Posts by Cheng Yao

MoE Training Best Practices on AMD GPUs

• 16 December 2025
Wen Xie , Xiaoming Peng , Xiaobo Chen , Lihuan Zhang , Liying Li , Ruibin Zhang , Cheng Yao , Kyle Zhao , Zhen Huang , Yuankai Chen , Yanyuan Qin , Mou Li , Chaojun Hou , Yao Fu , Liz Li , Andy Luo , Zhenyu Gu
English
Software tools & optimizations
AI/ML Performance Optimization

This blog covers best practices for training Mixture-of-Experts (MoE) models on AMD Instinct™ MI300/MI355-series^[a] GPUs with the ROCm ecosystem. Whether you’re new to MoE distributed architectures or optimizing trillion-parameter models, this guide will help you identify bottlenecks and maximize efficiency on AMD hardware.

Read more ...

---