Fix: sink layer id and tensor write-back scheduling#96
Open
sjoks93 wants to merge 2 commits intoKULeuven-MICAS:masterfrom
Open
Fix: sink layer id and tensor write-back scheduling#96sjoks93 wants to merge 2 commits intoKULeuven-MICAS:masterfrom
sjoks93 wants to merge 2 commits intoKULeuven-MICAS:masterfrom
Conversation
sjoks93
changed the title
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
I created this fix for sink layer ids. The logic is as follows:
Current implementation looks for layer ids where ALL nodes associated with id have no successors. This is wrong, as intra-layer partitioning leads to edges between sink nodes. Instead, if ONE node with a certain id has no successor, then ALL nodes with that id should be considered as sink nodes.
EDIT: I have added a second bug fix related to scheduling tensor write-back. Without this fix, a tensor write-back may overlap with another tensor transfer (from off-chip to a core). I wanted to do this in a more proper way with two separate pull request, but I have realized my mistake too late. Hence, I combined both fixes in this pull request. (Sorry)
Please review each commit separately and decide if any or both are acceptable