RailOptimization

RailOptimization v2.0-snapshot 1

This snapshot introduces the next-generation rail update engine — with state-changing updates 7-9x faster and unchanged updates up to 30x faster than vanilla.

🚀 Performance

• State-changing updates (toggles): 7.5-9x faster than vanilla
• Unchanged neighbor updates (repeated redstone pulses on an unchanged rail line): up to 30x faster
• Long rail lines with powerLimit=64: about 12x faster than v1.6
• Figures are statistical medians over 32 benchmark rounds on our test machine; real-world results vary with hardware and load

✨ New Features

Zero-Allocation Update Engine

• Rebuilt the powered rail update pipeline on pooled, per-thread update contexts with no allocation in the mod's own hot paths
• The rail search cache and update memo store their metadata in single bit-packed machine words, keeping the hot tables inside the L1 cache
• Straight and ascending rail depowering share a single step-table path with no branching per rail

Memoized Unchanged Updates

• Unchanged rails validate repeated neighbor updates against a memoized state and skip the rail line re-search entirely
• Redstone clocks and repeated pulses no longer re-traverse the full powered rail chain when nothing changed

🛠 Improvements

• Powered rail searches prune with the exact cached cost instead of re-walking known distances
• Signal searches skip conductor directions already known to be invalid
• Rail power limit values are clamped to the valid range

🐞 Fixes

• Fixed out-of-range rail power limit values being stored as-is instead of being clamped

⚠️ Snapshot Notes

• This is a snapshot intended for testing before the final v2.0 release
• This release changes no observable rail or redstone behavior — it is a pure performance update, verified by the parity test suite