Repository navigation
fit-config: co-sweep coordinator.max_in_flight (roadmap follow-up) - #2
Merged
Merged
Conversation
The fit finder now varies coordinator in-flight depth as a fourth candidate axis alongside grid shape / op distribution / memory split. Every in-flight slot costs LUTs in wau_resource_model_v1 but the depth only buys makespan when independent flows overlap, so swept depths are workload-bounded (_fit_max_in_flight_values): single-flow workloads collapse to depth 1 (cycle-identical there, reclaiming unused-slot LUTs) while multi-flow workloads try 1, powers of two, the base value, and the flow-count ceiling (clamped to [1,16]). The LUT tiebreaker keeps the cheapest depth that doesn't cost makespan. Kept additive: CandidateKnobs.max_in_flight defaults to None so arch-search reports stay byte-identical (no _mif suffix, no knobs key); only fit-config sets it. --max-in-flight overrides the swept set; the set is reported as max_in_flight_swept and encoded per candidate as a _mif<N> id suffix (round-tripped by emit_fit_config). Tests: FitMaxInFlightSweepTests + arch-search byte-identity guard + two CLI tests. Full Python suite 112 pass; iverilog RTL suite passes. Co-Authored-By: Claude Opus 4.8 <[email protected]>
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
The fit finder now varies coordinator in-flight depth as a fourth candidate axis alongside grid shape / op distribution / memory split. Every in-flight slot costs LUTs in wau_resource_model_v1 but the depth only buys makespan when independent flows overlap, so swept depths are workload-bounded (_fit_max_in_flight_values): single-flow workloads collapse to depth 1 (cycle-identical there, reclaiming unused-slot LUTs) while multi-flow workloads try 1, powers of two, the base value, and the flow-count ceiling (clamped to [1,16]). The LUT tiebreaker keeps the cheapest depth that doesn't cost makespan.
Kept additive: CandidateKnobs.max_in_flight defaults to None so arch-search reports stay byte-identical (no _mif suffix, no knobs key); only fit-config sets it. --max-in-flight overrides the swept set; the set is reported as max_in_flight_swept and encoded per candidate as a _mif id suffix (round-tripped by emit_fit_config).
Tests: FitMaxInFlightSweepTests + arch-search byte-identity guard + two CLI tests. Full Python suite 112 pass; iverilog RTL suite passes.