschrieben) [i.e. memory snapshot
written] eddy_openmp --imain=${sbj}_dir81_merged_LR_RL_gdc
--mask=${sbj}_b0_topup_brain_mask --acqp=acqparms.txt
--index=eddy_param.txt
--bvecs=${sbj}_dir81_merged_LR_RL.bvec
--bvals=${sbj}_dir81_merged_LR_RL.bval
--topup=${sbj}_b0_topup_results
--out=${sbj}_eddy_corr_data
Dear Theodore,
>
> Thank you for your answer:
>
> "ls ${FSLDIR}/bin/eddy*" gives you
>
> "eddy_combine eddy_correct eddy_cuda eddy_openmp”
the executables you want to use are eddy_cuda or
eddy_openmp depending on if you have a CUDA-gpu box on
your system or not. If you do, then I strongly recommend
using eddy_cuda. If not you should use eddy_openmp.
The latter is multi-threaded and will potentially take as
many threads as it can get its hands on, which may not be
what you want. In that case you can use the environment
variable OMP_NUM_THREADS and set it to a suitable value.
If you prefer to have an executable called eddy I suggest
you make a link from ${FSLDIR}/bin/eddy to
${FSLDIR}/bin/eddy_* where * is the version you decide to
use.
Jesper
|