You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@mxnet.apache.org by GitBox <gi...@apache.org> on 2021/07/13 23:58:42 UTC
[GitHub] [incubator-mxnet] ptrendx opened a new pull request #20445: [WIP][FEATURE][CONTRIB] Spatial parallel convolution support
ptrendx opened a new pull request #20445:
URL: https://github.com/apache/incubator-mxnet/pull/20445
## Description ##
This PR introduces Gluon blocks enabling spatial parallel convolutions (convolutions spanning multiple GPUs). This functionality was used in the NVIDIA's UNet-3D submission for MLPerf 1.0.
## Checklist ##
### Essentials ###
- [x] PR's title starts with a category (e.g. [BUGFIX], [MODEL], [TUTORIAL], [FEATURE], [DOC], etc)
- [ ] Changes are complete (i.e. I finished coding on this PR)
- [ ] All changes have test coverage
- [ ] Code is well-documented
### Changes ###
- [x] SpatialParallelConv2D module
- [x] SpatialParallelConv3D module
- [x] SpatialParallelSplit module
- [x] SpatialParallelAllgather module
## Comments ##
- Currently the code has multiple restrictions: supports only the channels last format (NHWC and NDHWC) and requires that the filter/padding in first spatial dimension (D in 3D case, H in 2D case) fulfill `pad = (filter - 1)/2` equation (so e.g. for filter size 3, pad has to be 1).
- In order to bootstrap NCCL communicator I use out of band MPI communication via mpi4py, which is why I submit it to contrib module
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: commits-unsubscribe@mxnet.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [incubator-mxnet] mxnet-bot commented on pull request #20445: [WIP][FEATURE][CONTRIB] Spatial parallel convolution support
Posted by GitBox <gi...@apache.org>.
mxnet-bot commented on pull request #20445:
URL: https://github.com/apache/incubator-mxnet/pull/20445#issuecomment-879480107
Hey @ptrendx , Thanks for submitting the PR
All tests are already queued to run once. If tests fail, you can trigger one or more tests again with the following commands:
- To trigger all jobs: @mxnet-bot run ci [all]
- To trigger specific jobs: @mxnet-bot run ci [job1, job2]
***
**CI supported jobs**: [clang, sanity, windows-gpu, centos-gpu, unix-cpu, centos-cpu, unix-gpu, website, windows-cpu, edge, miscellaneous]
***
_Note_:
Only following 3 categories can trigger CI :PR Author, MXNet Committer, Jenkins Admin.
All CI tests must pass before the PR can be merged.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: commits-unsubscribe@mxnet.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [incubator-mxnet] ptrendx commented on pull request #20445: [WIP][FEATURE][CONTRIB] Spatial parallel convolution support
Posted by GitBox <gi...@apache.org>.
ptrendx commented on pull request #20445:
URL: https://github.com/apache/incubator-mxnet/pull/20445#issuecomment-879480271
Marking this WIP for now, since there are no tests nor documentation in this PR yet.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: commits-unsubscribe@mxnet.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org