packages/chip/docs/pd/signoff/pdn-current/local-budget.md
Status: draft_local_evidence
Release use: prohibited_until_external_review
This budget is derived from local OpenLane artifacts only. It is useful for triage and first-article planning, but it is not a tapeout or board-fabrication release approval.
pd/openlane/runs/RUN_2026-05-18_05-41-42
Key local artifacts:
final/metrics.jsonfinal/metrics.csv57-openroad-irdropreport/irdrop.rpt20-openroad-generatepdn/openroad-generatepdn.logsignoff-run.yamlbenchmarks/power/local-estimates/e1-npu-openlane-npu-estimates.json| Rail | Board net | Voltage | Local current basis | Local budget |
|---|---|---|---|---|
| VDDCORE | +1V8 | 1.8 V | 5.554 mW / 1.8 V = 3.086 mA | 6.172 mA with 2x local margin |
| VDDIO | +3V3 | 3.3 V | unavailable | blocked pending IO/package/board load model |
For first-article planning only, both rails should start with a 25 mA bench current limit until workload-calibrated power and board loads replace this local estimate.
The local estimate artifact also records the NPU architecture-model result
(47.663 to 49.152 modeled INT8 TOPS) and the invalid arithmetic that would
result from dividing it by OpenLane power (8581.556 to 8849.730 TOPS/W).
That arithmetic is kept only as a blocker tripwire: the power and TOPS values
come from different substrates and cannot support sustained efficiency claims.
VPWR and VGND.VPWR worst local IR drop is 87.6137 uV at nom_tt_025C_1v80.VGND worst local bounce is 105.98 uV at nom_tt_025C_1v80.VPWR and VGND.