Browser Performance Comparison

Wednesday March 19th, 2003

Raphael Wegmann writes: "I've found a browser benchmark comparison (IE 5.5, Mozilla 1.3b and Opera 7.01) that might be interesting to the community."

#1 Nothing

by willll

Wednesday March 19th, 2003 4:56 PM

Wohoo! Mozilla can do nothing fatest!

#2 Re: Nothing

by AlexBishop

Wednesday March 19th, 2003 5:02 PM

"Wohoo! Mozilla can do nothing fatest!"

It can concatenate strings in no time at all! (Literally)


#3 Cool! But where's the rest of it?

by flacco

Wednesday March 19th, 2003 5:09 PM

You know, the performance results on non-Windows platforms?

#15 Other platforms

by leafdigital

Thursday March 20th, 2003 2:51 AM

Well, there's not much point comparing on Unix - does anybody seriously choose a different browser on Unix? (If so, you're on crack, and Javascript performance results certainly won't affect your decision. :>)

On OS X I suppose you could usefully compare browsers, but I would assume that whoever did the tests doesn't have a Mac.

I do wonder why an outdated version of IE was used for tests, though. I mean the Mozilla version was probably current when the tests were run, and same for Opera.


#16 outdated versions

by warpozio

Thursday March 20th, 2003 3:28 AM

At the moment all these browsers are outdated. Re-run the test with IE 6, Mozilla 1.4(.1)? and Opera 6.03

#17 Re: outdated versions

by pepejeria

Thursday March 20th, 2003 5:18 AM

Re-run with Opera 6.03? Are you kidding? You mean 7.03 I guess, but there is no performance difference at all comparing Opera 7.00 - 7.03. Maybe Mozilla though.

#28 Agreed

by rothkj1022

Friday March 21st, 2003 9:32 AM

I agree that the browsers tested are out of date. But what about the Operating System? NT4? Gimme a break. I suggest re-running the tests with an up-to-date OS as well.

#30 yea thats pretty disapointing...

by ph1nn

Saturday March 22nd, 2003 3:37 PM

I mean i always knew IE OWNED every other browser in the initial page load time because of the way it incorrectly sends packets to the server without handshacking or whatever. But overall the results are pretty disapointing... almost makes me wish there was IE for linux :-P

I SERIOUSLY think Mozilla should switch from adding new features that people could care less about (bloat) and work on the small, lighter weight, FAST browser i thought it was supposed to be.

I havent encountered a Mozilla crash since like 1.0, 1.3 is great... WORK ON PERFORMANCE and nothing else please :) i wanna have the fastest browser

#4 Really

by bandido

Wednesday March 19th, 2003 5:24 PM

Read again. Winning results are in bold. Mozilla came ahead in 7 benchmarks, Opera in 3 and IE5.5 won the rest. I wonder how IE6 would have fared. I have read that IE 6 is slower than 5.5.

#5 Re: Really

by willll

Wednesday March 19th, 2003 6:04 PM

I meant it performed fastest in the empty loop category. it does nothing the fastest

#10 Re: Really!

by schapel

Wednesday March 19th, 2003 8:15 PM

Looping isn't nothing. It's looping!

#29 Re: Re: Really!

by Tanyel

Saturday March 22nd, 2003 1:13 PM

Maybe the optimizations of Mozilla cause it to bypass loops that have no instructions to execute. That would explain the execution time.

#6 Re: Nothing

by quin

Wednesday March 19th, 2003 6:21 PM

I 've have and intel celeron 1200 mhz 256 RAM and mozilla 1.3 final release load pages faster than opera and IE6.0

I don't know how those benchmark are made. and what components on mozilla are evaluating but , those results can't seem be all the necesary benchmark to compare mozilla with other browsers, maybe there something missing, more components to evaluate and the make a more reliable beackmark.

i don't saying that those results are bad, i only saying that is necesary make a more extensive bechkmark beacuse mozilla seems faster loading pages than other browsers

#7 Re: Re: Nothing

by willll

Wednesday March 19th, 2003 6:55 PM

the benchmark is evaluating speed of JavaScript, not speed of pageload, downloads, or javascript. it is not designed to test anything else

#19 Re: Re: Re: Nothing

by coda

Thursday March 20th, 2003 10:47 AM

Are you confused? ;)

#26 Re: Re: Nothing

by TheK

Thursday March 20th, 2003 4:51 PM

at first Mozilla scales much better with CPU performance (that's because it's so fast on modern Computers and starts to become unusable below 500MHz) and second this only tests JS performance. Interesting, that the self called "fastest browser on earth" it way slower than Mozilla or MSIE in around everything. I fact this doesn't affect real-life that much I think. IMHO other values are more visible (redraw-times or obening a new window), also making the different windows more independant, would give much visible performance - it's not that funny, if rechecking a big newsgroup prevents the browser from working.

