Thanks. I installed the patch, but see another error that I suspected might happen:split_parts_gpu: /lib64/libc.so.6: version `GLIBC_2.14' not found (required by split_parts_gpu)As far as I know, there isn't a "safe" way to upgrade glibc since many other programs could break. Do I have to do a manual install (into e.g. /usr/local) and then point LD_LIBRARY_PATH there?ChrisHi Chris,There is a bug in the 6.0.0 version of bedpostx_gpu ( actually in xfibres_gpu ). The upcoming patch ( FSL6.0.1) will have a fixed binary, but until then Moises has generated a patch ( see below for details ):
Hi Mina,A version for FSL 6.x is now available.You can find the files here (FSL 6.x FILES):
Same for probtrackx_gpu:
This should solve that problem, but let me know if it doesn't.Thanks,Moises.Hope this helps,
Kindest RegardsMatthew
--------------------------------
Dr Matthew Webster
FMRIB Centre
John Radcliffe Hospital
University of Oxford
On 28 Dec 2018, at 04:38, Chris Watson <[log in to unmask]> wrote:
Hi Moises et al.,
I am trying to run "bedpostx_gpu", yet it finishes pretty quickly with an error. I have successfully run "eddy_cuda" on the same system.
The error message (in dwi.bedpostX/logs/bedpostx_gpu.eXXXX.1) is some variant of:
"/bin/sh: line 1: 4320 Segmentation fault (core dumped)"
and then the command for "xfibres_gpu" follows.
In "/var/log/messages", I see e.g. (note the "..." are added by me)
kernel: xfibres_gpu[4320]: segfault at 0 ip 00000035a2c897f2 sp 00007ffd0657d358 error 4 in libc-2.12.so[35a2c00000+18b000]
abrt[4328]: Saved core dump of pid 4320 to /var/spool/abrt/ccpp-2018...
abrtd: Directory 'ccpp-2018...' creation detected
abrtd: Executable '/parietal/fsl/fsl-6.0.0/bin/xfibres_gpu' doesn't belong to any package and ProcessUnpackaged is set to 'no'
abrtd: 'post-create' on '/var/spool/abrt/ccpp-2018...' exited with 1
abrtd: Deleting problem directory '/var/spool/abrt/ccpp-2018...'
Here is some system info:
OS: CentOS 6.10 (64-bit)
Kernel: 2.6.32-754.6.3
Graphics card: NVIDIA GF100GL (Quadro 4000)
Kernel driver: kmod-nvidia-390.87-1.el6_10.elrepo.x86_64
CUDA version: 7.5-18.x86_64
FSL v6.0.0
I ran "bedpostx_gpu" with the verbose flag, but no additional info was provided.
"ldd $FSLDIR/bin/xfibres_gpu" does not show any errors. Same with "libbedpostx_cuda.so", "merge_parts_gpu", "split_parts_gpu".
Could this be a problem related to "glibc"? It is v2.12-1.212.el6, which is the current version for CentOS 6.
I am happy to give more information if necessary. Thanks,
Chris
To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1
To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1
To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1