[FX] elfurbe: Nightly [Branch|Trunk/OSX] (G3/G4/G5/Intel)

[jram]

Thanks for explaining, and again, keep up the great work.

[elfurbe]

Alright, heck with it. Live fire time. Updates enabled. Post back with problems, I tested it on my MBP and my G5. I can't remember for sure when I started building the update packaging, so if you're not getting updates for your nightly, grab a new nightly and then wait a day. Fingers crossed...

[jram]

Just tried to update on my MBP, went fine, no problems.

[elfurbe]

Good to know someone even tried. Glad it's working for you. Looks like you're running 3.7, and I've been successfully updating for a week or more with no problems on my MBP with 3.6, so I think it's safe to call Intel builds update-enabled. Has anyone had luck with any of the PPC builds?

[drseuss9311]

i would have tried if my ibook G4 didn't look like this right now: http://friendfeed.com/ccheath/6ea0a61d/bad-hard-drive -- still holding out hope that i might be able to resurrect the hard drive and then copy the data to a new one so i can put it all back together again -- hopefully ...

[LordBodak]

Try putting it in an external enclosure and then standing it on its side. When the drive failed in my PowerBook G4 I was able to get all the data off of it while it was on its side.

[drseuss9311]

i've already bought a 'jumper' cable (if you will) http://friendfeed.com/ccheath/e8ae8027/ ... hard-drive but haven't been able to get my spare pc to boot any repair discs (i think something's fishy with that pc) but when i do get another donor pc to work with (which will be soon hopefully) i'll give that 'on its side' thing a try - thanks!

[darm1k]

Has anyone had luck with any of the PPC builds?

I just installed FXG3-3.7a1pre-2009.08.26 and let it update to today's - worked great!

Thanks!!!!!

- Dan.

[jram]

I just updated, noscript is broken. It won't let me surf, the options on the toolbar are blank. When disabled I can surf, got flash block until fixed. Tried older versions, same thing. Anyone else ?

[elfurbe]

@drseuss - Harsh on the iBook... One nice thing about working in university IT, there are a ton of spare computers around.

@darm1k - Boss. Sounds like maybe I did this right!

When 3.5.3 comes out, we'll find out if stable updates work, and then it's all set.

[ppc]

I see no performance gain with the optimized G4 build. It takes more than a minute on this iBook to launch Firefox and reload a set of tabs from the last session. Same with the optimized build. It could go 90 seconds from launch to the completion of session restore and I'm still not absolutely sure that I noticed a 1-second difference on the clock between the stock build and the optimized build. I have lots of free RAM and the CPU load is typically below 10% until I launch the browser. This app has total freedom to demonstrate what it can do with the CPU, and I see no improvement. But Safari is easily twice as fast as any Firefox build.

If I have overlooked something and other people are seeing a noticeable speed boost with these alternate builds, I would certainly like to know. For example, if there was some kind of benchmark page that everyone could use to collect performance statistics, I might reconsider... but right now I dont see any reason to stop using the official builds.

[darm1k]

On my Smurf, the optimized builds are waaaay faster than the regular release. On my 1.5 GHz PowerBook G4, the diff isn't so big, but still noticeable. Of course, Safari 4 (with or without webkit nightly) is always faster.

Firefox can be greatly slowed by grody long bookmark lists. And page loading can be greatly slowed if those you've choosen are flash or java heavy.

- Dan.

[ppc]

@darm1k: Thanks for the feedback.

I dont have any bookmarks. I was using NoScript to block Flash but now using Flashblock. Maybe the extensions I have installed are negating any performance gain. I guess this goes back to the Javascript engine, which seems like the primary cause of Firefox slowness. Thats why Safari is so much faster:

http://webkit.org/blog/189/announcing-squirrelfish/

I can see why a G3 optimized build would make a larger difference on that platform, but if the goal is to support old hardware I think there are millions more systems which would benefit from an OS 9 version of Firefox. The main reason those systems are no longer useful is the lack of a modern, standards-compliant browser. In any case, what Firefox seems to need most is a bit of source code optimization.

http://www.besttechie.net/2009/06/12/fi ... ad-slower/

[ppc]

[just sharing this as a general reference]

Sunspider Javascript Benchmark

This is SunSpider, a JavaScript benchmark. This benchmark tests the core JavaScript language only, not the DOM or other browser APIs. It is designed to compare different versions of the same browser, and different browsers to each other.

Webkit browsers

Radon 7164 ms (average)

3d: 1300 ms
access: 880 ms
bitops: 593 ms
controlflow: 63 ms
crypto: 442 ms
date: 503 ms
math: 941 ms
regexp: 1014 ms
string: 1426 ms

Safari 7769 ms (average)

3d: 1495 ms
access: 1085 ms
bitops: 584 ms
controlflow: 67 ms
crypto: 465 ms
date: 468 ms
math: 1168 ms
regexp: 1022 ms
string: 1413 ms

Sunrise 7211 ms (average)

3d: 1327 ms
access: 928 ms
bitops: 575 ms
controlflow: 63 ms
crypto: 431 ms
date: 474 ms
math: 984 ms
regexp: 1002 ms
string: 1425 ms

OmniWeb 7253 ms (average)

3d: 1325 ms
access: 880 ms
bitops: 566 ms
controlflow: 64 ms
crypto: 425 ms
date: 492 ms
math: 904 ms
regexp: 1082 ms
string: 1512 ms

Shiira (crashed)
__________________________

Mozilla browsers

Firefox 12484 ms (average), with extensions

3d: 1628 ms
access: 1989 ms
bitops: 1601 ms
controlflow: 157 ms
crypto: 731 ms
date: 1012 ms
math: 1609 ms
regexp: 914 ms
string: 2839 ms

Firefox 12579 ms (average), safe mode-no extensions

3d: 1585 ms
access: 1991 ms
bitops: 1656 ms
controlflow: 157 ms
crypto: 726 ms
date: 1083 ms
math: 1688 ms
regexp: 928 ms
string: 2760 ms

Shiretoko 12632 (average), safe mode-no extensions

3d: 1699 ms
access: 2018 ms
bitops: 1653 ms
controlflow: 181 ms
crypto: 729 ms
date: 1041 ms
math: 1691 ms
regexp: 849 ms
string: 2769 ms

__________________________

*All tests run on a G4 under Mac OS 10.5.8

My conclusions:

1. JavaScript is not the only measure of performance, but Webkit browsers are typically 2 to 3 times faster than Mozilla browsers due to the JavaScript engine.

2. The greatest performance gains are in these JavaScript functions:

access (2x)
bitops (3x)
controlflow (2x)
date (2x)
string (2x)

3. Some WebKit browsers might be faster than others, but there is no appreciable difference between them in Javascript performance. Speed gains are dependent on page content and functionality is probably more relevant to the user.

4. Running Firefox with a full suite of extensions has negligible impact on JavaScript performance unless an extension is buggy.

5. The optimized G4 build does not provide better JavaScript performance than the official build.

[darm1k]

@darm1k:
I can see why a G3 optimized build would make a larger difference on that platform, but if the goal is to support old hardware I think there are millions more systems which would benefit from an OS 9 version of Firefox. The main reason those systems are no longer useful is the lack of a modern, standards-compliant browser. In any case, what Firefox seems to need most is a bit of source code optimization.

Updated WaMCom Mozilla, for OS 9 ....
http://www.floodgap.com/software/classilla/