DeepPiC: xPU-PIM Cluster Architecture with Adaptive Resource-Aware Task Orchestration for DeepSeek-Style MoE Inference.
Zixu Li, Manni Li, Zijian Huang, Jiayu Yang, Wending Zhao, Yinyin Lin, Chengchen Wang, Haidong Tian, Xiankui Xiong
Browse the full ASPDAC paper archive.