2025 年 7 月 16-19 日，第五届 RISC-V 中国峰会在上海张江科学会堂拉开帷幕。这场汇聚全球芯片产业目光的盛会中，知合计算以 "多场主题演讲 + 核心 产品发布" 的强势姿态登场 —— 不仅在主论坛、高性能分论坛、人工智能分论坛系统阐述技术理念，更正式公布新一代高性能 RISC-V 内核进展，并发布 "阿基米德" 系列通推一体 CPU 产品。这一系列动作，被业内视为 RISC-V 架构向高性能计算领域发起冲击的 "冲锋号"，也让中国芯片企业在全球架构竞 争中的话语权再获提升。 一、架构变革临界点：RISC-V 为何成为 AI 时代新引擎？ 全球芯片架构的演进始终与时代需求深度绑定：PC 时代 x86 以封闭生态垄断市场，移动时代 ARM 靠 "公版授权" 降低设计门槛普及，而在 AI 与万物互 联时代，RISC-V 正凭借开源、精简、可扩展的特性，成为重构产业格局的新变量。 经过十余年发展，RISC-V 已在物联网、嵌入式领域实现规模化落地，但向高性能计算场景的突破却始终面临 "瓶颈"。知合计算 CEO 孟建熠博士指 出："当前 RISC-V 阵营的第一梯队产品，性能仍落后 ARM 与 x86 ...

Core Viewpoint - The article discusses the evolution of chip architectures, highlighting the rise of RISC-V as a new engine for transformation in the global chip industry, particularly in the context of AI and IoT, due to its open-source, simplified, and cost-effective nature [1][2]. Group 1: RISC-V's Current Status and Challenges - Despite the growing interest in RISC-V for high-performance applications, there is a lack of benchmark products that can compete with ARM and x86 architectures [2][3]. - The first-tier RISC-V products, such as SiFive's P870-D and others, still lag behind mainstream ARM and x86 offerings in terms of performance [2][3]. - The absence of high-performance RISC-V chips may hinder the development of a robust software ecosystem, as developers are less inclined to invest in a less popular architecture [2][3]. Group 2: Key Developments and Innovations - Knowledge Computing has introduced the "Archimedes" series of CPUs, which aim to integrate general computing and AI inference capabilities efficiently [6][15]. - The new RISC-V CPU core is positioned among the top-tier high-performance CPUs, capable of matching the performance of leading ARM and x86 products [7][14]. - The CPU core supports a wide range of RISC-V extensions and is designed to provide a complete software stack, including tools and operating systems, to facilitate easier adoption [9][12]. Group 3: Market Strategy and Ecosystem Development - Knowledge Computing is focusing on specific application scenarios, such as edge servers and AI model inference, to drive the adoption of RISC-V and build a more robust ecosystem [21][22]. - The company aims to establish benchmark products that can attract more developers and partners to participate in the RISC-V ecosystem, thereby enhancing market confidence [28][26]. - The strategy emphasizes collaboration over competition among domestic manufacturers to strengthen RISC-V's position in the global market and enhance its standard-setting capabilities [34][35].