Camino Builds w/Custom BM Bar [PPC7450, 1.8 Branch, 10.4+]

Talk about the native Mac OS X browser.

Moderator: Camino Developers

Post Reply
Old davedit
Posts: 0
Joined: December 31st, 1969, 5:00 pm

Camino Builds w/Custom BM Bar [PPC7450, 1.8 Branch, 10.4+]

Post by Old davedit »

You can find builds of Camino optimized for the PPC7450 (Newer G4) here!
If that URL doesn't work for you, try this mirror. (Thanks to batmanppc)
All updates are now posted in my RSS feed, as this thread is getting extremely long.

These builds are being compiled by Phoenix2005, and I'm posting them. If you have any thoughts, critiques, suggestions, opinions, questions, etc, please lemme know. Enjoy!

UPDATE: These builds now use a custom bookmark bar.
Last edited by Old davedit on June 28th, 2006, 10:40 pm, edited 8 times in total.
Old davedit
Posts: 0
Joined: December 31st, 1969, 5:00 pm

Post by Old davedit »

12/27 build is up, this time in 3 different optimization levels: O3, O2, and O1. If you like, try them all out and see if find any noticable differences between them (rendering, scrolling, etc).

EDIT: [Widget problem resolved]
Last edited by Old davedit on December 28th, 2004, 3:16 pm, edited 1 time in total.
User avatar
cflawson
Posts: 4721
Joined: December 26th, 2004, 2:54 pm
Location: Flying over your house in a red, white, and blue jet
Contact:

Post by cflawson »

Back-link to the old thread from whence this came:

http://forums.mozillazine.org/viewtopic.php?t=160902

Just in case.

cl
mrmister
Posts: 90
Joined: July 18th, 2003, 10:29 pm

Post by mrmister »

What are the differences between the three builds--I can't tell by their names.
Old davedit
Posts: 0
Joined: December 31st, 1969, 5:00 pm

Post by Old davedit »

I'm afraid I don't really know (I just post them!), but try this site... Hopefully that helps

Apparently optimization levels can go from -1 (completely disabled) to 6, and GCC supports from 1-3 (or so I was told). Camino needs some tweaking with level 3 (because of inlining?)
Old davedit
Posts: 0
Joined: December 31st, 1969, 5:00 pm

Post by Old davedit »

12/28 is up now... And good news, the widget problem is fixed.

I'm looking for some more feedback on these builds if possible. I'm curious if you guys are finding it faster or even slower, and what kind of system you have. I need to know if they're worth continuing, as they do take time, space, and bandwidth. Thanks :)
User avatar
Hangnail
Posts: 306
Joined: October 9th, 2003, 5:45 pm
Location: Boston, MA

Post by Hangnail »

I'm using a g4 powerbook 12 inch (1.3 ghz) with 1.25 gig ram. I've noticed mostly a very fast startup time on the optimized builds, especially the o1 variety. Page rendering is just about the same or alittle faster. The 'faster' seems associated with complex pages with a lot of graphics. Table based pages like the forums load so fast in any flavor that it is hard to see any speed differences.
valkraider
Posts: 99
Joined: January 15th, 2004, 4:51 pm
Location: Portland, OR

Post by valkraider »

davedit wrote:12/28 is up now... And good news, the widget problem is fixed.

I'm looking for some more feedback on these builds if possible. I'm curious if you guys are finding it faster or even slower, and what kind of system you have. I need to know if they're worth continuing, as they do take time, space, and bandwidth. Thanks :)


Nice. Any word on what it was causing the problem? It is a curious one....

Does anyone know of a good site for browser performance testing? ;)
Old davedit
Posts: 0
Joined: December 31st, 1969, 5:00 pm

Post by Old davedit »

Ya, I did notice a faster startup time as well, though it was about the same for me regardless of the optimization level.

As for the widget problem, not sure what caused it. I noticed the problem wasn't there in powerbook's old optimized Camino build, so my friend started over from scratch, and whatever happened, the widget problem disappeared.

