This repository has been archived by the owner on Nov 17, 2023. It is now read-only.
Multi_sum_sq review, AtomicAdd removal #17002
Merged
Conversation
This file contains 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
Thanks ! I think I corrected all of them |
eric-haibin-lin
approved these changes
Dec 10, 2019
ptrendx
pushed a commit
to ptrendx/mxnet
that referenced
this pull request
Dec 20, 2019
* Update multi_sum_sq to avoid AtomicAdd * Add specific test for multi_sum_sq * Add a determism test and lint issues * better test for cheching op is deterministic * Follow MXNet letters case format * Reduce dimensions of tensors in the test
ptrendx
added a commit
that referenced
this pull request
Dec 20, 2019
* Improve the speed of the pointwise fusion graph pass (#17114) * Debug the long startup time * Optimize backward fusion * Figure out why the fusion pass is called twice * Cleaning * Small optimization * [BUGFIX] Fix trainer param order (#17068) * fix trainer param order * Update trainer.py * Update trainer.py * Update trainer.py * [reproducibility] multi_sum_sq review, AtomicAdd removal (#17002) * Update multi_sum_sq to avoid AtomicAdd * Add specific test for multi_sum_sq * Add a determism test and lint issues * better test for cheching op is deterministic * Follow MXNet letters case format * Reduce dimensions of tensors in the test Co-authored-by: Haibin Lin <linhaibin.eric@gmail.com> Co-authored-by: MoisesHer <50716238+MoisesHer@users.noreply.github.com>
shuo-ouyang
reviewed
Aug 11, 2021
|
|
||
| if (threadIdx.x == 0) { | ||
| block_reductions[(start_tensor_id + tensor_loc) * param.max_chunks_per_tensor + | ||
| param.block_to_chunk[blockIdx.x]] = final; |
There was a problem hiding this comment.
Maybe we should change the variable name here? = final specifies that a virtual function cannot be overridden in a derived class.
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
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.
Description
Modified multi_sum_sq operator to avoid nondeterministic behavior, which was potentially caused by AtomicAdd operation on GPU kernel.
Checklist
Essentials
Please feel free to remove inapplicable items for your PR.
Changes