I have found WebSpeed at

wyciwyg://1/http://www.numion.com/YourSpeed/Checkup.php?L=us

to be THE BEST Browser speed tester. It's great for speed tweaking your overall system and browser settings.
Instead of communicating with 1 site, WebSpeed averages the actual through-put speed to 40 of the most popular sites. It also bypasses the effects of the browser's cache.

Incidentaly, as measured by WebSpeed, K-Meleon beats IE handily and totally blows-away Phoenix and other pretenders to the throne. On my system K-Meleon beta 0.6.5, build 44 is noticeably faster than the current build.

Bugs or no, K-Meleon beta 0.6.5, build 44 and IE 5.5 are my Browsers.

GHM

Where is a link to see the performance comparison? I didn't see it there.

Measure it yourself by running the test (several times) with each browser on your machine.
Note: other programs that may be running in the background; the browsers' configuration settings, the inherent speed of your system, how well it's tweaked, etc. can have BIG effects on through-put speed.

GHM

I ran the test several times on both Kmeleon and Netcaptor. Netcaptor was at least 20 percent faster each time I ran it. There was considerable variation each time on each browser but Netcaptor was always much faster.

Dont know what this means if anything, but just thought I would report my results, since they are so different from yoursl

K-meleon was on my machine 2.7744 nano-seconds faster than IE, and 1.2764 nano seconds faster than Phoenix!!!!!11111

K-MELEON R0X0RZ!!!!!

</dumbass i-need-a-life mode>

thats what I think of speed comparisons.

to <lame username>: Isn't speed sort of a part of K-Meleons purpose? Benchmarking it to see if it's faster is a given, because without actual numbers, optimizations cannot be tested. Speaking as someone who has done optimization in the past, I can tell you that it's incredibly hard to be objective concerning optimizations you're proud of -- only by having actual numbers on hand can you really tell if you're going faster or not. This is especially important because some things which under most circumstances should be speeding things up can slow things down if something unexpected happens (case in point -- I once designed an ultra-efficient routine to draw to the screen in one program, but it was so much bigger than the original routine, that it ran slower because it wouldn't fit into my pentiums puny cache.)

For users, it's more a measuring stick to find out if things really are going faster, or if it just *feels* faster. Not quite as critical as it is to the developer, but if you are trying to get the fastest software for the job, "feels faster" won't cut it.

Opera says it's the fastest browser, and with the new version out sometime, we need a way of comparing them. Numion's webspeed still won't do Opera, so has anyone any other suggestions? I feel that Opera is presently slightly faster, but offset by K-meleon being more IE 'compatible'.

> I feel that Opera is presently slightly faster, but offset by K-meleon being more IE 'compatible'.

Too bad Opera is nowere near Geckos standards support either.
At least with IE you can use simple parsingbugs to work around lots of the problems, but Opera 6.05 is almost as buggy with CSS as IE 6, but has no good way to work around the bugs
Really hope Opera 7 will be a lot better, so it's actually usable on anything but the slowest of comps (that can't run K-Mel).

Correct me if I'm wrong but this speedtest has nothing to do with "browser speed" asu such unless you consider your line speed as browser speed. Browser rendering speed has nothing to do with loading 40 sites at once. If one was to measure browser rendering engine speed then it should be done with complex markup test.

As for differences between browsers - using pipelining, for example, makes great difference in concurrent loading test. I'm not sure IE has it enabled as default, latest Mozilla based browsers tend to have, I think.

I have no idea why, sven, but K-Meleon did have higher transfer rates than the other two browsers when I tested them. I thought it might be just the nature of the internet, but Moz and IE consistently got 26k, and KM consistently got 31k.

> unless you consider your line speed as browser speed

Actually, since you compair several browsers on the exact same line, you do test the RL performance of the browsercode, when compairing 2+ browsers.
Netcode optimization and GUI CPU overhead makes a difference...

> If one was to measure browser rendering engine speed then it should be done with complex markup test.

40 sites at once make up for a resonably complex markuptest don't you think?

> using pipelining, for example, makes great difference in concurrent loading test. I'm not sure IE has it enabled as default

Then by default IE is broken. What else is new

> 40 sites at once make up for a resonably complex
> markuptest don't you think?

Not really. Look at the code of the page. These sites with all their markup are never loaded only images from according servers are loaded repeatedly. This measures average loading speed from different locations but not really rendering speed

To measure rendering speed you'd have to actually load those sites and get them rendered, in small iframes, for example, then measure their loading times, sum those up and calculate from there.

Then again - that'd depend also on loading speed