======================================================================== LOCK-FREE FLIP — FROZEN ANALYSIS (ledger §8) scipy available: True ======================================================================== === Throughput inference [variant A — disjoint (PRIMARY)] (write_ops_per_sec) === CONTROL : n=30 mean=2,720.1 sd=401.9 TREATMENT: n=30 mean=17,234.8 sd=2,337.3 diff (T-C) = 14,514.7 ops/s (+533.61%) 95% Welch CI [13,631.2, 15,398.1] Welch t = 33.5213 df = 30.71 p(two-sided) = 9.4038e-26 Cohen's d = 8.6552 (pooled) Mann-Whitney U = 900.0 p(two-sided) = 3.0199e-11 === S1 — checkpoint pause (veto: TREATMENT ≤ CONTROL) === [S1 PRIMARY] writer-observed p999 (us): control=262004.7 treatment=7147.1 [secondary 'ALSO'] checkpointer-side mean (us): control=127497.2 treatment=89703.7 [secondary 'ALSO'] checkpointer-side p99 (us): control=272911.7 treatment=305765.5 === S2 — write tail p50/p99/p999 (veto: T ≤ 1.10×C on p99/p999) === write_p50_us: control=77.7 treatment=342.2 ratio(T/C)=4.403 write_p99_us: control=118552.8 treatment=2395.2 ratio(T/C)=0.020 write_p999_us: control=262004.7 treatment=7147.1 ratio(T/C)=0.027 === S3 — peak RSS (single-arm-per-process pass; veto: T ≤ 1.25×C) === peak RSS (KiB): control=3,178,396 treatment=1,318,888 ratio(T/C)=0.415 === Throughput inference [variant B — contended (S4)] (write_ops_per_sec) === CONTROL : n=30 mean=3,799.2 sd=115.2 TREATMENT: n=30 mean=15,204.5 sd=582.3 diff (T-C) = 11,405.3 ops/s (+300.20%) 95% Welch CI [11,184.4, 11,626.3] Welch t = 105.2376 df = 31.27 p(two-sided) = 1.9016e-41 Cohen's d = 27.1722 (pooled) Mann-Whitney U = 900.0 p(two-sided) = 3.0199e-11 S4 cas_retries (treatment, per round): mean=101.1 max=207 === §8 DECISION RULE === -> BRANCH 1: PROCEED. throughput +14,514.7 ops/s (95% CI [13,631.2,15,398.1]), p=9.404e-26, d=8.66≥0.8, no S1/S2/S3/S4 regression ======================================================================== VERDICT: branch 1 = PROCEED EVIDENCE: throughput +14,514.7 ops/s (95% CI [13,631.2,15,398.1]), p=9.404e-26, d=8.66≥0.8, no S1/S2/S3/S4 regression ========================================================================