Demo build. This is a software product for sale — not an investment, token sale, or financial service. Coins shown run on testnet and have no monetary value. Buyers are responsible for their own licensing, compliance and any mainnet launch.

Demo pool testnet only no real mining revenue
BTC testnet XMR testnet ETC testnet RVN testnet KAS testnet GCC testnet
Rebuilding a decade of writing

The blog.

Plain-English writing on the stack, mining, security and the GCC project. A decade-deep archive, being rebuilt and reorganised here after my last site went down. No paywall, no popups, no AI-generated filler.

Editor's pick · this week

Mining tutorial Featured
./asic-bench --model s21-pro
nominal: 234 TH/s
measured: 218 TH/s · 93% of spec
power: 3,460W ± 4%
efficiency: 15.9 J/TH
firmware: official v2.4.1 · no lock-in detected
./energy --kwh 0.18 --hours 24
draw: 83.0 kWh/day
electricity: £14.96/day @ £0.18/kWh
noise: 76 dB @ 1m
resale: check firmware lock before you buy
testnet demo · coins have no monetary value
hardware benchmark

May 18, 2026 · 9 min read

Choosing your first ASIC in 2026: what actually matters.

Wattage per terahash is the obvious metric. The non-obvious ones — firmware lock-in, resale value, electricity contracts, noise floor — are what burn newcomers. I benchmarked four current-gen miners against their spec sheets, and the gap is uncomfortable.

TL;DR

  • Spec sheets lie by ~7% in the real world
  • Firmware lock-in is the silent risk
  • Electricity and noise decide the build, not headline hashrate

Archive

Everything I've published.

sort:
Low volume · no spam

The occasional note.

An email when there's something worth sending: a new article, a notable change to the demo stack, and a couple of links worth your time. Unsubscribe is one click.

Writing since 2014 · unsubscribe in one click