-
Notifications
You must be signed in to change notification settings - Fork 1.4k
support GPU tensor for GridPatch and GridPatchDataset
#6246
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
support GPU tensor for GridPatch and GridPatchDataset
#6246
Conversation
* `GridPatch`, `GridPatchd`, `RandGridPatch` and `RandGridPatchd` * `GridPatchDataset`, `PatchIter`, `PatchIterd` and `iter_patch`
for more information, see https://pre-commit.ci
|
/black |
Signed-off-by: monai-bot <monai.miccai2019@gmail.com>
|
By checking the tests in CI, it seems |
thanks, let me skip the tests for this new feature for torch<1.11... |
Signed-off-by: Wenqi Li <wenqil@nvidia.com>
|
there are still some typing issues, I think you can debug with |
DCO Remediation Commit for Qingpeng Li <qingpeng9802@gmail.com> I, Qingpeng Li <qingpeng9802@gmail.com>, hereby add my Signed-off-by to this commit: 4d178e2 Signed-off-by: Qingpeng Li <qingpeng9802@gmail.com>
for more information, see https://pre-commit.ci
|
/black |
Signed-off-by: monai-bot <monai.miccai2019@gmail.com>
Signed-off-by: Qingpeng Li <qingpeng9802@gmail.com>
|
/build |
1 similar comment
|
/build |
|
/build |
wyli
left a comment
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Fixes #6245 .
Description
support GPU tensor for
GridPatch,GridPatchd,RandGridPatchandRandGridPatchdGridPatchDataset,PatchIter,PatchIterdanditer_patchChanges
GridPatch,GridPatchDatasetand their siblings need to copy GPU tensor to numpy array (or not support GPU tensor) in their pipeline. This PR enables them to support end-to-end GPU operations.pad_ndis used heavily, its robustness is improved. Some strange behavior introduced byPaduses numpy though pytorch backend is available #6066 is fixed.GridPatch,GridPatchwithpin_memory=Truesignificant slow-down in following epochs #6082 might be fixed by this PR?RandGridPatch's mode defaults toNone(consistent with docs)Need to Notice
PatchIter,PatchIterdanditer_patchare kept asNumpyPadMode.WRAP. However,GridPatchdefaults toNone. (might be inconsistent?)Types of changes
./runtests.sh -f -u --net --coverage../runtests.sh --quick --unittests --disttests.make htmlcommand in thedocs/folder.