You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This is a known issue. When an application is started with virtuaGL then torch fails to locate the cudnn library and fails. This impacts all torch based modules in Slicer. See the posts on VGL repo:
Note that this is independent on the of the NVIDIA driver related torch problem reported here #66. I have replicated the current issue with 550 series non grid driver on a vanilla exosphere g3.xl instance.
This is a known issue. When an application is started with virtuaGL then torch fails to locate the cudnn library and fails. This impacts all torch based modules in Slicer. See the posts on VGL repo:
VirtualGL/virtualgl#227
VirtualGL/virtualgl#209
Note that this is independent on the of the NVIDIA driver related torch problem reported here #66. I have replicated the current issue with 550 series non grid driver on a vanilla exosphere g3.xl instance.
Since the Slicer shortcut in MorphoCloud is indeed configured to use vgl, it means applications that rely on torch fails. To avoid confusion about when to invoke Slicer with or without vgl, I propose we stop using vgl and try the native accelerated X recipe provided by the JetStream documentation
The text was updated successfully, but these errors were encountered: