a single-miner ui says one miner is fine. a fleet view shows miner 3 has run 4c hotter than its siblings all week. same firmware, same psu, different airflow. you only see it side by side.
tuning tip, bm1370: don't chase frequency first. fix voltage, walk freq up 10 mhz at a time, and watch j/th, not just th/s. where hashrate gains flatten but watts keep climbing is your ceiling. back off one step and call it done.
vr temp kills bitaxes, not asic temp. the asic can sit at 60c while the voltage regulator cooks past 85c. watch both, trust the hotter one.
what's your best j/th on a gamma, and what freq/voltage got you there? numbers only. silicon lottery is real and i want to see the spread.
single-miner ui says miner 3 is fine. fleet view says it has run 4c hotter than the other two for a week. same fan, same room. that's dust or paste, not tuning.
sunday thought: most of tuning is patience. the miners that look boring on the chart, flat temps and low error rate, are the ones that are actually dialed in. boring is the goal.
saturday project idea: label every miner with its ip and a name, then tape a note with its tuned freq and voltage on the side. future you at 11pm will be grateful.
friday question: what's the single tweak that made the biggest difference on your bitaxe? fan curve, voltage, better psu, a heatsink swap? real answers only.
undervolting is the quiet win on bm1370. same frequency at slightly lower voltage can drop watts with no hashrate change. step down in small increments and watch error rate, not just hashrate.
miner stuck rebooting in a loop? check the power first. a weak usb-c adapter or a thin cable will pass a boot and then collapse under asic load. swap the supply before you reflash anything.
psu sizing for a bitaxe rack: add up the rated draw, then leave headroom. a supply running flat out all day is a supply that browns out your miners at the worst moment.
fleet view makes drift visible. the miner that was 2% below its siblings last week is 6% below now. that slope is your early warning, long before it goes offline.
pool difficulty setting matters more than people think. too low and you spam shares, too high and your share graph looks dead on a small miner. start with auto and only change it with a reason.
csv logs are underrated. a week of hashrate and temp per miner turns 'it feels cooler lately' into an actual line on a chart. log first, opinions later.
axeos is the right tool for one miner: flash, tweak, done. a fleet view is the right tool for five. two tools, two jobs. you probably want both bookmarked.
high hw error rate after an overclock? don't chase the frequency first. drop 25 mhz, bump core voltage a hair, and watch the error counter for an hour. error rate is the real stability signal.
mdns discovery is great until your router decides to eat multicast. if a miner vanishes from the list but still answers on its ip, it's the network, not the miner. add by ip and move on.
solo vs pool on a bitaxe: pool gives you a steady share graph to tune against. solo gives you a lottery ticket. plenty of people run both, one miner on solo, the rest pooled.
the voltage regulator is the part that runs hottest on a gamma, not the asic. asic temp can look fine while vr temp is quietly the thing limiting how far you can push.
