/css-variables-benchmark

2023 CSS variables benchmark. A way to measure performance of vars vs. raw CSS.

Primary LanguageTypeScript

css-variables-benchmark

This is simple CSS variables benchmark generator. It generates n colored squares in two flavors, one with classic CSS and other using CSS variables.

Note: colors are reused once for better benchmarking.

Clone and:

yarn 
yarn start [number of colors] [reuse factor]

You should get two files in bundle directory.

  • dynamic.html
  • static.html

Now use your browser` tools to benchmark which runs faster.

Running web server

You can use VSC Go Live plugin to run it in browser.

Or...

You can install http-server

npm install -g http-server
cd bundle
http-server ./

Build

You can build this repo to run it faster:

yarn build
node dist/index.js -- 666 4

Results (2021-08-25)

5000 colors, reused once. Guest profile, no browser add-ons at all.

Linux 
Google Chrome Version 92.0.4515.159 (Official Build) (64-bit)

Static CSS (ms)

Try Loading Scripting Rendering Painting System Idle Total
1 33 4 329 17 110 4948 5441
2 31 3 317 12 113 4928 5404
3 28 3 351 19 117 4950 5468
4 29 4 314 14 117 4943 5421
5 31 4 332 18 112 4944 5441
SUM 152 18 1643 80 569 24713 27175

Dynamic CSS (ms)

Try Loading Scripting Rendering Painting System Idle Total
1 33 3 381 16 110 4906 5449
2 38 2 358 20 114 4962 5494
3 36 2 371 24 106 4978 5517
4 36 2 321 11 106 4934 5410
5 33 3 353 22 108 4977 5496
SUM 176 12 1784 93 544 24757 27366

Conclusion

As you can see using CSS variables on 10 000 elements is 0.7% slower than raw CSS.

This is very simple usage case. You might want to create your own more advanced benchmarks. Feel free to fork this code.