This repository tries to explore the question if:
- We can monitor a website over time and can see if the energy cost for rendering changes
- If different websites have different energies for rendering which is uncorrelated to the pure time it takes for rendering (aka different power draw)
That these differences do exist is already known. Parsing complex CSS structures or going through large DOM structures is more extensive than small pages.
The main question is: Can a single change of say a reduction of DOM nodes really be measured or will it just drown in the noise / standard deviation of a normal parsing flow of a browser. The culprits here are async parsing, the OS scheduler, network latency etc.
We use the Green Metrics Tool to setup a simple Playwright Headless Browser based benchmark.
There are different types of pages:
- bbc.co.uk - Media. Lots of images and tracking
- theguardian.co.uk - Media. Lots if images and tracking
- michaelkors.de - Playing video
- green-coding.io - low fi
- svgator.com - SVG Animation