As for browser testing performance, you could try this stopwatch site, though I dunno how accurate it is. And I'm not sure if the optimized builds should affect javascript and/or flash performance...
User avatar
krmathis
Posts: 4699
Joined: May 24th, 2004, 9:40 pm
Location: Oslo, Norway
Contact:

Post by krmathis »

User avatar
cflawson
Posts: 4721
Joined: December 26th, 2004, 2:54 pm
Location: Flying over your house in a red, white, and blue jet
Contact:

Post by cflawson »

When I first grabbed the November build that powerbook did, I noticed a HUGE speedup on anything that involved complex page rendering or lots of graphics. Sites that are mostly text, like this one, don't seem all that much faster.

I haven't grabbed a recent official nightly to compare, but I don't doubt the optimisation is having an effect. I'll run those speed tests suggested in the previous post tomorrow night and see what I come up with.

EDIT: I just grabbed that 12/28 version off the site, Dave, and it's b0rk3n. Not sure what's wrong, but it downloaded in about 1:30 on a 28.8 dialup and gave me a nice ugly corrupt .tgz file. Haven't re-tried it, though; might be on my end. Just a heads-up in case.

cl
Old davedit
Posts: 0
Joined: December 31st, 1969, 5:00 pm

Post by Old davedit »

Weird... I think it's probably on your side. I just tried downloading it, and no problems.

12/29 is up now though. You can try grabbing that instead. :]

I tried some of those speed tests. In the beginning, the official nightly was faster in the first and third tests listed there, but slower in the second. But I tried again later, and the speeds of the official and optimized nightly were about the same. So yeah, I dunno...
valkraider
Posts: 99
Joined: January 15th, 2004, 4:51 pm
Location: Portland, OR

Post by valkraider »

davedit wrote:I tried some of those speed tests. In the beginning, the official nightly was faster in the first and third tests listed there, but slower in the second. But I tried again later, and the speeds of the official and optimized nightly were about the same. So yeah, I dunno...


I think "speeds" might be misleading when using a faster machine. For example, on my 1.5ghz powerbook, I think that for the most part you can't tell the difference between the optimized and non optimized version. (Except the optimized version for some reason puts 0000000000 (v0.8+) in the "about" info for a version number :) )

What could be better to watch is processor usage while running the tests. It is possible that the optimized versions consume less CPU - but not noticable compared to the official builds because we have extra CPU left over.

Where the difference might be noticable is where you have a slow machine, or a machine under heavy load. Then every CPU cycle counts tremendously.

Also, an optimized build might be better on battery life and heat generation - not taxing the CPU quite as much. But then again, it could be worse - stressing the CPU more as you get rid of latency issues...

All in all, I am not a processor engineer, I just play one on Mozillazine. So all this is hyper-theorhetical.

But the optimized builds feel Snappier(tm)!
herbs
Posts: 568
Joined: November 22nd, 2003, 8:39 pm

Post by herbs »

Howdy,

What impresses me more is that the search through the Cache to set up the favicon in the bookmark menu seems to go much faster. Outside of that I think it's difficult see the speed differences. It feels snappy although I couldn't say significantly snappier than the standard nightly.

I still see Camino take up to 80% of the processor for at least some time when waking from sleep or when there are several (>=3) swap pages. This is true of the standard build too.

I'm running 10.3.7 on a 550MHZ Powerbook with 512MB RAM and the 12/28 build.

Good Luck,
Herb Schulz

P.S. I'm still seeing the same hesitation in text boxes as the regular build.
Old davedit
Posts: 0
Joined: December 31st, 1969, 5:00 pm

Post by Old davedit »

valkraider wrote:I think "speeds" might be misleading when using a faster machine. For example, on my 1.5ghz powerbook, I think that for the most part you can't tell the difference between the optimized and non optimized version. (Except the optimized version for some reason puts 0000000000 (v0.8+) in the "about" info for a version number Smile )

What could be better to watch is processor usage while running the tests. It is possible that the optimized versions consume less CPU - but not noticable compared to the official builds because we have extra CPU left over.

Where the difference might be noticable is where you have a slow machine, or a machine under heavy load. Then every CPU cycle counts tremendously.


Ya I was thinking that as well... that the difference would be bigger on slower machines. I can't tell much of a difference on my 1.5ghz Powerbook, but herbs can't tell much of a difference, other than the Cache, on his 550mhz Powerbook either. So... *shrugs*
Post Reply