#27 Re: re: Nothing

by rajbhaskar

Friday March 21st, 2003 3:30 AM

> at first Mozilla scales much better with CPU performance (that's because it's so fast on modern Computers and starts to become > unusable below 500MHz)

Er, I run Mozilla on my secondary machine (a P166 with 80MB RAM) and it's nowhere near unusable. It's slow to load but once it loads, it's perfectly usable. As always, YMMV.


#31 Was faster before, but now I notice it slower

by pulsorock

Monday March 24th, 2003 9:00 PM

Hello, When I fist installed Mozilla 1.3 I noticed it was way faster than IE6.0 on my AMD Athlon XP 2200 1.8 Ghz, 512MB RAM. But after two weeks, I notice it slower on page loads. I un-installed it and installed it again, and it continues slower than IE. Have anyone have notice this or know anything about this?

#8 And?

by user4321

Wednesday March 19th, 2003 7:24 PM

So now we know that unless your JS/DOM app is freakin' huge and complex beyond reason, all the browsers will probably run it more than fast enough. Cool. Already knew that, but gee, I guess we all like validation.

I was kinda happy to see that IE is so speedy, though. That means all the extra code I have to write to work around all its bugs and half implemented features won't slow down my apps _too_ much...

#9 Re: And...

by schapel

Wednesday March 19th, 2003 8:13 PM

obviously you've never tried to navigate JS menus using Opera 7. It's so incredibly slow that it seems like the browser hung. Mozilla developers even acknowledge Mozilla has some performance problems with DHTML. The performance of IE should be a goal of all browsers.

#13 Re. And...

by user4321

Thursday March 20th, 2003 12:21 AM

Well, Opera 7's DOM support is nothing to write home about, that's for sure, but the last time I pointed that out I was treated to an extremely long diatribe about CSS' inventor being their CTO and Mozilla and IE sucked eggs and Opera was perfect and I was an idiot... it seemed best not to go down that road again.

As for Mozilla's performance, I don't recall saying it was perfect. However, I implied that it is good enough, and in my experience, it is. I develop apps which are cross platform to the maximum extent possible (a browse through this site should turn up mention of the one which isn't) and I use the W3 DOM whenever possible.

My Moz development machine is an old Celeron 300A overclocked to 450. My IE machine is a 900 MHz Athlon. I see no speed difference between them in actual usage. A benchmark shows IE is faster? What the hell do I care? They're both speedy enough for my clients.

What matters to me is that I spend about 10% of my coding time working in IE, and about half of my debugging time. Mozilla ain't perfect by any means, but IE is a friggin' mess, and all the benchmarks in the world won't change that. I don't give a damn how fast IE does something if it can't do it right.

I'm sure IE fanboys can come up with all kinds of reasons why it ain't IE's fault and Mozilla sucks rocks, but the reality is that what Mozilla does poorly it almost always does well enough to be workable, and what it does well, it does better than any other browser out there. The same can most certainly *not* be said of IE.

#18 Re: DHTML Performance Problems

by schapel

Thursday March 20th, 2003 6:21 AM

> As for Mozilla's performance, I don't recall saying it was perfect. > However, I implied that it is good enough, and in my experience, it is.

There are at least one hundred people who disagree with you. See here: Why would hundreds of people vote for bugs on JS and DHTML performance when it's good enough? That's all my point was. Are you saying they're all wrong?

#23 Then go fix it

by Kommet

Thursday March 20th, 2003 1:30 PM

I don't agree with your logic there. People vote on a bug to say "We would love to see this fixed" not because their votes mean the bug suddenly becomes Priority Numero Uno for ANYONE. Go read the etiquet FAQ and see what I mean.

Voting in Bugzilla is like voting in Iraq or Florida. People might care, but they are under no real obligation to do so.

