gpuview

GPU is an expensive resource, and deep learning practitioners have to monitor the health and usage of their GPUs, such as the temperature, memory, utilization, and the users. This can be done with tools like nvidia-smi and gpustat from the terminal or command-line. Often times, however, it is not convenient to ssh into servers to just check the GPU status. gpuview is meant to mitigate this by running a lightweight web dashboard on top of gpustat.

With gpuview one can monitor GPUs on the go, though a web browser. Moreover, multiple GPU servers can be registered into one gpuview dashboard and all stats are aggregated and accessible from one place.

Thumbnail view of GPUs across multiple servers.

Setup

Python is required,gpuview has been tested with both 2.7 and 3 versions.

Install from PyPI:

$ pip install gpuview

[or] Install directly from repo:

$ pip install git+https://github.com/fgaim/[email protected]

gpuview installs the latest version of gpustat from pypi, therefore, its commands are available from the terminal.

Usage

gpuview can be used in two modes as a temporary process or as a background service.

Run gpuview

Once gpuview is installed, it can be started as follows:

$ gpuview run --safe-zone

This will start the dasboard at http://0.0.0.0:9988.

By default, gpuview runs at 0.0.0.0 and port 9988, but these can be changed using --host and --port. The safe-zone option means report all detials including usernames, but it can be turned off for security reasons.

Run as a Service

To permanently run gpuview it needs to be deployed as a background service. This will require a sudo privilege authentication. The following command needs to be executed only once:

$ gpuview service [--safe-zone] [--exlude-self]

If successful, the gpuview service is run immediately and will also autostart at boot time. It can be controlled using supervisorctl start|stop|restart gpuview.

Runtime options

There a few important options in gpuview, use -h to see them all.

$ gpuview -h

• run : Start gpuview dashboard server
  • --host : URL or IP address of host (default: 0.0.0.0)
  • --port : Port number to listen to (default: 9988)
  • --safe-zone : Safe to report all details, eg. usernames
  • --exclude-self : Don't report to others but to self-dashboard
  • -d, --debug : Run server in debug mode (for developers)
• add : Add a GPU host to dashboard
  • --url : URL of host [IP:Port], eg. X.X.X.X:9988
  • --name : Optional readable name for the host, eg. Node101
• remove : Remove a registered host from dashboard
  • --url : URL of host to remove, eg. X.X.X.X:9988
• hosts : Print out all registered hosts
• service : Install gpuview as system service
  • --host : URL or IP address of host (default: 0.0.0.0)
  • --port : Port number to listen to (default: 9988)
  • --safe-zone : Safe to report all details, eg. usernames
  • --exclude-self : Don't report to others but to self-dashboard
• -v, --version : Print versions of gpuview and gpustat
• -h, --help : Print help for command-line options

Monitoring multiple hosts

To aggregate the stats of multiple machines, they can be registered to one dashboard using their address and the port number running gpustat.

Register a host to monitor as follows:

$ gpuview add --url <ip:port> --name <name>

Remove a registered host as follows:

$ gpuview remove --url <ip:port> --name <name>

Display all registered hosts as follows:

$ gpuview hosts

Note: the gpuview service needs to run in all hosts that will be monitored.

Tip: gpuview can be setup on a none GPU machine, such as laptops, to monitor remote GPU servers.

etc

Helpful tips related to the underlying performance are available at the gpustat repo.

For the sake of simplicity, gpuview does not have a user authentication in place. As a security measure, it does not report sensitive details such as user names by default. This can be changed if the service is running in a trusted network, using the --safe-zone option to report all details.

The --exclude-self option of the run command can be used to prevent other dashboards from getting stats of the current machine. This way the stats are shown only on the host's own dashboard.

Detailed view of GPUs across multiple servers.

License

MIT License