Web browser showdown May 2019: Who's the fastest, Chromium vs Firefox
Which web browser is the fastest and snappiest on GNU/Linux? Here's some extensive testing of Firefox and Chromium both with out of the box settings and performance-tuned settings for better performance.
All tests were done at 1440p on a Ryzen 2600 with a RX 570 GPU running Fedora 30.
Getting maximum performance out of Chromium and Firefox
Graphics performance of both Firefox and Chromium is bad out-of-the-box on all the GNU/Linux distributions. Thus, some changes are required to get the most out of these browsers.
When you start Firefox on all the GNU/Linux distributions you get "Basic" rendering. You can check what is used by typing in the special url
about:support in the URL bar. This takes you to a special status-page. Look for Compositing under Graphics. This can be changed to two better types of rendering: OpenGL and Webrender. OpenGL is better than Basic. Mozilla is moving away from OpenGL to shiny new Webrender.
If you want to use OpenGL you need to go to
about:support (type it into the address bar) and set
true. To get the full benefit you will also need to right-click and select New->Boolean and name it
gfx.canvas.azure.accelerated. Now set this to true. Restart the browser and go to
about:support to see that you are using the OpenGL renderer.
If you want to use Webrender then all you have to do is to go to
about:support and set
gfx.webrender.all to true. You do not need to change any OpenGL settings to use Webrender, they will simply be ignored. Go to
about:support and look at the Graphics section to verify that Webrender is now in use.
Check the special Chromium url
chrome://gpu to see what graphics features you have enabled then go to
chrome://flags. You will want to change the following to Enabled:
- Override software rendering list
- GPU rasterization
and that's it.
Chromium wins this test. Jetstream doesn't tested so the results are the same regardless of GPU settings. The results do differ and they differ a lot in the MotionMark synthetic test of graphical performance:
Chromium wins over Firefox out of the box by a large margin in MotionMark 1.1. Chromium's default settings only provide two thirds of potential performance - that's a pretty big difference between two settings being disabled or enabled.
Firefox just miserably fails this test. A score of 30 compared to chromium's 207 out of the box is just pathetic. Further, more than doubling that stock configuration score by enabling Webrender doesn't help: Firefox is still miles behind Chromium in the graphics department. This is just pathetic.
So how do these scores translate to real-work usage?
Real-world like benchmarks
Basemark's Web and WebXPRT3 from Intel's "independent" front company Principledtechnologies give a better idea as to how browsers perform than the more synthetic benchmarks.
Firefox wins the WebXPRT3 by a small margin. It's interesting to note that this test doesn't have any graphical elements so graphics settings don't come into play. This gives Firefox a win in it's simulated performance.
Basemark does test graphical performance and it weighs heavily in on the result. Thus, Chromium wins this test by a large margin with stock settings and an even larger margin with tuned settings. Firefox does better with Webrender than it does with stock settings but it still falls way behind.