How open-source works: You (the impersonal, rhetorical "you", not "schapel") want the bug fixed? You can go fix it. You want a tracking bug fixed? Fix one or one hundred of its dependant bugs. Don't bitch to me. Don't bitch to the bug owners. Just don't bitch.

By the way, it is not too smart to hold up a tracking bug as your (this is the personal "your", as in "schapel's") example. Tracking bugs only group families of issues together. It won't be closed until EVERY bug under it is closed. Voting for tracking bugs are almost like voting for Ralph Nader: a point is made, but not a good one.

And for the record, the phrase "... in my experience..." is about as subjective as they come. You then replying with "Are you saying they're all wrong?" is a total non sequitur. You are putting words into the original poster's mouth. How in the hell did "... in my experience..." come to mean "... for everyone who uses Mozilla..." in your head?

That people disagree with the original poster's SUBJECTIVE statement is not interesting or surprising. I want Mozilla to run on UNIVAC, but performance will have to come WAY up before that is feasable.

#24 LOL!

by schapel

Thursday March 20th, 2003 2:29 PM

LOL! Go read the original message again, and then try my suggestion and use Opera 7 to navigate DHTML menus. Contrary to what the OP originally said, even simple DHTML is way too slow for many users.

#32 Try my no-JS DHTML menus

by LPetrazickis

Tuesday March 25th, 2003 4:43 PM

Wow, it's lightning fast in Opera and Mozilla and yet doesn't work in MSIE. I guess real browsers aren't slow when they aren't fed bugtimised code.:)

Yes, I did borrow a lot of the code from

#14 Re: And?

by gbukovics

Thursday March 20th, 2003 12:58 AM

Huge and complex? Like, say, the front end of Mozilla?

I'm guessing that speeding up JS/DOM would lead to significant usability improvements in Mozilla.

#22 Re: And?

by wo_ow

Thursday March 20th, 2003 12:38 PM

Actually, I've written a JS/DOM app that is freakin' huge and complex beyond reason , and it turns out that Mozilla is much faster than IE, at least in the core part (try the last example on the "examples" page; but be warned that Mozilla will crash after calculationg a couple of hours if you don't stop it, namely at node 179797). Before I tweaked the code to make it faster with IE, Mozilla was more than 10 times faster, now it's only about twice as fast. I don't really know why IE is so slow here. Maybe the benchmark tests just aren't very significant because in reality you never want to sort exactly the same array 100000 times in a row. Or maybe I'm just using untypical operations that aren't tested in common tests.

#11 JS and DOM Performance problems

by schapel

Wednesday March 19th, 2003 8:30 PM

Here are some links to Bugzilla bug reports on some of Mozilla's JavaScript and DOM performance problems:

Object creation is slow:

Array creation is slow:

Concatentation of local strings is slow:

Improve overall JavaScript performance:

Improve DHTML performance:

#12 Date Creation Performance Test

by schapel

Wednesday March 19th, 2003 9:35 PM

I was about to file a new bug about Date object creation being too slow in Mozilla, but according to the tests on this page Date creation is only a little slower in recent Mozilla nightlies than in IE 5.5. Maybe the original browser benchmark comparison has a bug in the Date performance test.

#20 Re: JS and DOM Performance problems

by pbreit

Thursday March 20th, 2003 11:00 AM

It's mind-boggling that these items seem to get less attention than new features, non-browsing-related modules (mail, news, chat, compose) and superfluities (splash screens, skins).

#21 Re: Re: JS and DOM Performance problems

by bzbarsky

Thursday March 20th, 2003 11:45 AM

It's mind-boggling how some people simply cannot get the concept of people just working on things OTHER THAN THE LAYOUT ENGINE. And how said people would simply NOT BE WORKING instead of working on the layout engine if they were not doing what they are doing.

If you think JS/DOM perf should improve, improve it. Stop whining about the fact that the one person who has the guts to work on it is not working fast enough for your taste.

#25 Re: JS and DOM Performance problems

by schapel

Thursday March 20th, 2003 2:32 PM

Uh, the one instance in those bugs where I saw a lack of attention was a developer who quit working on performance bugs temporarily to go work on correctness bugs instead. Do you want a browser that's buggy and fast, or one that works correctly and is a bit slow? My experience is that people drop buggy software no matter how fast it runs!