490-line production proof-of-concept. Measured outcomes on the reference bench: 36.5% power reduction, clock jitter cut from 67.5ps to 30.5ps, timing violation rate from 59 to 13 per second. Every state transition is SHA-256 chained into an audit trail. Ship-ready — needs packaging.
CSEL continuously samples the operating state of the controlled system — clock domain, thermal envelope, switching activity — and locks into a coherence band that minimizes wasted power and timing slack consumption. Every state transition is hash-chained, so the savings claim is auditable, not asserted.
Compact, auditable, fully instrumented. The proof-of-concept is the shipping core, not throwaway research code. Reviewable by a single engineer in an afternoon.
Bench measurement against the unlocked baseline. The savings come from coherence-state locking — fewer wasted switching events, tighter clock-tree behavior, less rework cycles burned on timing-marginal paths.
More than 2x reduction in measured jitter. Direct consequence of locked-coherence operation. The same envelope that buys power savings buys margin against setup/hold failures on critical paths.
Aggregate timing-violation rate falls more than 4x under lock. Fewer paths blowing setup, fewer error-replay cycles, less downstream rework. The savings compound in any pipelined system.
Every transition into and out of the locked coherence state is hashed and chained. Tamper attempts surface at verification time. Compliance teams and energy auditors get a forensic record, not a vendor dashboard claim.
The lock decision uses a dual-force spectral metric — opposing inputs balanced against a coherence target. The same engineering primitive used elsewhere in the WHL stack for governed equilibrium control.
The core is production-grade. The remaining work is integration packaging — installer, telemetry endpoint, customer-facing dashboards. Pilot deployments can begin against the current PoC under a co-engineered SOW.
CSEL does not require replacing controllers, PDUs, or platform firmware. It runs as a governed observer-and-locker alongside existing operations infrastructure and posts decisions through the audit chain.
The same 490-LOC core ships into three named verticals with different packaging and different SLAs. The underlying measurement — 36.5% — has to be reproduced on customer hardware before any contract closes.
Rack-level or row-level deployment. The power-savings line item is large enough to fund the audit. CSEL produces the cited evidence energy procurement teams need to defend the projected savings to finance.
Critical-load environments where timing-violation reduction matters as much as kilowatt savings. The SHA-256 audit chain aligns to existing biomedical equipment governance documentation.
Air-gapped or sovereign-cloud deployments. CSEL operates inside the enclave, posts its audit chain into the local trust store, and surfaces savings without exfiltrating telemetry to a vendor cloud.
CSEL is built for organizations whose energy line item and timing-violation rate are both first-class operational metrics.
Instrumented benchmark output from the CSEL proof-of-concept run.
$ python whl_csel/benchmark.py --runs 10
=== WHL CSEL Power & Timing Benchmark ===
Baseline (uncontrolled):
Mean power: 287.4 W
Jitter (ps): 67.5
Timing violations: 59/sec
With CSEL active:
Mean power: 182.5 W (-36.5%)
Jitter (ps): 30.5 (-54.8%)
Timing violations: 13/sec (-78.0%)
✓ SHA-256 chained audit: 1000 events, chain verifiedMeasured: 36.5% power reduction, 54.8% jitter reduction, 78% drop in timing violations. SHA-256 audit chain verified across 1000 events.
We bring the 490-LOC core and the measurement protocol. You bring the baseline. The audit produces a hash-chained record your energy and compliance teams can both sign off on.