Chrome and Firefox resource usage experiment

Browser WAR!

Why bother fighting over features in browsers when you can decide using some hard facts?

I had recently managed to make Firefox work for me (it still isn’t as fluid as Chrome, but is now close, thanks to using the beta version which has e10s support). Still confused about whether its really better than Chrome, I decided to put them to the test!

Results first, because.. well.

Since majority would come here for the results, well…

drum roll

Firefox wins hands down here. Personally, this was not a surprise.

Chrome was always know to be a RAM hog, but it also seems to have a huge lot more power usage as compared to Firefox as well.

Update I’m back to Chrome these days. I noticed it works much more smoother, and hangs much less often.

RAM

Not a new thing, but yes, Firefox uses lower RAM than Chrome consistently. For description of the actions during the course of the benchmark, look at the next section.

Don’t laugh at the plot :P

Also, the higher one is Chrome, although it’s not apparent.

Bad gnuplot skills ^

Power

I measured the power output of my laptop’s battery while doing various actions on both browsers, and Firefox again demonstrates much lower usage. Really good for people who keep running out of battery and complain they cannot figure out the reason (Yes, I’m also talking about myself). Firefox, good job there!

The experiment

I designed a casual benchmark for testing the browsers’ resource usage.

Here’s the timeline of the benchmark:

Start Time Duration of step Action
0 seconds 25 seconds Startup of browser
25 seconds 20 seconds Open Viva La Vida video on Youtube
45 seconds 20 seconds Open Instant Crush video on Youtube in a new tab
65 seconds 20 seconds Open The Scientist video on Youtube in a new tab
85 seconds 20 seconds Open Facebook in a new tab
105 seconds 20 seconds Close all tabs but keep browser open

NOTE All videos were running simultaneously by 85 seconds. No tab was closed prior to 105 seconds.

I executed these exact actions on both browsers, taking care to start with exactly the same conditions in both experiments.

Things done to ensure a clean test environment:

  • A fresh boot in both cases.
  • No other applications were running, except the light weight window manager (XMonad) and core services.
  • Browser was not loaded any time before the start of the test.
  • The battery level (95%) and the battery usage rate (power) were same before the start in both cases.
  • All plugins (except ad-blocker) were disabled in both browsers prior to the start of the experiment.
  • The browsers used were the publicly available binaries, not custom compiled.
  • Both browsers did not use dedicated GPU for any purpose (nvidia module was not loaded).
  • Yes, both browsers had the same volume set on system as well as Youtube.

NOTE I left ad-blocker running in both the browsers since Ads would just introduce variability and delays in video starts.

Here are the links used for the experiment:

The experiments were conducted on a Lenovo Y50-70 with a 4 cell (and dismal) battery, 8 GB RAM, and an i7 4705HQ Processor.

Scripts used

Data recording

Here is the script used to record the data. Feel free to test it on yours as well.

Data format

The output data files were of the format:

<time>, <ram_used>
<time2>, <ram_used2>

Plotting script

The (below-par) script used for plotting the plots is here. It takes as argument the 2 files (like: chrome-ram and firefox-ram).