Chip Failure
Chip failure is when one chip on a hashboard stops responding. The log shows fewer chips than it should, and the board's hashrate drops or goes to zero.
What chip failure is
A hashboard is a chain of dozens of chips, around a hundred per board on current models. Data moves along the chain, so one silent chip can drag the whole chain down with it: the controller stops seeing both that chip and everything behind it.
The causes are almost always physical. Overheating and constant cycles of heating and cooling crack the solder under a chip. Dried thermal paste and a skewed heatsink roast one patch of the board. Moisture, or immersion without preparation, brings corrosion. Factory defects happen too, and they show up in the first months.
In the log it looks like an incomplete chip count on a chain, or a line like chain 2: find 0 asic. A reboot sometimes brings the chip back, but not for long: a crack in the solder only grows with heat.
The cure is soldering. The chip gets replaced at a rework station, with a stencil for the balls, then the board is washed and run under load. At home, without the experience and the gear, you are more likely to finish the board off than fix it.
Quick reference table
| What the log shows | an incomplete chip count on a chain |
| Common causes | overheating, cracked solder, corrosion, dried paste |
| What is lost | the board's hashrate, sometimes the whole chain |
| What cures it | chip replacement at a rework station |
| What does not | a reboot: the chip comes back briefly |
What to do when a chip drops
Read the kernel log and write down which chain and how many chips it found. That line means more to a repair shop than any description in words.
Check the simple things before soldering: heatsink seating, dust, fan speeds, the state of the paste. Sometimes the chip is alive and overheating is the culprit.
Watch when the chips disappear. If it happens once things warm up, it is almost certainly the solder, and from there only the station helps.
Hand over the board whole, together with the miner or at least the power supply: the shop needs to test the repair under its native load.
An example from a log
A miner holds two thirds of its rated hashrate. The log says chain 1: find 0 asic, the other chains are intact. That is a dropped board: one of three is not working, and the hashrate figure matches the arithmetic. After a chip swap and a wash the board passed an overnight run and came back to spec.
Related terms
Where to go next on the site
Hardware
- ASIC Miner knowledge base
- ASIC Chip and Hashboard knowledge base
Guides
- ASIC Manufacturers knowledge base
Questions and answers
Can the failure clear up on its own
A chip sometimes comes back to life after cooling down, but that is a pause, not a recovery. The crack in the solder grows with every heat cycle, and the chip will go silent again.
Is the firmware to blame
Firmware does not break solder, but reckless overclocking heats the board harder and brings the failure closer. Autotuning with a sane profile is safer than manual overclocking.
What does the repair cost
Replacing one or two chips costs noticeably less than a new board, so diagnostics almost always pays. A board after corrosion costs more, and there the bill is not counted in chips.
How the terms connect
Every link in the chain is clickable. Orange marks where you are now.
Looking for an ASIC miner
The catalog holds 212 models. You can compare them by hashrate and by joules per terahash, then plug your own rate into the calculator and see what stays in your pocket.
Page written and checked by Denys Klimchuk. Updated .