NCCL issue 1214: Cannot use P2P in Azure GPU cluster

Fields

id

sources/mgpu-nccl-1214-azure-t4-shm

URL

https://github.com/NVIDIA/nccl/issues/1214

Publisher

NVIDIA/nccl

Kind

repo

Published

2024-03-05

Accessed

2026-10-03

notes

4x Tesla T4 (Standard_NC64as_T4_v3) under AzureML MpiDistribution; records 'OSError: [Errno 28] No space left on device' and the DataLoader 'Bus error ... out of shared memory' signature from an undersized container shm.

Notes

4x Tesla T4 (Standard_NC64as_T4_v3) under AzureML MpiDistribution; records 'OSError: [Errno 28] No space left on device' and the DataLoader 'Bus error ... out of shared memory' signature from an undersized container shm.

Cross-references

This record references no other records.