DIZK: A Distributed Zero Knowledge Proof System
summary
A distributed ZK-SNARK system that actually runs the setup and prover across a cluster and reaches billions of gates instead of millions. Implements a tailored witness reduction that avoids data-skew by replicating only a few Z entries with a hybrid join, beating block and skew-join approaches. Prototype is Java ~10k LOC on Apache Spark, evaluated on EC2 (256 machines → ~2 billion gates), reports ~10μs amortized cost per gate and linear scaling with machine count.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 0:30Zero Knowledge Proofs
- 1:51Peer to Peer Payments
- 4:53Challenges
- 6:39Architecture
- 10:49Offtheshelf approaches
- 13:54tailored approach
- 15:40system
- 16:03Evaluation
- 17:53Conclusion
- 18:42Questions
related talks
22mDIZK: A Distributed Zero Knowledge Proof System
48mzkSNARKs — Cryptographic Proofs for Scaling and Security Challenges
33mAn Introduction to Zero Knowledge Proofs
18mzkSaaS: Zero-Knowledge SNARKs as a Service
17mZK for the Enterprise - Quod Erit Demonstrandum (What Will Need to be Proven)
13mUnveiling the Future of Zero-Knowledge Proofs
38mNew Techniques for Zero-Knowledge: Leveraging Inefficient Provers to Reduce Assumptions, ...
36mScalable Zero-Knowledge Proof Made Easy