2010/10/25 Nick Schmansky <nicks@nmr.mgh.harvard.edu>

Freesurfanados and GPU fans,

We (namely Richard) have fixed some of the problems with the existing
freesurfer gpu binaries, which were the result of building against cuda
3.0 on our build platforms, which seemed to be incompatible with 3.1.
We've built against 3.1, although now i see nvidia has come out with
3.2, but i've tested against that and they seem to work.

So you can get the binary tarball here:

ftp://surfer.nmr.mgh.harvard.edu/pub/dist/freesurfer/misc/linux-centos4_x86_64/fscudabins-linux-centos4_x86_64.tgz

and here for the 32b linux:

ftp://surfer.nmr.mgh.harvard.edu/pub/dist/freesurfer/misc/linux-centos4_x86_64/fscudabins-linux-centos4.tgz

copy the extracted contents of those tarball into your freesurfer/bin
dir.

The binaries are:

mri_ca_register_cuda
mri_em_register_cuda
mris_fix_topology_cuda
mris_flatten_cuda
mris_inflate_cuda
mris_sphere_cuda
mri_vol2vol_cuda

No doubt people will encounter problems, so let us know. But hopefully
you will see a dramatic reduction in the runtime of the subcortical
segmentation stage (mri_ca_register).

Also, to remind people, Richard is targeting our gpu development for the
Fermi class of gpu card, which has better memory management which leads
to better performance of some algorithms, so if you're buying new
hardware, be sure to spec that class of card (see nvidia site).

Nick

_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer

The information in this e-mail is intended only for the person to whom it is
addressed. If you believe this e-mail was sent to you in error and the e-mail
contains patient information, please contact the Partners Compliance HelpLine at
http://www.partners.org/complianceline . If the e-mail was sent to you in error
but does not contain patient information, please contact the sender and properly
dispose of the e-mail.