[zen]

For my own interest I'd like to compare the relative browsing speeds of Safari, Firefox, IE, Camino etc. Is there an app that can do this, or even a website of some kind? I guess it would mean rendering speed, page loading speed etc etc.

Cheers,

Zen

[cpac]

I'd question the value of any such comparison - figure out which browser works best for you and stick with it - the speed differences are so miniscule these days as to make any speed difference inconsequential for anything except specialized tests of the sort you're seeking to perform.

[rytc]

Originally posted by cpac:
I'd question the value of any such comparison - figure out which browser works best for you and stick with it - the speed differences are so miniscule these days as to make any speed difference inconsequential for anything except specialized tests of the sort you're seeking to perform.

I'd agree with this when describing Firefox, Camino and Safari, however, IE is in a league of its own in terms of how slow and out of date the Mac version is.

[f1000]

Originally posted by zen:
For my own interest I'd like to compare the relative browsing speeds of Safari, Firefox, IE, Camino etc. Is there an app that can do this, or even a website of some kind? I guess it would mean rendering speed, page loading speed etc etc.

Ars Technica published a "Macintosh Browser Smackdown" last year,

http://arstechnica.com/reviews/003/s...rowsers-1.html

[VEGAN]

Originally posted by f1000:
Ars Technica published a "Macintosh Browser Smackdown" last year,

http://arstechnica.com/reviews/003/s...rowsers-1.html

Didn't read the article, but anyway... there have been many updates to the browsers rather reseantly, therefore, one year old comparison is a bit dated

[Mike S.]

I'm going to get vitriol saturated replies for this but since somebody asked:



These numbers are not scientific nor is it trying to be but it is consistent with every other similar test I've done over the years due to my own curiosity.

The test were done on my local LAN with archived copies of the site's front pages, four runs, two with caches full and two emptied and the results displayed are averaged. Times recorded via a stop watch and based on the browsers' various status indicators.

Because these were direct archived pages some sites connected to ad server which would have a minimal effect on results but when taken as a way to show performance relative to each other these numbers are perfectly consistent with every other test I've run just loading sites right off the web over the last few years and consistent with other results I've seen at various web sites who have posted comparisons.

If you don't like the numbers then don't bother to complain about how my testing sucks for reason xyz, I don't care. Everybody likes to say "this browser is fastest!" but nobody has bothered to even try to back it up with any kind of testing. Perceptions are not fact, these times are even if they are not statistically or scientifically perfect.

That'll all I'll say on the matter, take them how you will.

[cpac]

Originally posted by Mike S.:
I'm going to get vitriol saturated replies for this but since somebody asked:
...
That'll all I'll say on the matter, take them how you will.

will do - as you seem to concede, these number don't mean anything.

PS - an upgraded 8600??

[Mike S.]

Originally posted by cpac:
will do - as you seem to concede, these number don't mean anything.

PS - an upgraded 8600??

Actually, I conceded that they weren't scientific but they were valid as relative indicators.

Please don't put words in my mouth.

[ooninay]

Originally posted by cpac:
will do - as you seem to concede, these number don't mean anything.

PS - an upgraded 8600??

Why don't these numbers mean anything? They seem to be rather rigorously derived load times for popular websites using different browsers and platforms. How is this information meaningless?

[Mike S.]

Originally posted by ooninay:
Why don't these numbers mean anything? They seem to be rather rigorously derived load times for popular websites using different browsers and platforms. How is this information meaningless?

You can't take the numbers as absolute values, in other words just because it took 3 seconds to load a site on my eMac doesn't mean that browser will always load that page in 3 seconds on every piece of hardware or even another eMac of similar configuration.

Many people will dismiss numbers that aren't done under heavily controlled situations and other people will dismiss anything that doesn't agree with their pre-conceived notions no matter how the tests were done, they'll find some little issue they don't like and use it to invalidate the tests. I'm not accusing cpac of doing that, mind you, but it's true of benchmarks in general.

