OpenProof offers a decentralized ZKP (Zero-Knowledge Proof) infrastructure solution that leverages NVMe technology for scalable and efficient computations. Their ZKPU product enables large-scale, privacy-preserving computations across various ecosystems using open-source technology. This allows for interoperable ZKP implementations in modern systems.
Funding
Funding not disclosed
Founders
Product
Problem
Generating Zero-Knowledge Proofs (ZKPs) is computationally intensive, hindering their widespread adoption across various applications. Existing hardware solutions are often expensive, incompatible with modern infrastructure, and rely on closed-source designs, limiting accessibility and trust.
Solution
OpenProof provides an open-source infrastructure solution for scalable and efficient ZKP computations, leveraging a modular architecture and NVMe protocol support. Their ZKPU chip is designed to integrate seamlessly with existing server infrastructure and cloud environments, offering a cost-effective alternative to GPU and FPGA-based systems. By utilizing UC Berkeley's Chipyard framework and a modular design, OpenProof enables on-demand, high-throughput processing while ensuring transparency and adaptability to evolving ZKP workloads. The ZKPU architecture incorporates an integrated RISC-V CPU, NVMe endpoint, dedicated algorithm modules, and high-bandwidth memory to optimize performance and reduce data movement latency. OpenProof's approach facilitates distributed proof generation and elastic zero-knowledge workloads, paving the way for broader ZKP adoption in real-world applications.
Target Audience
The primary target audience includes developers, researchers, and enterprises seeking scalable, efficient, and transparent solutions for ZKP acceleration in blockchain, AI, and other privacy-preserving applications.
Features
- Integrated RISC-V CPU built on UC Berkeley’s Chipyard framework for efficient witness generation and reduced data movement latency
- NVMe endpoint for seamless data transfer and compatibility with existing operating systems and server infrastructure
- Dedicated algorithm modules optimized for MSM (Multi-Scalar Multiplication) and NTT (Number Theoretic Transform)
- High-bandwidth memory (GDDR/HBM) for memory-intensive operations
- Modular design that allows for the integration of eFPGA cores for enhanced flexibility and scalability
- Support for NVIDIA GPUDirect and ZKLink for high-speed, peer-to-peer communication between ZKPUs and GPUs
- Open-source architecture for transparency and community-driven innovation
- Python-based driver utilizing Linux’s open-source nvme-cli command-line tools for efficient host-FPGA communication