Accelerating Model Training on Ascend Chips: An Industrial System for Profiling, Analysis and Optimization

Training large-scale deep learning (DL) models is a resourceintensive and time-consuming endeavor, yet optimizing training efficiency poses significant challenges. The sporadic performance fluctuations during long training require advanced profiling capabilities. It is not easy to perform comprehensive and accurate bottleneck analysis amidst numerous influencing factors. Selecting effective optimization strategies without proper guidance further complicates the process. This paper shares our practical insights on optimizing training on Huawei Ascend chips based on three years of experience with 135 typical cases. We propose a systematic optimization system, Hermes, including a lightweight profiling approach, a hierarchical bottleneck analysis framework, and an optimization advisor. Our real-world experiments demonstrate significant acceleration in training for models like PanGu-α, MobileNetV1, and MoE (Mixture of Experts), with respective speedups of 3.05×, 1.91×, and 1.19×.