Job Description:
Design the GPGPU compute core, including streaming multiprocessors (SMs), warpschedulers, and arithmetic logic units (ALUs), to deliver high general-purposecomputing performance.ptimize core designfrparallelworkloads andcompatibility with CUDA kernels. Collaborate with the CUDA compatibility team toensure hardware support for key CUDA features. Lead performance analysis andoptimization of GPGPU cores for LLM inference and general Al workloads.
任职要求
Requirements:
7+ years of experience in GPGPU core design, with deep knowledge of SMarchitecture and parallel computing.Proficiency in RTL design (Verilog/VHDL) and GPU performance modeling.Strong understanding ofthe CUDAprogrammingmodelandGPUhardware-software interaction.Experience with high-performance GPU projects (e.g., data-center, HPC) is a plus.
