[autoparallel] fix C version rotor inconsistency #1691
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
Merge ColossalAI
Daily merge
| setattr(gm, "__sequence__", sequence) | ||
|
|
||
| # set __opttable__ attribute to GraphModule | ||
| setattr(gm, "__opttable__", opt_table[0]) |
There was a problem hiding this comment.
because I think don't need what table to be attached to the graph module, the returned opt_table by solver is a table containing opt_table[0] = opt, opt_table[1] = what, and I think we only need opt to check the value.
super-dainiu
approved these changes
Oct 12, 2022
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Outdated suggestions cannot be applied.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.
What's New?
Previously in PR #1679 , @super-dainiu reported the inconsistency between C version and python version of rotor solver, I dived into the C source code and found that the problem is that the C version compute the value in different order, with older version of profiler, the computation order will not damage the result, but currently some of the output of profiler has memory size of 0 (depends on newer version of graph analysis, some of the operation will discard the input), thus, the different order will cause solver failure. In this PR, I fix this problem and now the C version is consistent with python version.