Incident Library
API · MCP
한국어
Organisation
Google
Google ·
google
1 case · 2024
Root cause
Hardware failure
Component
NIC, cable and switch port link
TPU
Contributing factor
–
Symptom
Throughput degradation
Training job failure
Cases
IL-0418
Machine, ICI link and optical circuit switch failures interrupted training jobs on Google's TPU v4 supercomputers
Google
· 2024
Root cause
Hardware failure
한국어