Component Benchmark：大规模推荐系统的分层模型性能分析
原标题：Component Benchmark: Hierarchical Model Profiling for Large-scale Recommendation Systems
AI 摘要
论文提出 Component Benchmark (CB)，一个面向大规模推荐系统的分层性能分析系统。CB 基于 PyTorch module hooks 捕获子模块输入，独立地对每个组件进行基准测试，并重建树状性能视图，帮助定位复杂模型架构中的性能热点。系统采用插件架构（Component Provider、Preprocessor、Profiler、Result 插件），核心仅约 1000 行代码，可在单 GPU 上运行，并在开源模型上验证了有效性。
正文节选
Component Benchmark: Hierarchical Model Profiling for Large-scale Recommendation SystemsThanks: *Equal contribution. Abstract Large-scale recommendation models pose distinct, under-explored profiling challenges. Most recommendation model architectures are structurally heterogeneous, intermixing memory-bandwidth-bound operations, small compute-bound dense layers, dynamic shapes from jagged categorical features, and low-arithmetic-intensity operations. Recommendation models evolve rapidly as model