We run our Freesurfer (recon-all) jobs on a grid (Sun Grid Engine) which uses a single NFS file server for all data. Since we have begun using Freesurfer 5.0.0 we have noticed very intense I/O activity as soon as recon-all jobs begin. The level of activity is markedly greater than we have seen in previous versions. The effective cumulative impact on our file server, when scores of jobs are launched at once, is significant and degrades the server's response.
Has something changed about how data files are accessed at the beginning of a recon-all job? If so, is there a way to "ease into it" so that the initial activity is more distributed as processing continues?
< david
David Mischel
Manager of IT
Center for Imaging of Neurodegenerative Diseases (CIND)
http://www.cind.research.va.gov/ http://www.cind.research.va.gov/
VA Medical Center
4150 Clement Street, 114M
San Francisco, CA 94121
voice: 415-221-4810 x3864
fax: 415-668-2864
freesurfer@nmr.mgh.harvard.edu