Both in response to my recent review of Midori and some argument… ehm, discussion, in yesterdays Ubuntuesday OMG! Podcast, I have taken on the task of running Midori and Google Chrome through a few tests to see which one performs better and faster in some basic web tasks. While, when I wrote my original review, I did a few ‘eye ball’ guessing tests, those were slightly less than… measurable. So, on to the tests! I’m running a Dell Studio XPS 13″, with a 2.66 Ghz Core 2 Duo, 4 gigs of ram, and off of a wireless internet connection.
I tried this test a few different ways. First, I set both Midori and Chrome to OMG!s homepage, and (with both set to resume session) opened them both by opening desktop shortcuts. Midori tended to open in 2 seconds or less, Chrome just a hair behind, but Chrome would load OMG! much quicker, usually within about 6 seconds, with Midori at about 12. Amusingly enough, however, Chrome crashed after my 5 try, and gave me one of those “Could not load your Profile” Errors. Awesome. Also amusing was the time that Midori loaded OMG! all the way before Chrome even showed up to the party. I then ran the same test, but with blank tabs. I thought at first it was a sure thing that Chrome was faster, but after the first two, Midori dominated, primarily because it was consistent. It loaded in less than 3 seconds every single time, usually more like 2, and never more. Chrome, while sometimes nearly instantaneous, was also sometimes 5, 6, 7 seconds, or never showed up at all and crashed. So? Consistency over sometimes just slightly faster, in my opinion.
In case you didn’t notice, Midori did better.
Chrome, while getting a 100/100, fails to properly render objects(Edit: As commenters have pointed out, Chrome also gets perfect marks, and Chromes failings on my end are a mix of some tests with extensions, and some with remnants of previous installs of other versions of Chrome), while Midori does fantastically, literally perfectly rendering the Acid3 test. Its pretty fantastic to see that Midori has succeeded where Chrome has not too :D. As a comparison? Heres Acid 2:
Looks pretty good, right? Actually:
Chrome is totally f’d up on Acid2 as well, scrolling and breaking the image, while again, Midori does perfect. As far as complex rendering of webpages, Midori definitely does better, and by my eye, faster.
One thing that Chrome definitely has over Midori is the full sandboxing of tasks. However, with just one tab open, Midori sat right at 23.4 MBs. Chrome, meanwhile, had 4 tasks open (obviously with all extensions disabled) at 19.6 mb, 9.4 mb, 600 kb, and 43.8 mb. Fairly obviously, Midori waaaaaaaaay wins out as far as lightness. However, Chrome does have, (as I’ve experienced and don’t know exactly how to replicate in a test) a better handle on memory leak, keeping things from getting out of hand.
Honestly, its not speed that keeps me from using Midori as my primary browser. Its fast, so fast. But I can’t, for the life of me, get my bookmarks imported, and theres a few other little design interface issues (discussed in my review) that I’d like to see get addressed before it moves to slot number one.
However, you know what would make Meerkat a perfect 10 out of 10 for me? Midori as default browser. If the devs can work out a few kinks (and I know they are, because they’ve already worked on some in response to my review) then I think Ubuntu can have a real winner.
EDIT! – DISCREPANCIES! It would appear that something odd is going on with my install of Chrome, as I’m getting odd results. I’ve run the tests with all extensions disabled, but there are still problems. I’ll be back when I figure out what the crap is altering some of Chromes scores. Thank you, watchful commenters.