5 papers
Completion-Path Credits: Multi-Resource Control for Scale-Up Fabrics
Fan Yang, Jiaqi Liu, Tao Jiang +1
Scale-up fabrics connecting GPUs and AI accelerators carry tensor transfers together with remote reads, writes, atomics, and notifications over shared target-side receiver resource…
Predict Before Replay: Joint FEC and Flight Control for Reliable Scale-Up Links
Fan Yang, Jiaqi Liu, Tao Jiang +1
Scale-up accelerator fabrics send latency-sensitive flits over serial links at hundreds of gigabits per second. Their reliability pipeline first relies on FEC, then detects residua…
CCL-D: A High-Precision Diagnostic System for Slow and Hang Anomalies in Large-Scale Model Training
Yida Gu, Fakang Wang, Jianhao Fu +17
As training scales grow, collective communication libraries (CCL) increasingly face anomalies arising from complex interactions among hardware, software, and environmental factors.…
JingZhao: A Framework for Rapid NIC Prototyping in the Domain-Specific-Network Era
Fan Yang, Zhan Wang, Ning Kang +4
The network is becoming domain-specific, which requires on-demand design of the network protocols, as well as the microarchitecture of the NIC. However, to develop such a NIC is no…
FNCC: Fast Notification Congestion Control in Data Center Networks
Jing Xu, Zhan Wang, Fan Yang +5
Congestion control plays a pivotal role in large-scale data centers, facilitating ultra-low latency, high bandwidth, and optimal utilization. Even with the deployment of data cente…