Nockchain ($NOCK) 101: mining with zero-knowledge proofs
Nockchain miners don't hash. They generate zero-knowledge proofs, and the proof itself is the lottery ticket. What zkPoW means, what hardware wins, and where the pools are.
Every chain we cover replaces Bitcoin's hash lottery with something it considers more useful. Pearl picked matrix multiplication. Nockchain picked zero-knowledge proofs: miners repeatedly prove a fixed puzzle computation, and the hash of the proof itself serves as the block nonce. Your mining power is literally your proving power, which is why Nockchain describes itself as a protocol for compute markets and, by its own count, the largest proving network running.
zkPoW in one paragraph
A Nockchain miner takes a deterministic computation, generates a zero-knowledge proof that it ran correctly, and hashes that proof. If the hash clears the target, the block is yours. It keeps the lottery structure that makes proof of work fair and permissionless, but the ticket is a ZK proof instead of a meaningless digest. ZK proving is exactly the workload that rollups and verifiable compute pay real money for, so the capacity this network builds has a second life.
CPU first, GPUs now
Nockchain launched as CPU mining, and for a while a fast many-core box was the right tool. That era is over in practice: GPU provers now dominate, with miner support across NVIDIA's consumer lineup. It's the familiar arc every provable workload follows, CPU to GPU to whoever writes the fastest kernels. If you're sizing hardware today, you're sizing a GPU prover. We'll get a prover on our own bench and publish numbers under the benchmark methodology before recommending specific cards.
The pools
Two names come up for pooled NOCK mining. NockPool is the community pool, running PPLNS payouts with a 4% fee on withdrawals, a 50 NOCK minimum payout, and KYC verification, and it reports hundreds of active miners pushing past twenty thousand proofs per second. Golden Miner runs its own pool alongside a GPU miner that supports effectively any NVIDIA card. The pool-vs-solo math is the same as on any proof-of-work chain; if the variance argument isn't familiar, Owen walked through it in the pool article.
How it compares to Pearl
Same family, different muscle. Pearl wants sustained int8 tensor throughput and pays out a block roughly every three minutes network-wide. Nockchain wants fast proof generation, which leans on different parts of the GPU and on memory in ways we intend to measure rather than guess about. If you're allocating cards between the two, that's exactly the comparison our Benchmarks section exists to answer. New Nockchain coverage lands under the Nock tag.
