[BWARE] Track more compressed-friendly ops in FederatedWorkloadAnalyzer#2481
Open
Baunsgaard wants to merge 2 commits into
Open
[BWARE] Track more compressed-friendly ops in FederatedWorkloadAnalyzer#2481Baunsgaard wants to merge 2 commits into
Baunsgaard wants to merge 2 commits into
Conversation
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## main #2481 +/- ##
============================================
+ Coverage 71.37% 71.42% +0.04%
- Complexity 48749 48830 +81
============================================
Files 1571 1572 +1
Lines 188912 189109 +197
Branches 37067 37106 +39
============================================
+ Hits 134845 135064 +219
+ Misses 43601 43579 -22
Partials 10466 10466 ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
Extends the federated workload counter so that compression decisions account for additional instruction shapes beyond AggregateBinary. - Pass the right-hand column count to incOverlappingDecompressions so the cost model reflects the actual decompression size rather than counting a single column - Count MMChainCPInstruction as one LMM and one RMM contribution per invocation - Count AggregateUnaryCPInstruction: when reducing columns with a sum/mean operator, treat it as a dict-op (compression-friendly); otherwise count it as a decompression - Minor formatting cleanup in compressRun
Cover the instruction-shape branches in incrementWorkload that drive federated compression decisions, which previously had no direct tests: - AggregateBinary: RMM/LMM counting, overlapping-decompress sizing by the right-hand column count, and the validSize row/column guards - MMChain: one LMM and one RMM contribution per invocation - AggregateUnary: dict-op vs decompression classification across ReduceAll/ReduceRow/ReduceCol with sum, mean, product, and max operators - Instance-level dispatch and compressRun threshold behavior, asserting async compression materializes when the cost model would compress
Baunsgaard
force-pushed
the
split/federatedAnalyzer
branch
from
eb81e3b to
89f7059
Compare
Contributor
Author
|
Hey @christinadionysio , I have seen this error before for the CI of Scuro, where the machine runs out of disk space, maybe you want to take a look at the logs? |
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
1 participant
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.
Extends the federated workload counter so that compression decisions account for additional instruction shapes beyond AggregateBinary.