Tabby: A Synthesis-Aided Compiler for High-Performance Zero-Knowledge Proof Circuits
summary
Tabby is a compiler that converts high-level ZK programs into optimized arithmetic circuits by lowering to a domain-specific IR, decomposing the dataflow into independent subproblems, and applying sketch-based program synthesis to generate compact low-level implementations. Equivalence is checked with SMT-based verification, codegen targets multiple proof backends, and artifacts are available. Evaluation on real ZKP apps shows ~70–80% reductions in proving time and circuit size versus mainstream compilers while keeping compilation time comparable.
AI-written summary · about
chapters (9)
- 0:00Introduction
- 1:08Zero-knowledge proofs
- 3:33Efficient ZK circuit challenges
- 5:00Motivating ZK optimization
- 8:55Invariant synthesis with program analysis
- 11:20Tabby’s compiler pipeline
- 12:24Evaluation vs mainstream compilers
- 14:32Ablation study and takeaways
- 14:55Summary and open-source
related talks
35mEfficient Zero Knowledge Protocols: The Modular Approach
41mGKR based Zero-Knowledge Proofs
25mHigh-Speed zkSNARKs without Trusted Setup
30mPractical Zero Knowledge Proofs
51mDistributed Zero-Knowledge Proofs and Applications to Secure Computation
32mSoK: Private Programmability on Blockchain Systems
14mStatic Analysis for Zero-Knowledge Programming Languages
22mTurning Catala into a Proof Platform for the Law
