Lume #
Lume is a simple library to display current usage of resurces with LEDs only. It is designed to mimick the look of CM-1 and CM-2 machines from the Thinking Machines Corporation. The overall goal of the project is to achieve a pleasing and modern look for the visualization while preserving functionality.
Currently supports CPUs and NVIDIA GPUs.
Usage #
CPU #
To visualize cpu load for the current machine in TUI run
cargo run cpu
Workload is visualized in columns, one column per cpu core. Number of lit lights in the column visualize how much core is occupied (all lit - full load, no lights - no load). Visualization was designed to have esthetic look while allowing to estimate resource in an eyes glance.
Very high load - all 32 core are working
Medium load - roughy half of cores are working, some are utilized in only some percent
Low load - only few cores are utilized
Combination #
There is a way to customize display with cpu and gpu monitoring. In order to combine various monitoring tools one can use yaml configurations files.
modules:
- name: cpu
region: { start: 0, end: 8}
config:
simple: false
reduce: 4
- name: nvidia
region: { start: 8, end: 9 }
config:
devices: [0,]
measure_type: Util
- name: nvidia
region: { start: 9, end: 10 }
config:
devices: [0,]
measure_type: Memory
step: 2
This will display in first 8 columns cpu utilization. Keep in mind to scale everything to display in 8 columns! In this example machine is 32 core, after reducing by a factor of 4 we get only 8 cores as requested. Last two columns will be dedicated to Nvidia gpu monitoring (only one device, utilization and memory separatelly). To run the config one can use
cargo run combined --path path/to/file.yaml