Make DistributedDataParallel usable with CPU models #20236
Closed
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Differential Revision: D15245428 Differential Version: 81320054
pytorchbot
added
oncall: distributed
This was referenced May 7, 2019
mrshenli
reviewed
May 7, 2019
torch/nn/parallel/distributed.py
Outdated
| self.broadcast_bucket_size) | ||
| # Synchronize buffers across processes. | ||
| # The process with rank 0 is considered the authoritative copy. | ||
| self._broadcast_coalesced(self.modules_buffers[0], |
Contributor
There was a problem hiding this comment.
This is where it might be confusing. We have two broadcast_coalesced, but one is distributed and another is local. Is this intentional?
Contributor
Author
There was a problem hiding this comment.
They come from different modules. For clarity we can proxy them with a function on the class, one for local broadcast and the other for distributed broadcast.
Differential Revision: D15245428 Differential Version: 81326665
mrshenli
approved these changes
May 8, 2019
Differential Revision: D15245428 Differential Version: 81461820
Contributor
|
@pytorchbot retest this please |
Contributor
|
This pull request has been merged in 558c6c4. |
pytorchmergebot
pushed a commit
that referenced
this pull request
Oct 27, 2022
Fixes: ``` /dev/shm/rbarnes/tempfs/pytorch/aten/src/ATen/native/cuda/UnaryFractionKernels.cu(125): warning #20236-D: pragma "diag_suppress" is deprecated, use "nv_diag_suppress" instead /dev/shm/rbarnes/tempfs/pytorch/aten/src/ATen/native/cuda/UnaryFractionKernels.cu(125): warning #20236-D: pragma "diag_suppress" is deprecated, use "nv_diag_suppress" instead /dev/shm/rbarnes/tempfs/pytorch/aten/src/ATen/native/sparse/cuda/SparseMatMul.cu(73): warning #20236-D: pragma "diag_suppress" is deprecated, use "nv_diag_suppress" instead /dev/shm/rbarnes/tempfs/pytorch/aten/src/ATen/native/sparse/cuda/SparseMatMul.cu(73): warning #20236-D: pragma "diag_suppress" is deprecated, use "nv_diag_suppress" instead ``` cc @ngimel Pull Request resolved: #87712 Approved by: https://github.com/soumith
kulinseth
pushed a commit
to kulinseth/pytorch
that referenced
this pull request
Nov 5, 2022
Fixes: ``` /dev/shm/rbarnes/tempfs/pytorch/aten/src/ATen/native/cuda/UnaryFractionKernels.cu(125): warning pytorch#20236-D: pragma "diag_suppress" is deprecated, use "nv_diag_suppress" instead /dev/shm/rbarnes/tempfs/pytorch/aten/src/ATen/native/cuda/UnaryFractionKernels.cu(125): warning pytorch#20236-D: pragma "diag_suppress" is deprecated, use "nv_diag_suppress" instead /dev/shm/rbarnes/tempfs/pytorch/aten/src/ATen/native/sparse/cuda/SparseMatMul.cu(73): warning pytorch#20236-D: pragma "diag_suppress" is deprecated, use "nv_diag_suppress" instead /dev/shm/rbarnes/tempfs/pytorch/aten/src/ATen/native/sparse/cuda/SparseMatMul.cu(73): warning pytorch#20236-D: pragma "diag_suppress" is deprecated, use "nv_diag_suppress" instead ``` cc @ngimel Pull Request resolved: pytorch#87712 Approved by: https://github.com/soumith
kulinseth
pushed a commit
to kulinseth/pytorch
that referenced
this pull request
Dec 10, 2022
Fixes: ``` /dev/shm/rbarnes/tempfs/pytorch/aten/src/ATen/native/cuda/UnaryFractionKernels.cu(125): warning pytorch#20236-D: pragma "diag_suppress" is deprecated, use "nv_diag_suppress" instead /dev/shm/rbarnes/tempfs/pytorch/aten/src/ATen/native/cuda/UnaryFractionKernels.cu(125): warning pytorch#20236-D: pragma "diag_suppress" is deprecated, use "nv_diag_suppress" instead /dev/shm/rbarnes/tempfs/pytorch/aten/src/ATen/native/sparse/cuda/SparseMatMul.cu(73): warning pytorch#20236-D: pragma "diag_suppress" is deprecated, use "nv_diag_suppress" instead /dev/shm/rbarnes/tempfs/pytorch/aten/src/ATen/native/sparse/cuda/SparseMatMul.cu(73): warning pytorch#20236-D: pragma "diag_suppress" is deprecated, use "nv_diag_suppress" instead ``` cc @ngimel Pull Request resolved: pytorch#87712 Approved by: https://github.com/soumith
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Stack:
:black_circle: #20236 Make DistributedDataParallel usable with CPU models 💚
:white_circle: #20235 Refactor core DistributedDataParallel tests 💚
:white_circle: #20234 Add c10d::broadcast_coalesced and tests 💚
Use the new version of broadcast_coalesced that deals with both CPU
and CUDA models. Add tests that evaluate correctness of
DistributedDataParallel for CPU models.
Closes #17757.
Differential Revision: D15245428