Hi all,

When I run recon-all on bert using CUDA, the process completes with "finished without error" message, but in the log there is a cuda-related error in addition to the errors mentioned in my other email (Subject "bert recon-all errors, non-cuda").

I'd appreciate any comments on whether this is significant, how to prevent it, and so on.

Thanks,

Graham

========================
[3982] CUDA Error in file 'mriconvolve_cuda.cu' on line 945 : unload of CUDA runtime failed.
Abort
ERROR: mri_ca_register with non-zero status 134
but continuing despite the error
=============================================

#------------------------------------------

Started at Fri May 31 13:35:28 PDT 2013
Ended   at Fri May 31 16:38:50 PDT 2013
#@#%# recon-all-run-time-hours 3.056
recon-all -s bert4 finished without error at Fri May 31 16:38:50 PDT 2013