TwistPhysics · 3SOS-C research · 2026-10-02

LOCK v1 RC1

A public research checkpoint for verified execution. The page keeps the published benchmark results, declared scope and source provenance visible while the public evaluation materials are being reduced to a minimal verification demo.

PASS WITH SCOPE · controlled software workbench

Certified execution research · guarded execution · independent validation · commit or quarantine

15.83%lower published median wall time for LOCK_TRACE vs strict VERIFY_EVERY_MOVE

12 / 12paired timing rounds favored LOCK_TRACE in the published campaign

6,240 / 6,240masked compensating fault pairs detected by exact trace evidence

320 / 320declared single input-delta faults detected

Correct interpretation: LOCK_TRACE recovered part of the cost of strict verification, but it remained 22.77% slower than RAW_SUBSET. This is not a claim that LOCK is faster than unprotected Klipper.

Public evaluation materials

The full RC1 workbench is no longer distributed from this page. A reduced verification demo will be published here after review, while the benchmark results, declared scope and source provenance remain public.

Pinned Klipper commit

Published RC1 package identity retained for provenance: SHA-256 7beb1092226941b3e7ccae732e90e8ec09b237a849486d8699a4f2d955e3e4a3.

Scope

✓ Audited logical LOCK core and controlled source-derived Klipper subset.
✓ Masked-fault and dependency/epoch guards exercised in the declared software classes.
✓ Reproducibility manifest included.

Not yet claimed: full Klipper daemon integration, MCU/trapq control, physical printer validation, hardware cache/memory counters, safety certification, or final LOCK v1 freeze.