Steve

Posts tagged with "Firefox"

More speed tests

Saturday, August 26, 2006 11:00:34 PM

Seeing as I now have a different computer, and Opera and Firefox have new versions, I should update my earlier post comparing them.

First of all, my "new" computer ... it used to be my brother's computer, he got it about 3 years ago. He finally decided to get internet service, but as his antivirus was 3 years out of date and he had no firewall, his computer lasted about a week before the various worms and adware caused it to grind to a halt. Being who he is, he doesn't try to fix stuff - he looked at it as a good excuse to by a new computer. Since I was still using a 6-year-old computer, he gave me his old one for free.

It's a 1.5 GHz P3 system (my old one was a 500 MHz K6-2) with 512 MB RAM (448 on my old one) and nVidia graphics card. It originally had Windows XP, but as I had to format the drive anyway I installed Linux (Mandriva 2006.0 Free). Obviously the results will be at least 3 times as fast, maybe even faster if they involve graphics.

I'll also note that the website I used in my previous test no longer logs previous results, the results listed for "Last results" are dated back in early July.

Anyway ... I also observed that settings for popup and tab handling could change the results dramatically (for test 2), so I now have several sets of results (besides different versions of various browsers).

So here's my first set of results - Opera 9.01, with various settings for tab/popup handling:

popups as windows    popups as tabs
                   foreground   background   maximized
     1.702            1.673        1.724       1.696
     4.11             5.927        4.271       1.757
     1.612            1.793        1.833       1.641
     0.704            0.726        0.725       0.673
     0.2              0.189        0.205       0.213
     4.944            4.879        4.96        4.721
     0.844            0.846        0.899       0.855

There is considerable variability in the tests, even if you rerun the same test; I think we can treat all the results except test 2 as the same. Note that using "Open windows instead of tabs" (referred to as SDI mode by us experienced users) is actually somewhat faster here than opening them as tabs - I was surprised by that. Though setting New tabs as "Maximize all, including popups" was the real surprise. (Note that in this test the content of the popups is injected by the script, there is no loading time for any of them.)

Okay, next set of tests ... Firefox has just released 2.0 Beta 1. I had 2.0 Alpha 3 and 1.5.0.6 installed originally, I added the new beta for completeness:

   popups as windows               popups as tabs
1.5.0.6  2.0 a3   2.0 b1      1.5.0.6  2.0 a3   2.0 b1
  7.11    7.021    6.582        7.281   7.192    6.264
 19.629   7.893    7.88         4.652   2.543    3.044
  3.415   3.019    3.083        3.464   3.097    2.95
  5.481   1.384    2.153        5.668   1.36     2.122
  0.582   0.566    0.53         0.619   0.561    0.518
 13.74    9.45     9.282       13.83    9.435    8.915
  1.883   3.123    2.848        1.874   3.118    2.801

Okay, Firefox 2 is better overall than 1.5, and markedly better at opening new windows and at the layers test (where it assembles a sentence out of floating letters). Oh, and the image swapping test. For some reason, the new version is quite a bit worse at date calculations (test 7 computes day of the week for the next 10,000 Christmases). There actually seems to be (at least, as far as I can tell given the variability of the test results) some improvement from Alpha 3 to Beta 1 in the loop test (Test 1, which shows that progress bar as it counts to 10,000) and also in the date calculations, but they actually lost ground on the image swapping test.

Note if you're trying to compare this with the earlier results that Firefox tabs are always maximized, there is no way to display a popup as a floating child window like there is in Opera. So the first 3 columns compare to Opera's first column, the other 3 columns should be compared to Opera's fourth column.

So how's Firefox 2.0 look? If you're a Firefox user, it looks like it is going to be good. Of course none of your extensions are available for the new beta yet, best wait a little while for the extension authors to update them. I don't use Firefox regularly (in case you couldn't guess), I really can't say how stable the new beta is. Still nowhere near as fast as Opera at these tests, seems like Opera is 2-3 times better on pretty much all of them, but we knew that.

