Sage-Bionetworks/nvml-statsd

Name: nvml-statsd

Owner: Sage Bionetworks

Description: This Python project integrates the NVIDIA management library with statsd to report GPU metrics to a monitoring tool

Created: 2016-10-04 17:34:06.0

Updated: 2017-05-30 18:42:43.0

Pushed: 2016-11-01 21:51:01.0

Homepage: null

Size: 20

Language: Python

GitHub Committers

User	Most Recent Commit	# Commits

Other Committers

User	Email	Most Recent Commit	# Commits

README

nvml-statsd

This Python project integrates the NVIDIA management library with statsd to report GPU metrics to a monitoring tool.
Example run command:

er run -d \
tatsd_host=localhost \
vice /dev/nvidia0:/dev/nvidia0 \
vice /dev/nvidia1:/dev/nvidia1 \
vice /dev/nvidia2:/dev/nvidia2 \
vice /dev/nvidia3:/dev/nvidia3 \
vice /dev/nvidiactl:/dev/nvidiactl \
vice /dev/nvidia-uvm:/dev/nvidia-uvm \
usr/lib64/nvidia/libnvidia-ml.so.1:/usr/lib64/libnvidia-ml.so.1:ro \
usr/lib64/nvidia/libnvidia-ml.so.367.48:/usr/lib64/libnvidia-ml.so.367.48:ro \
start unless-stopped \
uset-cpus "0" \
mory 512m \
mory-swap 0m \
me nvml-statsd \
er.synapse.org/syn5644795/nvml-statsd

Note: using nvidia-docker run rather than docker-run allows you to omit the device mounts, which nvidia-docker will add for you. First, you need to start your Dockerfile with a CUDA-enabled Docker image.