Submitted by “WebKit’s core JavaScript engine just got a new interpreter, code-named SquirrelFish.” According to the Webkit blog, SquirrelFish is 1.6 times faster than WebKit’s previous interpreter. This detailed blog entry explains exactly why the new engine is so much faster and the Webkit team’s plans for continued improvement in rendering speeds.
Webkit Introduces New Javascript Interpreter
About The Author
Adam Scheinberg
Technology Executive • Web Developer • Father • Foodie • Music Snob • OS enthusiast
Follow me on Mastodon @[email protected]
I wonder if this thing will find its way back to khtml.
Hasn’t KHTML been replaced by the KDE (or is it Qt?) port of WebKit?
It has been included into Qt.
http://www.osnews.com/story/19720/Qt_4.4_Released
But unfortunately Konqueror does not use it.
Edited 2008-06-03 19:36 UTC
Not likely. KHTML devs have done their own Javascript performance work. http://www.kdedevelopers.org/node/3476
Too bad really to see this duplication of effort still going on, but lots of changes between the two engines have accumulated over the years…
WebKit – Safari, OmniWeb, Epiphany, Konquerer
Gecko – Mozilla, Firefox, Camino
Trident – Internet Explorer
Have I missed any that are relevant today? I believe Netscape had a different one, but Gecko replaced it.
Presto – Opera
Epiphany uses Gecko. There is an experimental build that uses Webkit, however.
It is switching to Webkit by default for 2.23. Not that I can actually find any updated code for this since 2.22, it must be hiding somewhere private…2.22’s Webkit backend basically works but leaves a lot to be desired (no extensions work with it, for instance, so no AdBlock).
Weren’t the KDE devs still fighting over KHTML vs WebKit? Because right now, even though WebKit was forked, KHTML is a separate rendering engine. There were plans to fuse again, but I’m not sure if that has gotten enough support in the community.
there is a google SoC application for a webkit kpart that you can use in konqueror http://websvn.kde.org/trunk/playground/libs/webkitkde/
Another WebKit/Cocoa browser: Shiira
http://shiira.jp/en.php
I like Shiira but it is way too unstable for me, crashes every few minutes…
Add Arora to Webkit.
http://code.google.com/p/arora/
There’s also Swift, but it’s not quite as far along as Arora.
http://try.swift.ws/
How “major” do these have to be in order to be counted?
There’s gtkhtml – does dillo use this, or another?
chimera2 has its own pretty bad enjine, but I admit it’s ancient
HotJava has its own engine
Hv3 uses Tkhtml3
Do we count JS engines seperately? As we have seen, the JS engine just changed for webkit and may change for Gecko.
Do we count text mode browsers like links, lynx, w3m, etc?
Webkit officially doesn’t suck and is a good contender for first place in the browser wars 2.0 as soon as this gets released.
best codebase, very fast, and one of the most standards compliant. It’s achilles heel has been js for quite awhile now.
Officially “doesn’t suck?” Have you seen the canvasing abilities in Webkit? The built in CSS rounding, reflections, masking, etc? Webkit is absolutely amazing.
You are right, it is amazing, and I have said as much in several places, including the second half of that post.
But since bad javascript means a bad experience on pretty much any website with any interactivity on it whatsoever, having bad javascript has been the big issue for webkit for quite awhile now. Now that it is being addressed, there isn’t really all that much holding it back from world domination.
I ran the javascript speed test from here:
http://celtickane.com/webdesign/jsspeed.php
The results are amazing. My main browser is Epiphany 2.20. It is built against Firefox 2.0.0.14. I also have the latest nighly build of webkit installed with the latest version of Midori for testing. Epiphany/Gecko scores a pitiful 746ms in comparison with Midori/Webkit’s 88ms. Now I have to install Firefox 3 to see how it compares on my machine.
These tests confirm the advantage:
http://www.jorendorff.com/articles/javascript/speed-test.html
Edited 2008-06-04 00:38 UTC
I forgot the Sunspider benchmarks. Here are the results:
Epiphany/Gecko
http://webkit.org/perf/sunspider-0.9/sunspider-results.html?%7B…
Midori/Webkit
http://webkit.org/perf/sunspider-0.9/sunspider-results.html?%7B…
It’s good to see that Webkit’s performance is better across the board and not just focused on one particular area. That’s the danger with these sort of benchmarks…
Note that
means the same as 2.6 times as fast.
Do they actually mean 160% faster, or, more probably, 60% faster? It’s quite a significant difference, as the ratio between 1.6 and 0.6 is 2.67 (or 267% !)
The times are right there in the article, it’s 60%.
My experience has taught me that nobody ever uses a phrase like 1.6 time faster correctly. When they say that, what they mean to say is 1.6 times as fast.
It annoys me to death and I’ve given up on trying to educate the people I meet.
With the term “faster” people should use percentages. 60% faster….take 60% of the original and add that to the original.
With the term “as fast”, ratios are good. 1.6 times as fast.
Using ratios with “faster” is weird. “0.6 times faster” appears like it is slower.
Is there a Gnome browser that is capable of running this engine?
Is it really faster, or are they just bullshitting the benchmarks like they do whenever they show opera being the slowest browser… or if they are gaming it because their onload method fires prematurely.