Dear Nárlon
Thanks for the interest in our work!
You have several options:
1. If it’s only a couple of subjects, you could bite the bullet and run them
on the CPU. It’ll be slow, but hey, easy to run (just add the --cpu flag).
2. Changing the resolution at which the segmentations are genera
External Email - Use Caution
Hi Malte,
indeed the -g helped to execute on GPU, but it exhausted my resources.
Have only 23GB on GPU :(
Thanks and best wishes,
Martin
On Fri, Nov 8, 2024 at 1:34 PM Hoffmann, Malte,PhD
wrote:
>
> Hi Martin,
>
> In addition to the --gpu=all or 0
Hi Martin,
In addition to the --gpu=all or 0 flag for Docker, you'll need to include the
-g flag for SynthMorph to enable the GPU. You can choose the device by setting
CUDA_VISIBLE_DEVICES; SynthMorph will default to 0. Affine registration should
require at least 16 GB of device memory, and the
External Email - Use Caution
Hi,
Thanks a lot.
I am particularly excited to the Synth* tools. I am trying SynthMorph from
docker. I see that it does not use GPU (nvtop). I tried adding —gpus=0 but
still see that inference runs on CPU (htop).
Are there any experiences in register
External Email - Use Caution
Hello FreeSurfer Developers,
Firstly, Thanks for your great work.
Two different releases have the same MD5 hash
"1f107c22eb75821e1aa46b55a6eb8189", namely
freesurfer_ubuntu20-7.4.0_amd64.deb (Correct one)
freesurfer_ubuntu20-7.3.2_amd64.deb (MD5 b2d0