Anyway, I think they're perfectly valid in establishing relative performance because this isn't the first time I've done these kinds of things and the order has never changed except for FireFox getting a very impressive speed boost. In other words it "proves" that Camino is the fastest and OmniWeb is the slowest (love it as I do that's just the way it is).

[JKT]

The question is how do you actually determine how to measure browser speed? I suggest reading David Hyatt's (lead developer of Safari) enlightening blog on this issue. Perception of speed doesn't necessarily equate with actual speed...

[JKT]

Originally posted by Mike S.:
If you don't like the numbers then don't bother to complain about how my testing sucks for reason xyz, I don't care. Everybody likes to say "this browser is fastest!" but nobody has bothered to even try to back it up with any kind of testing. Perceptions are not fact, these times are even if they are not statistically or scientifically perfect.

Mike, it would be better if you also expressed your figures in relative terms (e.g. set the fastest or slowest browser as 1.0 and express the speed of the others as a fraction of that), so that people could relate to them more easily.

For instance, OW 5 is ~ 3 times slower than Camino on your G3 and only ~ 2 times slower on the G4. Safari is ~ 1.3 times slower on both.

[Mike S.]

Originally posted by JKT:
Mike, it would be better if you also expressed your figures in relative terms (e.g. set the fastest or slowest browser as 1.0 and express the speed of the others as a fraction of that), so that people could relate to them more easily.

For instance, OW 5 is ~ 3 times slower than Camino on your G3 and only ~ 2 times slower on the G4. Safari is ~ 1.3 times slower on both.

Sounds good but math has always been my weak subject (that and non-english languages) so I just put up the numbers and let the individuals determine the best way to interpret them.

[ooninay]

Originally posted by Mike S.:
You can't take the numbers as absolute values, in other words just because it took 3 seconds to load a site on my eMac doesn't mean that browser will always load that page in 3 seconds on every piece of hardware or even another eMac of similar configuration.

Yeah, but none of what you're saying makes your data "meaningless", it just makes it more difficult to extrapolate to other situations. E.g., the data has a lot of meaning from your point of view unless you don't care about sitting around three times as long for your websites to load as you need to--it's just that those results may or may not replicate in another part of the country (or world, for that matter). Not that you ever suggested that they would. Maybe I'll try your test here and see if it's any different, that would be interesting.

[Mike S.]

Originally posted by ooninay:
Maybe I'll try your test here and see if it's any different, that would be interesting.

Well, you'll get different results than me just because I've got a weird set-up. The pages were locally cached versions of the site front pages (gotten via FireFox's save feature) and they were locally served over a 1Mb "home line" phone network.

Doing the same tests over 10 or 100 will give noticeably faster absolute speeds. I found this out the first time I did them on the PC; it was considerably faster because it was accessing the server at 10Mb vs 1Mb for the eMac. Once I dropped it to 1Mb the results fell more in line with the eMac. Their relative positions remained unchanged, i.e.. FireFox was still faster than IE.

Come to think of it, I wonder how much faster the eMac would have been if it could access at 100.

If you want to do the tests go right ahead; I'd like to see other people's results.

[cpac]

as your last post points out - many factors, including potentially most importantly, network/server speed, factor into how fast a browser can render pages.... (why not cache the pages locally, since that would have eliminated the network margin of error?, --though, there'd still be the issue of whether they remained in some other sort of cache for recently accessed files...)

When I said you numbers were "meaningless" what I meant was that people should pay attention to their own real world experience, rather than any of these pseudo-scientific tests. My own feeling is that almost all browsers are close enough in speed so as to make their relative rendering speeds inconsequential for any practical use

Indeed, I personally prefer OmniWeb because of its pro feature set - this saves me much more time in the long run than any speed lag it may have compared to other browsers.

Bottom line--what I was trying to express to the original poster--Is that rendering speed doesn't matter anymore. It's all about features and user interface. Pick what works for you best - not what any "tests" claim is fastest.