技术解析Artela：为什么“并行EVM”关乎以太坊EVM生态的存续之战？

最近，Paradigm下重注领投了Monad一轮2.25亿美元的巨额融资，引发市场对“并行EVM”的强烈关注。那么，“并行EVM”到底解决了什么问题呢？发展并行EVM的瓶颈和关键在哪里？在我看来，“并行EVM”是EVM链迎击高性能layer1链的最后一博，事关以太坊EVM生态的存续之战。为什么呢？接下来，让我们来谈谈我的理解。 由于以太坊EVM虚拟机只能“串行”处理交易，这使得EVM- Compatible layer1链以及EVM兼容的layer2链受到了性能制约，因为它们都基于同一套框架处理状态和交易Finality。 然而，像Solana、Sui、Aptos等主打高性能的layer1链天生具有可并行的优势。在这种背景下，想要正面抗衡高性能layer1链的挑战，EVM链就必须提升“并行”处理能力。那么该如何做呢？涉及技术原理和细节，我将以并行EVM新锐链@Artela_Network为例展开说明。 以Monad、Artela、SEI等为代表的强化型EVM layer1链，它们在高度兼容EVM的基础上显著提升了TPS，并赋予了交易在拟EVM环境下的并行能力。这些独立的并行EVM layer1链具有独立的共识机制和技术特性，但仍以兼容并扩展EVM生态为目标，相当于以“换血统”的方式重塑EVM链，从而服务于EVM生态； 以Eclipse、MegaETH等为代表的扩容型layer2 EVM兼容链，则利用layer2链独立的共识和交易“预处理”能力，在批量交易上链之前对交易状态进行筛选和处理，并可选择其他任意链的执行层来确认交易状态。这种方法将EVM抽象为可插拔的执行模块，根据需求选择最佳“执行层”，实现了并行能力，但这种方案能够服务于EVM，却超出了EVM框架范畴； 以Polygon、BSC等为代表的等效型Alt-layer1链，在一定程度上实现了EVM的并行处理能力，但只是通过算法层的优化，并未对深层共识和存储层进行优化，因此这种并行能力更多是特定特性，而并未完全解决EVM的并行问题。 Aptos、Sui、Fuel等为代表的非EVM并行链，在一定程度上并非实现的是EVM链，而是在其天生具有高并发执行能力的基础上，通过中间件或编码解析方式，实现了与EVM环境的兼容。我们可以看到，像以太坊layer2的Starknet就是这种方式，虽然Starknet具备Cario语言和账户抽象，具备并行能力，但兼容EVM需要特殊的通道。这些非EVM链的并行能力接轨EVM链时大多存在这个问题。 以上四个方案各有侧重点，有并行能力的layer2注重模块化组合“执行层”链的灵活性；而EVM-Compatible链突出了特定功能的定制特性；至于其他非EVM链的EVM兼容特性更多致力于抽取以太坊流动性；目前只有强化型EVM layer1链是彻底巩固EVM生态、从根本上改变并行能力的最后一条赛道了。 那么，做强化型并行EVM layer1公链的关键是什么？如何重塑EVM链且服务于EVM生态呢？有两个关键点： 1、意义重大的状态I/O磁盘读取和输出能力。由于数据读取和写入需要时间，简单的交易排序和调度并不能根本提升并行处理能力，需要引入缓存、数据切片甚至分布式存储技术等，从根本状态存储和读取流程上平衡读取速度和状态冲突的可能性； 2、高效的网络通信、数据同步、算法优化、虚拟机强化以及将计算与IO任务分离等共识机制层各方面的组件优化，需要从底层组件架构、协作流程等方面综合优化和提升，最终实现高响应速度、可控的计算消耗、高准确度的并行处理交易能力； 具体到并行EVM layer1链项目本身，需要做哪些技术创新和框架优化来实现“并行EVM”呢？ 为了从底层架构层彻底实现资源协调和优化的“并行EVM”能力，Artela引入了弹性计算（Elastic Computing）和弹性区块空间（Elastic Block Space）。这两者的理解是什么呢？弹性计算使网络能够根据需求和负载动态分配和调整计算资源；而弹性区块空间则可以根据网络中交易数量和数据大小动态调整区块大小。整个弹性设计原理就像商场自动感应人流量进行扶梯工作一样，非常合理。 前文提到，State I/O磁盘读取性能对并行EVM至关重要，而像Polygon、BSC等EVM-Compatible链通过算法实现的“并行”能力可以实现2-4倍的效率提升，但这只是算法层的优化，共识层和存储层并未进行深度优化，那么真正的深度优化会是怎样呢？ 针对这一点，Artela借鉴了数据库技术解决方案，在状态读取和写入方面都做了优化。其中，在写入状态方面，采用了写入前日志（WAL）技术，实现在状态改变时先记录改变并提交到内存，避免立即进行磁盘写入操作，从而降低磁盘I/O操作。而在状态读取方面，则通过预加载策略提高读取效率，根据合约历史执行记录预测下一次合约调用可能需要的状态，并预先加载到内存中，提升磁盘I/O请求效率。 总的来说，这种通过内存空间换取执行时间的算法可从根本上提升EVM虚拟机的并行处理能力，从根本上解决了状态冲突问题。 此外，Artela通过引入Aspect模块化编程能力分析，以更好地管理复杂性并提高开发效率。它引入了WASM编码解析来增强编程灵活性，并具有底层API访问权限，实现了执行层的安全隔离。这使得开发者可以在Artela环境下高效地开发、调试和部署智能合约，激发开发者群体的定制能力。特别是，开发者会受到鼓励在智能合约代码层面朝着可并行化方向进行代码优化，以减少状态冲突的可能性，每个智能合约的调用逻辑和算法都至关重要。 总的来看，“并行EVM”实质上是在优化交易状态的执行过程，@monad_xyz号称每秒可处理10,000笔交易，其技术核心无非是通过专用数据库、开发者友好性、延迟执行共识、超标量流水线技术等方面来实现大规模交易的并行处理，这与Artela的弹性计算和I/O异步操作在本质逻辑上并无太大差异。 然而，我更愿意表达的是，这些高性能并行EVM链实质上是融合了web2产品和技术力量的成果，确实采用了web2应用市场上高负载下的“技术处理”精髓。 在一个遥远未来的Mass Adoption时代，实现“并行EVM”的确是EVM生态走向web2更广泛市场的基础infra，受到资本市场如此看好也在情理之中。

Recently, a huge round of financing of billions of dollars has attracted the market's strong attention to parallelism. So what problems has parallelism solved? What are the bottlenecks and key points of developing parallelism? In my opinion, parallelism is the last wave of the chain to meet the high-performance chain, which is related to the survival of the Ethereum ecology. Why? Next, let's talk about my understanding. Because Ethereum virtual machines can only process transactions in series, this makes the chain and compatible chains subject to performance constraints because they are all based on the same framework. Managing state and transaction, however, high-performance chains such as equality are inherently parallel. In this context, if you want to confront the challenge chain of high-performance chains, you must improve the parallel processing ability. So how to