(Oh, of course the statements I just made imply that Firefox 2.0 is running with no extensions installed. The extensions I have shouldn't really effect javascript anyway, and the tests wait until the page is loaded so my installed extensions shouldn't really matter here.)

No comments

Next great thing?

Wednesday, November 16, 2005 8:19:39 PM

Firefox, browsers

A few days ago, I noticed that Firefox 1.5 was finally out[*]. I do like to keep up-to-date, so I installed it over RC1. Yes, I had installed Beta 2 and RC 1 also ...

I suppose I should say something about upgrading from Beta 1 to Beta 2. You hear people in the forums talk about it all the time - they call it "extension hell". When B2 came out, most of my extensions wouldn't work with it, it took about a week for most of them to be upgraded so that I could get a version that worked with the updated browser. They did better with RC1 and the final - in RC1 I only lost one extension; in the final one extension had to be updated but it updated itself.

There are a couple of things I like in Firefox, don't get me wrong. In Linux, Firefox seems a bit more responsive when you click a link or button. The text entry speed is actually too fast, if I use one of the cursor keys I have a hard time stopping it at the right point. But Opera will still be my primary Linux browser.

The most disconcerting thing for me is how slow the menus are. In Opera, if I click on the menu bar, the menu pops up very quickly (for a 500 MHz PC anyway), in Firefox there's a quite noticeable delay. And it's even worse for context menus, there were several times when I right-clicked on a page and decided nothing was going to happen, then it finally did.

Also disconcerting, I had a popup get by their popup blocker. Opera (9.0 PR1 anyway) blocked the popup no trouble, there it was in Firefox.

Also, of course Firefox is not an MDI program. You only have two choices when it comes to desired popups - they open in a full-size tab, or they open in a separate window. Of course, you also can't "Restore" your tabs, they always occupy the full browser (excluding space for toolbars and panels). I don't like that ... though you can open a link in the panel without bookmarking it, something Opera may want to look into.

I was curious how they did on that javascript speed test BenchJS. I've previously compared Opera 7.x to some of the early versions of Firefox (search the forums if you want to see that). Keep in mind, this is on a 500 MHz PC running Mandrake Linux 10.0 and with no video acceleration to speak of. Opera on Windows runs faster. Anyway, here's the results:

                    Firefox 1.5        Opera 9.0 TP1
TEST 1 time:         32.115 sec.         13.784 sec.
TEST 2 time:         94.832 sec.         28.723 sec.
TEST 3 time:         28.262 sec.         14.927 sec.
TEST 4 time:         19.307 sec.          8.429 sec.
TEST 5 time:          5.478 sec.          1.862 sec.
TEST 6 time:         56.62 sec.          49.455 sec.
TEST 7 time:         14.06 sec.           4.862 sec.
Total:              250.67 sec.         122.04 sec.

The only test where the two were anywhere near comparable was test 6 (the layers test), the worst example was the popup test where Firefox took 3.30 times as long (though I'd attribute some of that to the popups being opened outside the browser window). Note that I use KDE as my desktop, perhaps Firefox would improve some in a GTK-based environment. (Both Opera and KDE are based on Qt.) Looks like they still have a ways to go, though.

Addendum: I should add, though I've said it before in the forums - no, I don't really consider BenchJS a test of anything meaningful. It's out there, it wasn't written to show off Opera (though it does so extremely well) ... but what does creating 8 red popups mean in terms of doing anything useful with javascript? Animating about 80 letters across the screen? Cute (if it weren't so slow here), but not very practical. And as those two are the slowest tests on my system, they overwhelm the tests which might be meaningful. But it's a test which is out there, and judging by the list of results page fairly popular, so I'll use it.

[*] As is now apparent, that must have been RC2 but the auto update notification only listed it as "Firefox 1.5" and left out the rest of the identification. Needs some work, then ...

2 comments