平头哥平头哥-AI芯片架构师-上海
任职要求
* Master's or Ph.D. in Electrical Engineering, Computer Science, or a related field. * A minimum of 10 years of experience in complex chip design and architecture, with a focus on AI or high-performance computing applications. * In-depth knowledge of AI algorithms and their computational and memory requirements. * Familiarity with chip design tools and methodologies, including RTL design using Verilog or VHDL. * Experience with programming languages such as C/C++ and scripting languages such as Python. * Strong analytical and problem-solving skills, with the ability to innovate and think creatively. * Good verbal and written skill for communication.
工作职责
We are seeking a highly skilled and innovative AI Chip Architect who will play a key role in the development of cutting-edge AI hardware. The ideal candidate is a visionary and a problem-solver, capable of designing complex chip architectures optimized for performance, efficiency, and scalability. In this role, you will work with software and hardware engineering groups to define state-of-the-art AI chip architecture for high-performance computing system in Data Center. Key Responsibilities: * Define the architecture for the next-generation AI chips, including high-performance computing system, hierarchically memory/cache system, and high-speed interconnects. * Collaborate with a cross-functional team of hardware engineers, software developers, and machine learning specialists to ensure designs meet the performance and power requirements of AI applications. * Propose and evaluate architectural innovations to improve throughput, latency, energy efficiency, and scalability of AI processing. * Produce thorough documentation to articulate design decisions and architectural trade-offs to stakeholders. * Participate in design reviews, providing critical feedback and insights to improve chip quality and performance. * Oversee and contribute to the entire lifecycle of the chip design process, from specification to production and post-production support. * Mentor junior engineers and contribute to a culture of technical excellence.
1、负责AI训练芯片架构的探索与设计; 2、负责AI训练系统的软硬件协同设计; 3、负责不同业务场景下AI模型训练流程中的关键任务和典型算子分析; 4、负责芯片架构的systemc建模。
1. 承接产品和市场需求,分解到芯片产品定义规格书,并进行相关系统架构设计,输出SOC上层和详细系统架构(包括计算系统,控制系统,内存系统,总线,互联接口,电源控制,功耗/面积/封装约束等) 2. 负责芯片SOC层面的架构/微架构设计,包括时钟、电源域、互联接口、总线优化、MMU,片上内存包括系统缓冲和一致性等设计,性能/功耗/面积优化和评估,相关领域的竞品分析,竞争力分析等 3. 负责芯片的封装选型和路标策略(比如chiplet方案) 4. 负责三方SOC层面的IP评估和选型 5. 联合相关团队,构建SOC层面的性能和功耗仿真和验证模型,支持芯片设计和验证
1. 承接产品和市场需求,分解到芯片产品定义规格书,并进行相关系统设计,输出AI计算子系统上层和详细系统架构(包括AI算力和算子支持,片上内存和缓存结构,相关编程生态环境和编译器支持,目标AI网络支持的优化支持和相应数据流等) 2. 负责AI计算子系统以及复杂模块的架构/微架构设计,包括和编译器的联合设计,性能/功耗/面积优化和评估,相关领域的竞品分析,竞争力分析等 3. 负责三方相关IP评估和选型 4. 联合相关团队,构建AI计算子系统性能和功耗仿真和验证模型,支持芯片设计和验证
1. 从功耗的维度参与软硬件协同策略和芯片总体架构方案设计 2. 负责芯片功耗场景的分析和定义,完成芯片总体功耗目标的制定,并将总体目标分解至各个子系统 3. 设计SoC低功耗架构,包括供电架构(Voltage domain, Power rail和Power domain)和Power Sequence,从PMIC到IP级的电源网络方案,Thermal方案,DVFS方案,Clock domain,以及不同层次的Clock gating方案等 4. 定义SOC和每个子系统的Power state,唤醒方案,以及相应功耗指标 5. 构建SoC层面的性能和功耗仿真平台和验证模型,负责芯片关键功耗场景的环境搭建、仿真验证、功耗测试及结果分析,指导优化算法和前端RTL设计 6. 根据功耗仿真与测试等结果,与软件和硬件团队合作,优化微架构和软硬件策略,提升芯片能效 7. 了解边缘AI芯片技术和性能/功耗优化技术的最新进展,探索新的性能/功耗方法学