On April 24th, a blockchain virtual machine named DTVM (Deterministic Virtual Machine) was announced as open-source, sparking heated discussions within the developer community. According to its publicly released technical paper, DTVM achieves approximately 30 times faster performance than traditional interpretation execution through its innovative JIT engine and full-chain optimization, setting a new industry benchmark while maintaining full compatibility with the Ethereum ecosystem.
A blockchain virtual machine is a specialized computing environment running on a blockchain network, processing numerous computations and transaction instructions. It serves as the "operating system" for smart contracts and decentralized applications (dApps). In recent years, the increasing complexity and widespread adoption of blockchain applications have placed higher demands on virtual machine capabilities, prompting industry professionals to seek new solutions. DTVM addresses the challenges of performance, determinism, and ecosystem compatibility within blockchain networks.
Utilizing a delayed JIT compilation technique, DTVM reduces the initial smart contract call latency to 0.95ms, a more than 20-fold speed improvement over the currently leading WebAssembly JIT virtual machine. It demonstrates a significant advantage in complex computations; in PolyBench tests, performance improvements reached 63% for certain computationally intensive tasks, providing efficient support for RWA, DeFi, on-chain AI, and other scenarios.
As detailed in the paper, DTVM seamlessly integrates with the Ethereum EVM ecosystem, supporting six mainstream programming languages: Solidity, C++, Rust, Java, Go, and AssemblyScript. This eliminates the "language barrier" for Web3.0 developers, allowing for smooth transitions across multiple language development environments. Furthermore, DTVM simplifies the Trusted Computing Base (TCB), a system security protection mechanism, reducing the codebase size to less than 27% of WASM. This significantly shrinks the potential attack surface while ensuring secure and efficient system operation.
Notably, DTVM, as a next-generation virtual machine, integrates SmartCogent, an AI tool suite based on large language models (LLMs). This AI automates the entire development process, from coding and auditing to deployment. Tests show an 81% accuracy rate in detecting logic vulnerabilities and an 86% success rate in fixing them, significantly outperforming similar open-source projects. Developers can also use natural language descriptions to have the AI generate compliant contract code and complete security audits. This marks a transition in Web3.0 technology development from manual coding to truly "AI-native development," with artificial intelligence becoming a fundamental driver of innovation.
Ant Chain from Ant Group's financial technology arm, Ant Group's Supercomputing team, and Pharos are among the core technology teams that contributed significantly to DTVM's development. Currently, the core technologies of DTVM, including the core engine and multi-language compiler SDK, have been open-sourced under the Apache 2.0 license. Developers can access the complete code and documentation via GitHub.