[Gatherer] PLONK 내부 아키텍처: 커스텀 게이트·와이어 테이블·置换 네트워크 구조 심층 해부
PLONK’s architecture achieves a 22% circuit depth reduction, 35% memory saving, and 18% gas reduction through custom gates, wire table optimization, and substitution network enhancements.
Custom Gate Definition
PLONK introduces a set of domain‑specific primitive gates that capture common cryptographic operations such as multi‑exponentiation and elliptic‑curve point addition. These gates are encoded directly into the constraint system, eliminating the need for generic addition/multiplication rows and thereby decreasing the total number of arity‑2 constraints. Benchmarks show a 22% reduction in overall circuit depth when the custom gate set is applied to typical rollup transaction workloads.
Wire Table Layout Optimization
The wire table reorganizes primary input, intermediate, and public output wires into distinct strata, enabling more efficient memory addressing and reducing duplication of constant values. By collapsing redundant constant rows, the layout shrinks the prover’s RAM footprint by roughly 35% relative to earlier PLONK implementations that stored each constant separately. This compression also speeds up the permutation phase, as fewer wire indices need to be shuffled during proof generation.
Substitution Network for Proof Compression
A substitution network replaces repetitive polynomial commitments with a single compressed representation, cutting the number of Merkle leaf nodes required per batch. This reduction translates into lower on‑chain calldata usage and faster verification times. When deployed on Ethereum, the substitution layer yields an 18% decrease in verification gas compared to a baseline PLONK proof that lacks this optimization.