130,827 Latest block · snapshot of Oct 20

Tool · Mining

Mining odds

How long one GPU would wait, on average, for a block

Mining odds

How long one GPU would wait, on average, for a block

Total: 600 TMAC/s

Average wait for one block

218 d 1 h

1.13 × 10²² multiply-accumulates per block (difficulty × 2.815 × 10¹⁴) ÷ 600 TMAC/s

Chance of at least one block in…

a day 0.46%
a week 3.16%
a month 12.9%
a year 81.2%
Network, by design
445 blocks/day
one every 194 s
Network rate
64.9 EMAC/s
snapshot, Oct 20, 2026
Your share
0.00092%
≈ 0.0041 blocks/day

Estimate, not advice Averages only: a block is luck, so the real wait can be much shorter or longer. GPU speeds are rough figures from pool data (arXiv 2606.04819), not benchmarks. This ignores pools, costs and prices. Many miners join a pool for small, steady payouts instead.

Runs in your browser · asks PRLScan only if you press “Use live” src/data/facts.ts · MINING.macsPerDifficulty

What it does

Type a GPU’s speed (or pick a rough preset) and how many you run. The tool divides the work one block needs by your speed and shows the average wait, the chance of at least one block in a day, week, month and year, and your share of the network.

How to read it

  • Work per block = difficulty × 2.815 × 10¹⁴ multiply-accumulates. That factor is 2⁶⁴ ÷ 65,535, because difficulty 1 is bits 0x1b00ffff.
  • Average wait = work per block ÷ your speed. At the snapshot difficulty a single H100 at about 600 trillion multiply-accumulates per second waits around 218 days.
  • Chances follow from blocks arriving at random: the chance of at least one in time t is 1 − e^(−t ÷ wait). Even a GPU with a seven-month average wait has a few percent chance in any given week, and no guarantee in a year.
  • Use live difficulty asks PRLScan for today’s figure; otherwise the dated snapshot is used.

Pearl specifics

  • Pearl’s hashrate counts multiply-accumulates, not hashes: every ticket costs h·w·k of them, and the winning bound scales with that work, so each multiply-accumulate has the same tiny chance whatever the matrix shape.
  • The network aims for one block every 194 seconds, about 445 a day, and retunes difficulty after every block.

What’s simplified

These are averages, not predictions. GPU speeds are rough figures from pool data (arXiv 2606.04819), not benchmarks; real throughput depends on the miner software and on how much of the work is noise and checking. Pools, fees, power and prices are left out on purpose. This is not profit advice.

Under the hood

Runs in your browser; it contacts PRLScan only when you press Use live difficulty. The constant comes from src/data/facts.ts (research doc D §b.17: work per block = 2²⁵⁶ ÷ target), and src/lib/block.ts’s difficulty and work functions are unit-tested against real mainnet bits.

Suggested