1. In one line, how would you explain to a non-technical user why they should use Artela versus Monad (highest throughput/best user experience) and MegaETH (Monad, but as an L2)?

   ---

   Artela is more than Monad plus Arbitrum Stylus.

2. Please compare the performance of Artela vs Monad and MegaETH, in metrics such as TPS, to the best of your ability

   ---

   Before conducting the performance comparison, we would like to share our two perspectives on L1/L2 performance.

   Perspective 1: All performance metrics are the result of trade-offs

   • Monad 强调通过低硬件要求实现最高性能，以实现去中心化，任何人都能运行全节点的高性能EVM L1，1w tps可能是这个硬件设定的极限
   • Artela与Solana、SUI等高性能网络的设定类似，通过高配节点实现超级性能和扩展性，解锁面向消费者的场景，代价是提高了验证者网络的参与门槛
   • MegaETH发挥中心化sequencer独特的架构优势，实现极致性能的乐观rollup。它的全节点需要有1TB级别内存的高端服务器，使参与验证的门槛巨高。

   Perspective 2: Technical implementations under different trade-offs cannot be simply substituted for one another

   • MegaETH的高性能秘密是使用1TB的内存加载所有世界状态，抛弃运行时对磁盘的依赖，这在L1节点里是不可能出现的设定，这种技术无法被L1采用
   • Artela在Cosmos SDK堆栈上使用模块化的方式实现高性能EVM，与自研堆栈的Monad之间诸多技术无法简单互相替换

   Artela的性能分两个目标：

   • 作为L1，将构建支持1B账户，50k TPS，可预测性能，功能可扩展的高性能公链
   • 作为EVM++执行层，它可以运行在Celestia上的基于Rollkit的高性能、可定制的sovereign rollup，它的目标是120kTPS

   与Monad和megaETH的比较：

   | | Artela L1 | Monad | Artela EVM++ Execution Layer | megaETH | | --- | --- | --- | --- | --- | | TPS | 50k | 10k | 120k | 100k | | Node Requirement | High-spec | Medium-spec | High-spec | Ultra-high-spec | | Account Capacity | 1B | unknown | 1B | unknown | | DB IO Throughput compared to geth | 20X | unknown | 20X | unknown | | DB IO Latency compared to geth | 10X | unknown | 10X | unknown | | DB Performance Degradation | No | unknown | No | unknown |

   细节解释：

   • Why does Artela emphasize 1B account capacity?

     MPT + LSM-based NoSQL DB的存储解决方案最大的问题是写放大。当数据库有100M个活跃账户，每秒持续超过100k的交易时，基于MPT + NoSQL的区块链DB的Throughput和Latency会急剧下降，导致L1的整体Performance急剧下降。

     要使100k的tps能实际发挥上作用，支持大规模账户下性能不衰减、数据可裁剪的高性能DB是更加关键的系统指标。

   • Why 1B&50k for L1？

     基于Artela团队过去支撑支付宝，作为世界最大的互联网支付基础设施，关键系统满足这两个指标，已经可以支撑起亿级消费者日常使用的场景。

   • Metrics details

     支持Artela高性能的核心技术，我们曾在蚂蚁链构建过（企业级区块链，商业用途非开源），做到：

     • 支持十亿账户规模，TPS 超过 12 万，交易平均时延低于 150ms
     • 测试环境：16个共识节点，通过专有网络连接
     • 节点配置：由于并行化和计算存储分离，每个node是cluster架构，每个node，由多台云服务器组成
     • 12万TPS的核心技术是并行执行，和计算存储分离的存储引擎：
       • a）存储层 IO 吞吐相比以太坊 MPT + LevelDB 等架构提升 10~20 倍
       • b）IO 延迟降低 90% 以上
       • c）在 7x24 高压力压测中，端到端 TPS 不随数据量增加而衰减
       • d）磁盘带宽减少 95%、空间占用减少 60%

   • Why can't 120k L1 TPS be rebuilt?

     • Antchain是在专网环境下构建，去中心化网络里共识算法成为瓶颈，无法到达这个数据
     • 在Cosmos SDK堆栈上构建，有其他技术约束，如p2p网络、iavl树的改进等问题

There’s some perspectives we would like to promote before doing the comparision:

Perspective 1: All performance metrics result from trade-offs.

• Monad emphasizes decentralization with low hardware requirements for EVM L1, achieving 10k TPS.
• Artela, like Solana and SUI, uses high-performance nodes for scalability, targeting large consumer scale scenarios but at the cost of increasing validator entry barriers.
• MegaETH leverages centralized sequencers for extreme performance in optimistic rollups, requiring 1TB RAM servers, raising validator participation costs.

Perspective 2: Different trade-offs require unique technical approaches

• The secret to MegaETH’s high performance is using 1TB of memory to load all world states, eliminating runtime dependency on disk. Such a setup is impossible for L1 nodes, and this technology cannot be adopted by L1.
• Artela implements a high-performance EVM using a modular approach on the Cosmos SDK stack, and cannot simply replace its components with those from Monad L1.

Artela aims for 50k TPS as L1 and 120k TPS as EVM++ rollup.

• As a Layer 1 blockchain, it aims to develop a high-performance blockchain that supports one billion accounts, processes 50,000 TPS, ensures predictable performance, and offers extensible functionality.
• As an enhanced EVM++ execution layer, Artela can operate as a high-performance, customizable sovereign rollup built on Rollkit and running on Celestia, targeting 120,000 TPS.