After reading the latest white paper released by Reddit, I found that the automated AI execution is indeed integrated into the EVM narrative, which is equivalent to filling the gap in the AI track of the entire Ethereum ecosystem. It makes sense, so why can the parallel EVM be seamlessly connected to AI? What is the logic and technical principle behind it? Let me briefly explain my understanding:
1) The “Parallel EVM” narrative has always been characterized as a key battle for the old and clumsy EVM ecosystem to bridge the gap with high-performance chain technology iterations such as Solana and Sui. Therefore, the previous market hype expectations for Sei and Monad’s huge financing of US$225 million have pushed Parallel EVM to an unprecedented height.
In comparison, Reddio, a parallel EVM public chain led by Paradigm, seems to be much more low-key. There is no hype about market expectations such as financing, ICO, and KOL rounds. It has only been showing off its stable TPS data of tens of thousands on its test network. Recently, the official snapshot was announced, and it is obvious that it wants to be the first to come on stage and verify the ecological niche value of parallel EVM in the Ethereum ecosystem.
2) So, why is parallel EVM an effective supplement to the technical bottleneck of the Ethereum ecosystem? Simply put, the original single-threaded execution + serial execution in transaction order of EVM is an inherent limitation. Parallel EVM achieves the simultaneous execution of large-scale batch transactions by utilizing the parallel computing power of modern hardware (CPU, GPU), plus some I/O asynchronous storage processing, state access optimization processing, etc.
The technical implementation logic revealed in the Reddio white paper is roughly as follows: Reddio has an execution network composed of GPU nodes, and through the CUDA "coding translator", it converts general EVM opcode programs into complex intensive computing tasks that can be calculated in the GPU. In addition, other I/O asynchronous storage optimizations, state access management optimizations, optimistic concurrency control, etc. have been added to achieve the ability to process transactions in parallel;
3) Since parallel EVM essentially takes advantage of the performance of "hardware", AI application scenarios naturally require large-scale parallel computing and intensive computing processing. A powerful set of hardware can be used in parallel EVM and AI application scenarios at the same time. In this way, another layer of narrative imagination space for parallel EVM + AI is opened. The parallel EVM chain can deploy large AI models on the chain, and allow smart contracts to directly control and schedule AI. At the same time, ZK, TEE and other related data privacy and verifiability capabilities can be applied, etc., to achieve the native integration of blockchain and AI. For example, AI real-time reasoning, AI Oracle, off-chain AI transaction strategy optimization, etc.
