Information

This project aims to create an observability system for real-time monitoring of hardware resources using C++ for data collection, InfluxDB for storage and Grafana for data visualization. The process is done automatically using docker containers. It collects nine data variants per second as showed below.

• CPU
  • Usage
  • Temperature
  • Power
• GPU
  • Usage
  • Temperature (edge)
  • Temperature (junction)
  • Temperature (mem)
  • Power
• RAM
  • Usage

The data can se analysed through Hardware Info dashboard.

Requirements

• Package:
  • git
  • lm-sensors
  • docker
  • docker-compose

The functionality is made exclusively for personal use, taking into account the author's computer hardware. Some adjustments to the files will be required to make them work on different computers. This project is applied to the machine with the following specifications:

• Specification:
  • OS: Arch Linux
  • CPU: Ryzen 5 5600G
  • GPU: RX 6750XT
  • RAM: 2x16GB

The software tool lm-sensors is required in order to extract temperature data. It's important to have previously used sensor detection so temp_info.hpp can properly get data. By default, the command is sensors-detect where all analysis suggestions should be denied. After that just use sensors to get something like below.

• k10temp-pci-00c3
  • Adapter: PCI adapter
    • Tctl: +26.2°C
• amdgpu-pci-0300
  • Adapter: PCI adapter
    • edge: +35.0°C
    • junction: +35.0°C
    • mem: +46.0°C

Temperature data is obtained as subfeature and checked by chip prefixes, so in this case it's k10temp and amdgpu, respectively. Each file in the headers folder has paths for data extraction.

Instructions

NOTE: If directories for data extraction or the temperature list are wrong, the service ends with an error.

To extract data and monitor it in real time, simply run the containers using docker compose. First clone this repository with the following command:

git clone https://github.com/aryelsoares/Hardware-Benchmark.git

Change current directory to the same where it was cloned. After that run the following command to start it:

docker-compose up --build

After that data extraction is made automatically. You can check the dashboard with Grafana browsing 127.0.0.1:3000. Both user and password are admin.

Remember to finish access later on with:

docker-compose down

Notes

Although this project is applicable for a specific computer, its use can be adjusted for other computers by making appropriate changes to the files. Its main use is to monitor hardware usage in real-time.