Recover dense bottom BGA fanout and improve benchmark to 20/36 - #180
Draft
seveibar wants to merge 25 commits into
Draft
Recover dense bottom BGA fanout and improve benchmark to 20/36#180seveibar wants to merge 25 commits into
seveibar wants to merge 25 commits into
Conversation
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
seveibar
changed the title
seveibar
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Dense bottom-facing BGA cases can now reserve every source escape, repair local source and plane conflicts, and route buses through their declared crossover layers. Bounded source shortening and length matching complete both the RK3308 bottom-center and bottom-left cases, including all 162 connections and the original timing constraints.
The default
./benchmark.shrun improves from 18/36 to 20/36 on the latest dataset already included in main:The full run uses the unchanged four-worker configuration and 120-second deadline per sample. Bottom-center passes in 87.95 seconds; bottom-left passes in 80.09 seconds. All prior passes remain, with unchanged SVGs. The remaining 16 cases time out; there are no errors or partial results.
benchmark-results/benchmark.jsonandbenchmark.mdrecord all 36 cases.benchmark-results/*.svgcontains exactly one committed snapshot for each of the 20 passing circuits. The new bottom-left snapshot was visually reviewed and matches the separately validated production run exactly.Candidates are derived from the original prepared geometry. The solver retains the original endpoints, layers, clearance, bus ordering, skew, and via rules; complete solutions still pass source/corner checks, the original solution validator, and emitted-copper DRC. Source repair now skips unused native-site searches. After deferred timing and source shortening, bounded periodic tuning avoids repeating a previously unsuccessful ordinary search.
Validation: complete default 36-case benchmark at
c158b62; focused visual source, crossover, and timing regressions; repository TypeScript and formatting. All eight CI test shards, the aggregate test job, formatting, and type checks pass on the final report/snapshot commit5eaf6ea.This remains a draft toward 9/12 Rockchip passes; the bottom-right case is still in progress. Main's latest performance improvements are included.