How does ETH 3.0 break through the performance bottleneck? This article reveals the breakthrough and upgrade plan of ZK technology behind it.

Written by: ZAN

Beam Chain: “ETH 3.0”

The just-concluded 2024 Devcon conference was lively and extraordinary, with the Ethereum ecosystem gathering in Bangkok. During this period, Ethereum and related ecological projects released the latest developments and updates. A series of new ideas, new trends, and new plans ignited the global Web3.

One particularly impressive plan is the ‘Ethereum 3.0’ proposal announced by Justin Drake, a researcher at the Ethereum Foundation.

He proposed a consensus layer upgrade proposal called “ETH Square 3.0” by the community – Beam Chain [1]。 The core idea is to reduce the computational and communication burden of validators with zero-knowledge proofs, lower the threshold required to become validators, and enable more validators to join the network to increase security, reduce computational redundancy, and improve the overall efficiency of the network. In addition, Justin also wants to take the opportunity of Snarkizing the consensus layer to solve the technical debt left by the Beacon chain due to the limitations of the times, and prepare for the future needs of quantum-resistant security.

In Vitalik’s The Verge roadmap, SNARKification of ETH is one of the important tasks, which involves the transformation of both the consensus layer and the execution layer into Zero-Knowledge Proofs. In Vitalik’s latest interpretation [2]Vitalik admits that the transformation of the consensus layer is a challenge, and it needs to be continuously considered and optimized in the subsequent work. The proposal of Beam Chain can be seen as Vitalik’s response to how to Snark-ify the consensus layer.

Beam Chain only follows the Consensus layer. However, making the Consensus layer Snark-friendly does not mean that all parts will be redesigned. The core of the blockchain is the State Transition Function, so the only modification required is to make the State Transition Function Snark-friendly, which happens to be an area where ZK excels. For the Consensus layer, the State Transition Function mainly consists of three parts, namely the per-slot transition function, per-block transition function, and per-epoch transition function. These three State Transition Functions include the execution of Consensus Algorithm and Staking logic, Block proposal, verification of execution layer results correctness, Merkle Tree Root verification, Consensus signature verification/aggregation, hash function calculation, stake and redemption operations involving account storage state access, etc. To make the Consensus layer Snark-friendly, it is essentially Snarkifying the above operations. It is worth noting that in the route to make the Consensus layer Snark-friendly, Ethereum still prioritizes Decentralization and diversity, so it does not force ZK Algorithm and protocol, but leaves it to the Proposer’s choice.

Image source:

Core Difficulty: Realtime Proving

To realize the vision of Beam Chain, the most crucial challenge is to improve the efficiency of proof, that is, how to compress the current proof time of tens or even hundreds of seconds to the millisecond level. This was discussed at the ZK Workshop II held on September 24th this year by AntChain OpenLabs, ZAN, and Pharos together.[3]Earlier, guest Justin shared a presentation titled ‘Realtime Proving’, during which he put forward core viewpoints in line with the vision of ‘Ethereum 3.0’ and proposed a solution to address performance bottlenecks. He encouraged the community to utilize ASIC acceleration methods to complete the entire process of zero-knowledge proof virtual machine from evidence generation to proof on the chip, aiming to increase proof generation time by over tenfold in the coming years, truly achieving Realtime Proving.

Empowering the performance and application of zk-SNARKs technology: we have always been on the road

AntChain OpenLabs has been committed to accelerating ZK proofs using hardware platforms such as GPU, FPGA, ASIC, etc., accelerating various ZK systems including Polygon ZKEVM Prover, Stone Prover, Halo2, and Plonky2, and forming multiple FPGA IP and GPU acceleration solutions, with some core operators achieving an acceleration ratio of over 100x.

The speedup ratio of some operators accelerated by CUDA relative to CPU in AntChain OpenLabs is shown in Table (1).

(Table 1)

Comparison of GPU-accelerated MSM implemented by AntChain OpenLabs and Open Source ICICLE (both with 4090 GPUs), as shown in Table (2).

( Table 2 )

Comparison of the performance of GPU-accelerated Circle FFT implemented by AntChain OpenLabs and Open Source Stwo Prover on a 16-core CPU, as shown in Table (three).

( Table 3 )

In addition, AntChain OpenLabs is also committed to the application exploration of Zero-Knowledge Proof, and in the direction of AI+ZK, it uses self-designed matrix multiplication Folding, Sumcheck protocol’s GPU+CPU collaborative optimization, efficient implementation of elliptic curves, etc., to verify large models, and zkLLM on 4 GPUs.[4][5](The original implementation does not include commitment and opening, the performance is tested after adding commitment and opening, for the LLama-7B model) The first token generation time has been reduced from 4 hours to about 18 minutes.

AntChain OpenLabs continually optimizes the performance and availability of large model verifiable computing based on GPU accumulation, and continues to make efforts in commitment scheme replacement and optimization, multi-instance Folding, and multi-token proof generation, aiming to build an efficient, easy-to-use, AI eco-friendly verifiable computing service, explore possible directions for the large-scale application of Zero-Knowledge Proof technology, and build a bridge linking Web3 and AI application connections.

At the same time, facing the threat of Quantum Computer, AntChain OpenLabs also actively explored, based on OpenSSL [6]Rebuilt a post-quantum version of the cryptographic library, supporting multiple NIST standard post-quantum algorithms. At the same time, facing the problem of more than 40 times storage expansion compared to ECDSA for post-quantum signatures, through consensus process and low-latency memory read optimization, the TPS of the anti-quantum blockchain can reach about 50% of the original chain.

It is worth following that the above-mentioned technical explorations have been applied in the software and hardware integrated acceleration solution Power Zebra launched by ZAN. This solution accelerates and optimizes the commonly used operators of Zero-Knowledge Proof, including multiple scalar multiplication MSM, number theory transformation NTT, polynomial solving H, polynomial opening proof, and other key components. It fully utilizes the computing resources, storage bandwidth, and PCIe bandwidth of the GPU, significantly improving the operation efficiency. Previously, after the zkWASM Open Source community Delphinus Lab introduced ZAN’s software and hardware integrated acceleration solution Power Zebra, the GPU computing performance increased by more than 20%, creating the highest record among similar services in the community.

Foresee

For a long time, the Ethereum community has mainly focused on optimizing the execution layer, including strategies such as improving BlockGasLimit and L2s. Among them, the execution layer’s Snarkification has already achieved significant progress with the help of L2 in the ‘Rollup Centric’ roadmap. The idea of Beam Chain opens the door to Snarkification of the Consensus layer. With the introduction of Beam Chain, it is believed that the community will be guided and inspired to focus on the transformation of the Consensus layer, bringing more exciting ideas.

It can be anticipated that the Snarkification of the consensus layer and execution layer will together complete the final puzzle of Ethereum Snarkification, making significant contributions to the scalability and decentralization of ETH. At the same time, with the advancement of the Beam Chain proposal, the Virtual Machine based on Zero-Knowledge Proof technology will play a core role in the ETH ecosystem. By introducing SNARK proofs and post-quantum cryptography, along with lower stake thresholds and faster block confirmation times, Beam Chain will bring stronger decentralization and higher performance to ETH.

References

[1]

[2]

[3]

[4]

[5]

[6]