do it? I will take the new parallel chain as an example to explain the enhanced chains represented by equality. They have significantly improved and given the parallel ability of transactions in a quasi-environment on the basis of high compatibility. These independent parallel chains have independent consensus mechanisms and technical characteristics. Sex, but still aiming at compatibility and expansion of ecology, is equivalent to reshaping the chain in the way of changing bloodlines to serve the ecology. The expanded compatible chain represented by equality uses the independent consensus and transaction preprocessing ability of the chain to screen and process the transaction state before the batch transaction is wound up, and can choose the execution layer of any other chain to confirm the transaction state. This method abstracts the pluggable execution module and selects the best execution layer according to the demand, which realizes the parallel ability, but this scheme can serve but exceeds it. In this paper, the equivalent chain represented by equality in the framework category has realized the parallel processing ability to a certain extent, but it has not optimized the deep consensus and storage layer only through the optimization of the algorithm layer, so this parallel ability is more a parallel problem with specific characteristics and has not been completely solved. The non-parallel chain represented by it is not a chain to a certain extent, but is compatible with the environment through middleware or code analysis on the basis of its inherent high concurrent execution ability. We can see that This is the way to go to a place like Ethereum. Although it has language and account abstraction and parallelism, compatibility requires special channels. Most of these non-chain parallelism have this problem when connecting the chains. Each of the above four schemes has its own emphasis and parallelism, and pays attention to the flexibility of modular combination execution layer chain, while the chain highlights the customization characteristics of specific functions. As for the compatibility characteristics of other non-chains, it is more committed to extracting the liquidity of Ethereum. At present, only the enhanced chain is to completely consolidate the ecology and fundamentally change. The last track of parallel ability, so what is the key to make an enhanced parallel public chain? How to reshape the chain and serve the ecology? There are two key points: the reading and output ability of state disk is of great significance. Because data reading and writing take time, simple transaction sorting and scheduling can not fundamentally improve the parallel processing ability. It is necessary to introduce cache data slicing and even distributed storage technology to balance the reading speed and the possibility of state conflict from the fundamental state storage and reading process, and an efficient network. Communication data synchronization algorithm, optimization of virtual machine, strengthening of computing and separation of tasks, etc. Component optimization in all aspects of the consensus mechanism layer needs to be comprehensively optimized and upgraded from the aspects of the underlying component architecture, collaboration process, etc., and finally achieve high response speed, controllable computing consumption and high-accuracy parallel processing transaction ability. Specifically, what technical innovations and framework optimization are needed for the