Archived
1
1
Fork 0
A Python module/TUI that provides AMD GPU statistics
This repository has been archived on 2024-03-02. You can view files and clone it, but cannot push or open issues or pull requests.
Find a file
2023-08-05 22:38:55 -05:00
.github/workflows deps: reflect Textual, allow Py3.8 2023-06-03 02:00:17 -05:00
docs move doc files where RTD can hopefully find them 2023-04-25 22:09:52 -05:00
screens add graphs screen 2023-06-03 02:05:56 -05:00
src/amdgpu_stats tui: use new notifications, update less often. efficiency++ 2023-08-05 22:38:55 -05:00
.gitignore start packaging 2023-04-23 13:39:03 -05:00
.readthedocs.yaml rtd: add config 2023-04-25 22:31:45 -05:00
LICENSE.md Create LICENSE.md 2023-04-23 13:40:45 -05:00
pyproject.toml pyproject: bump to 0.1.19, raise textual dep for new Notifications 2023-08-05 22:38:47 -05:00
README.md add graphs screen 2023-06-03 02:05:56 -05:00
requirements.txt requirements: reflect added pyyaml 2023-06-03 02:02:50 -05:00
todo.md lint, +low prio/maybe finished items 2023-07-31 12:24:59 -05:00

amdgpu_stats

pylint

A Python module/TUI for AMD GPU statistics

Screenshot of the main stats table Screenshot of the 'graphing' scroll bars

Tested only on RX6000 series cards; APUs and more may be supported. Please file an issue if finding incompatibility!

Installation

pip install amdgpu-stats

To use the TUI, run amdgpu-stats in your terminal of choice. For the module, see below!

Module

Introduction:

In [1]: import amdgpu_stats.utils

In [2]: amdgpu_stats.utils.AMDGPU_CARDS
Out[2]: {'card0': '/sys/class/drm/card0/device/hwmon/hwmon9'}

In [3]: amdgpu_stats.utils.get_core_stats('card0')
Out[3]: {'sclk': 640000000, 'mclk': 1000000000, 'voltage': 0.79, 'util_pct': 65}

In [4]: amdgpu_stats.utils.get_clock('core', format_freq=True)
Out[4]: '659 MHz' 

For more information on what the module provides, please see:

Feature requests are encouraged 😀

Requirements

Only Linux is supported. Information is completely sourced from interfaces in sysfs.

It may be necessary to update the amdgpu.ppfeaturemask parameter to enable metrics.

This is assumed present for control over the elements being monitored. Untested without.

See this Arch Wiki entry for context.