clock lock at cap shipped — verdict ON-BY-CARD written 2026-09-22T00:10:49Z card GPU-edff232c-7dbf-2bac-07fb-921a7c9eecff asked of THE CARD (clocks.sm against the lock's floor and against this board's own at-rest clock; utilization.gpu for whether it was at rest; clocks.max.sm against the ceiling). Never the unit. clocks.sm min/max 1192 / 1192 MHz over 12 reads clocks.max.sm 3090 MHz utilisation max 0 % — at rest for the whole sample: yes lock range declared 1200,2550 MHz (ai-perf.service) floor · tolerance 1200 MHz · 25 MHz (the driver snaps 1200 to 1192 on this board) evidence the card was AT REST for the whole sample (utilization.gpu read 0 % on all 12 reads) and clocks.sm never left the lock's floor: min 1192 MHz, max 1192 MHz, against a declared floor of 1200 MHz and the driver's 25 MHz quantization tolerance (the driver snaps 1200 to 1192 on this board, measured 2026-09-21T23:13Z). Unlocked and at rest this board reads 232 MHz or less (232 MHz at 17:30Z, 180 MHz at 18:11Z), so a resting card pinned at the floor is the lock holding it there. clocks.max.sm reads 3090 MHz, which is this board's own unlocked maximum with OR without the lock and is therefore not part of this verdict. unit state active — CONTEXT ONLY, NOT EVIDENCE: ai-perf.service is Type=oneshot and reads 'active (exited)' for the whole uptime regardless of a later 'nvidia-smi -rgc'. what this means benchbox, which produced the desktop rows this bench compares against, runs no clock lock. OFF-BY-CARD means like-for-like, and it is the REQUIRED verdict for the fixed and dynamic-boost passes; ON-BY-CARD means the board could not clock below its floor, and it is the REQUIRED verdict for the shipped pass, whose whole subject is what this box does with its own lock on; ON-CONSISTENT is the same confound without a proof (and has never been read on this board); UNDETERMINED means neither was established and no claim is made.