A Pioneer in Hardware Acceleration in the ZK Field: Interpreting Cysic's Vision and Layout

Recently, Ethereum founder Vitalik reiterated the importance of ZK-SNARKs in the Ethereum Danksharding roadmap at the Hong Kong Blockchain Summit, and pointed out the huge potential of ASIC chips for ZK acceleration. In fact, ZK technology is expected to bring paradigm-level changes in both Web3 and traditional industries, but currently, the generation efficiency and economic cost of ZK proofs remain key factors limiting its large-scale application.

To address this issue, several ZK acceleration projects have emerged. Among them, Cysic has become the focus of market attention due to its outstanding performance in the ZPrize competition and strong financing background. This article will delve into Cysic's product principles, business model, and the fundamental principles of ZK hardware acceleration.

Workflow of ZK Proof System

The core process of the ZK proof system includes:

1. Problem Setting: Define the content to be proven
2. Arithmetization and CSP: Transforming the content to be proved into logic gate circuits and polynomial form
3. Choose the appropriate proof system ( such as Halo, Plonk, etc. ) to generate ZKP programs.
4. The prover uses the ZKP program to generate a proof, which is submitted to the verifier for validation.

Currently, the widely used technology in blockchain is zk-SNARK, primarily leveraging its simplicity rather than its zero-knowledge property. The characteristics of zk-SNARK include high proof generation costs but low verification costs, and this asymmetry makes it very suitable for decentralized verification scenarios.

However, the high cost of ZKP generation will ultimately be passed on to user experience and transaction fees, limiting the large-scale application of ZK technology. To address this, some ZK acceleration projects such as Cysic and Ingonyama have emerged, attempting to reduce the generation cost of ZKP from different perspectives.

Main Computational Bottlenecks of ZKP Generation

In the process of ZKP generation, the two tasks that consume the most computing resources are MSM( multi-scalar multiplication) and NTT( number-theoretic transform), accounting for 80-95% of the proof generation time.

MSM is mainly used for computations on elliptic curves, which can be highly parallelized but require a large amount of memory. NTT is used to accelerate polynomial multiplication, involving random memory access, which is not hardware-friendly.

Optimizing the methods for these two calculations includes:

• Algorithm Optimization: New algorithms like Hyperplonk attempt to eliminate NTT operations.
• Hardware Acceleration: Improve computing efficiency using hardware such as GPU, FPGA, or ASIC.

Cysic's ZK Acceleration Strategy

The development strategy of Cysic is divided into four stages:

1. SDK developed based on GPU, providing solutions for ZK applications and integrating GPU resources from the entire network.
2. Utilizing FPGA to Implement Customized ZK Hardware Acceleration
3. Independently developed ASIC-based ZK Depin hardware
4. Build Cysic Network as a SAAS platform/pool to integrate various ZK computing resources.

Cysic Network: ZK Mining Pool and SaaS Platform

Cysic Network, as a ZK mining pool and SaaS platform, will integrate Cysic's own computing power and third-party computing resources. Its workflow includes:

1. ZK project direction agent submits proof task
2. Prover accepts the task and generates the ZK proof
3. The validator committee checks the validity of the proof and votes.

Cysic Network has also built a dedicated public chain, utilizing the Proof of Compute (PoC) consensus algorithm.

GPU, FPGA, and ASIC solutions

In terms of GPU acceleration, Cysic's self-developed CUDA SDK is 50%-80% faster than open-source frameworks.

In the FPGA field, Cysic has achieved the world's fastest MSM, NTT, and other modules. Its SolarMSM and SolarNTT perform excellently in large-scale computations.

In terms of ASIC, Cysic is developing two products: ZK Air for ordinary users and ZK Pro for large projects. These two devices are expected to be launched in 2025 and will build a stable and reliable ZK-DePIN network.

With the development of Cysic Network and the popularity of ZK hardware, the field of ZK computing may experience a boom similar to Bitcoin mining, with the market size expected to achieve explosive growth.