Incident Library

Component

PCIe link

PCIe 링크 · pcie-link

2 cases · 2025

Root cause Hardware failure1unknown1
Contributing factor No alert on limit1
Symptom Throughput degradation2
Organisation ByteDance1IBM1

Cases, by severity

  1. IL-0476 PCIe downgrade on one machine slowed a 128-machine V100 training job for 40 minutes through a PFC surgeByteDance · 2025

    Root cause Hardware failure

  2. IL-0420 PCIe links on IBM's Vela GPU nodes dropped to a lower generation's speed and slowed training jobs by up to 2xIBM · 2025

    Root cause unknown

View on the graph