parallel chain project itself to achieve parallelism? In order to completely realize the parallel ability of resource coordination and optimization from the underlying architecture layer, an elastic meter is introduced. What are the understandings of computing and elastic block space? Elastic computing enables the network to dynamically allocate and adjust computing resources according to the demand and load, while elastic block space can dynamically adjust the block size according to the number of transactions and the size of data in the network. The whole elastic design principle is as reasonable as a shopping mall automatically sensing people's flow for escalator work. As mentioned above, disk reading performance is very important for parallelism, and the parallel ability realized by equal chain algorithm can achieve twice the efficiency improvement, but This is only the optimization of algorithm layer, consensus layer and storage layer. What will be the real depth optimization? In view of this, the database technology solution has been optimized in both state reading and writing. In the writing state, the log before writing technology is adopted to record the changes and submit them to memory when the state changes, so as to avoid immediate disk writing operations, thus reducing the disk operation. In the state reading, the reading efficiency is improved by preloading strategy. Predict the state that may be needed for the next contract call according to the contract history execution record and pre-load it into memory to improve the efficiency of disk request. Generally speaking, this algorithm of exchanging execution time for memory space can fundamentally improve the parallel processing ability of virtual machines and fundamentally solve the problem of state conflict. In addition, by introducing modular programming ability analysis to better manage complexity and improve development efficiency, it introduces coding analysis to enhance programming flexibility and has the realization of bottom-level access rights. In order to ensure the security isolation of the execution layer, developers can efficiently develop, debug and deploy smart contracts in the environment, and stimulate the customization ability of developers. In particular, developers will be encouraged to optimize the code in the direction of parallelization at the code level of smart contracts to reduce the possibility of state conflicts. The calling logic and algorithm of each smart contract are very important. Generally speaking, parallelism is essentially optimizing the execution process of transaction state, which is called that transactions can be processed every second. Its technical core is nothing more than The parallel processing of large-scale transactions is realized by delaying the implementation of consensus superscalar pipeline technology by the developer of special database, which is not much different from the elastic computing and asynchronous operation in essence. However, I prefer to express that these high-performance parallel chains are essentially the results of the integration of products and technical forces, and the essence of technical processing under high load in the application market is indeed adopted. In a distant future era, it is indeed the basis for the ecology to move towards a wider market, and it is reasonable to be so optimistic about the capital market. 比特币今日价格行情网_okx交易所app_永续合约_比特币怎么买卖交易_虚拟币交易所平台

注册有任何问题请添加 微信：MVIP619 拉你进入群

文章转载注明 网址：https://netpsp.com/?id=53249