dangernoodle-io / TaipanMiner / 26734591816 / 2

90%

main: 90%

push

github

fix(webui): expected efficiency must use canonical pcore_mw (incl. board offset) (#476)

power_handler in routes.c was deriving expected efficiency from
vcore_mv × icore_ma, omitting BOARD_POWER_OFFSET_MW (5 W whole-board
offset). The canonical power source of truth is mining_stats.pcore_mw,
which asic_task.c computes as (vcore_mv × icore_ma / 1000) +
BOARD_POWER_OFFSET_MW. This made expected efficiency read ~30-40% too
optimistic — e.g. bitaxe-601 showed expected=11.46 J/TH vs actual=16.17;
with the offset, expected≈16.1 which matches actual.

fix: add mining_efficiency_jth(power_mw, hashrate_ghs) to the mining
component — pure math, host-testable, outside esp-platform — and route
all four efficiency fields in power_handler through it using pcore_mw as
the single power basis. the old vcore×icore re-derivation in the expected
path is removed entirely.

Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>

1653 of 1993 branches covered (82.94%)

Branch coverage included in aggregate %.

2914 of 3035 relevant lines covered (96.01%)

29.59 hits per line