Improve dense and peripheral fanout routing to 48/72 dataset cases - #189
Merged
Conversation
seveibar
changed the title
seveibar
marked this pull request as ready for review
September 9, 2026 17:53
Contributor
|
/benchmark |
1 similar comment
Contributor
|
/benchmark |
Dataset 31 — AM62L, RK3308, K230, i.MX6ULL, T113-S3, and AM3352 fanout benchmarkCommit: Full JSON/Markdown reports and logs Solved 27/72 selected samples. Completed 72/72; partial 0; errors 0; timeouts 45. Concurrency: 32; per-sample deadline: 120s; assignment budget: sample defaults. Only dataset-fanout31-am62l is benchmarked: 12 AM62L, 12 RK3308, 12 K230, 12 i.MX6ULL, 12 T113-S3, and 12 AM3352 cases. A case is solved only when all its SoC connections have validated fanout with the original constraints (135 for AM62L; 162 for RK3308; 171 for K230; 102 for i.MX6ULL; 128 for T113-S3; 322 for AM3352). RAM fanout and inter-chip routing are separate phases. Dataset revision: Per-sample results
|
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Dense BGA and peripheral fanouts can stall around shared source escapes, crowded boundary exits, and bus length constraints. This coordinates source reservations and complete bus groups, repairs boundary and source geometry, and adds bounded length repair through permitted signal layers.
The recorded complete benchmark passes 48/72, up from 18/72 with baseline solver
733d44aon the same dataset and budgets.637a6566a011be520b4fd170ad3f6da696ad4ff9.The full run completed all 72 samples with four workers, the original assignment budgets, and a 120-second deadline per sample: 48 solved, 24 timed out, zero partial results, and zero errors. Wall time was 22m57s. All 72 generated inputs match the original dataset inputs.
The report records measured runtime commit
090a27a8f5ecb38ab5f2000fcd1e8604312812db. A follow-up merges current main and repairs final normalization of shared plane copper; the full benchmark has not been rerun for that follow-up. Independent audits cover all 48 passing SVGs and 6,764 connections, with no geometry, ordering, bus/pair constraint, or source-pad reentry issues. The committed SVG hashes match the audited files.The AM62L reproduction with 60 real decouplers now completes all 135 fanouts, including 102 plane drops, and passes full routed-copper DRC. Final corner repair preserves same-net plane sharing introduced by main while rejecting foreign/unknown copper, signal-branch merging, and source-pad reentry. A focused visual regression also verifies that cached plane permissions cannot leak into signal candidates; the full AM62L snapshot is refreshed for the repaired corners.
Validation: the complete AM62L regression, focused geometry/clearance regressions, typecheck, and tracked-file formatting pass locally. CI on
d334643passes all eight test shards and their aggregate; the typecheck and formatting workflows also pass. The original benchmark implementation/report revision6c93ec5also passed all eight test shards and their aggregate.