Machine crash during multi-gpu ML inference

[boniek83]

Gist below contains dmesg just before machine became unresponsive. dmesg: https://gist.github.com/boniek83/6b7bce76b531f59e8b5b98a6b7cae76a Centos 7 (3.10.0-1160.21.1.el7.x86_64) ROCK 3.5 4x Radeon VII pytorch 1.6 + torchserve