Fundamentals

How Mining Works & Pools

SHA-256, shares, hashrate, difficulty, pools and payout schemes - the concepts behind what your miners are doing.

What a miner actually does

An ASIC miner repeatedly runs double SHA-256 (SHA256d) on a block header, changing a number called the NONCE each time, hunting for a hash below the network TARGET. It's a lottery of quadrillions of guesses per second. The chips do only this one calculation - extremely fast.

Hashrate = guesses per second. Units: TH/s (tera), PH/s (peta), EH/s (exa). A modern unit does ~100-270 TH/s.
Bitcoin mainnet retargets difficulty every 2,016 blocks toward a long-term ten-minute average. Actual intervals are random; hashrate changes affect block rate before the next retarget.
A single miner almost never finds a whole block - that's why farms join a POOL.

Shares: accepted, rejected, stale

A pool gives each miner easier 'share' targets to prove it's working. Shares are how the pool measures your contribution and pays you.

Type	Meaning	What high counts signal
Accepted	Valid share credited by the pool	Healthy - this is what you want
Rejected	Submission not credited by the pool	Inspect the reject reason and job/session context
Stale	Share arrived too late (old job)	Network LATENCY - go wired, closer pool endpoint
HW error	Chip produced a bad nonce	A weak/dying chip; under ~1% is normal

A rejected share is a submission the pool did not credit. Read its reason and job/session context first: stale, duplicate, target or authorization issues differ from invalid computation. Corroborate hardware faults with logs and operating data before replacing a board.

Pools & failover

A pool aggregates thousands of miners so payouts are steady instead of a rare jackpot. Miners connect via the Stratum protocol.

Pool URL format: stratum+tcp://host:port (e.g. a Foundry USA endpoint on :3333).
Worker naming often uses account.workername. The account controls credit; the worker suffix labels a device. A suffix typo may create a different worker under the same account. Authorization and payout behavior are pool-specific.
You configure up to 3 pools; they run in PRIORITY order for failover, not load-balancing. Pool 1 primary, 2/3 backups.
Stratum V2 adds encryption and lets the miner help build blocks; V1 is still the norm.

Payout schemes (why the farm cares)

How the pool pays out changes revenue smoothness and who eats the variance.

Scheme	Plain English
FPPS	Full Pay Per Share - pays expected subsidy value plus a pool-defined transaction-fee component per share, subject to pool fees and terms; the pool bears block-finding variance.
PPS+	Pay Per Share + a share of tx fees; steady income.
PPLNS	Pay Per Last N Shares - rewards loyalty, more variance, can pay more long-term.
SOLO	You keep the whole block if you find one - huge variance, usually only for very large hashrate.

Why uptime is everything

A stopped miner contributes no new work, though past-work payouts may continue. A throttled miner can still submit accepted shares; revenue and efficiency depend on its operating point and pool terms.

Efficiency (J/TH) sets running cost; uptime sets whether you earn at all.
Curtailment = deliberately pausing/throttling miners on command (for grid demand-response). The OPERATIONS side is a tech task; firmware can drop a unit to near-zero watts in seconds.
A dead board = ~1/3 of a unit's revenue gone until fixed - which is why fast diagnosis matters.

What a miner actually does

Hashrate = guesses per second. Units: TH/s (tera), PH/s (peta), EH/s (exa). A modern unit does ~100-270 TH/s.

Bitcoin mainnet retargets difficulty every 2,016 blocks toward a long-term ten-minute average. Actual intervals are random; hashrate changes affect block rate before the next retarget.

A single miner almost never finds a whole block - that's why farms join a POOL.

Shares: accepted, rejected, stale

A pool gives each miner easier 'share' targets to prove it's working. Shares are how the pool measures your contribution and pays you.

Type	Meaning	What high counts signal
Accepted	Valid share credited by the pool	Healthy - this is what you want
Rejected	Submission not credited by the pool	Inspect the reject reason and job/session context
Stale	Share arrived too late (old job)	Network LATENCY - go wired, closer pool endpoint
HW error	Chip produced a bad nonce	A weak/dying chip; under ~1% is normal

Pools & failover

A pool aggregates thousands of miners so payouts are steady instead of a rare jackpot. Miners connect via the Stratum protocol.

Pool URL format: stratum+tcp://host:port (e.g. a Foundry USA endpoint on :3333).

Worker naming often uses account.workername. The account controls credit; the worker suffix labels a device. A suffix typo may create a different worker under the same account. Authorization and payout behavior are pool-specific.

You configure up to 3 pools; they run in PRIORITY order for failover, not load-balancing. Pool 1 primary, 2/3 backups.

Stratum V2 adds encryption and lets the miner help build blocks; V1 is still the norm.

Payout schemes (why the farm cares)

How the pool pays out changes revenue smoothness and who eats the variance.

Scheme	Plain English
FPPS	Full Pay Per Share - pays expected subsidy value plus a pool-defined transaction-fee component per share, subject to pool fees and terms; the pool bears block-finding variance.
PPS+	Pay Per Share + a share of tx fees; steady income.
PPLNS	Pay Per Last N Shares - rewards loyalty, more variance, can pay more long-term.
SOLO	You keep the whole block if you find one - huge variance, usually only for very large hashrate.

Why uptime is everything

Efficiency (J/TH) sets running cost; uptime sets whether you earn at all.

Curtailment = deliberately pausing/throttling miners on command (for grid demand-response). The OPERATIONS side is a tech task; firmware can drop a unit to near-zero watts in seconds.

A dead board = ~1/3 of a unit's revenue gone until fixed - which is why fast diagnosis matters.