Slashdot Mirror
Mac OS X Slow for Web Browsing?
Atryn writes "Wired News has reportedly confirmed user performance complaints in their own tests. From the article: 'That was a conscious decision Apple made,' Mac MSIE project manager Jimmy Grewal said. 'They optimized for user experience rather than raw performance.'" My hunch is that you can take care of many Mac OS X performance issues by logging in as user ">console" ...
Looking at just web browsing speed on an OS is not a great reason to choose one over an another.
Strange women lying in ponds distributing swords is no basis for a system of government.
Chimera is, according to these tests, the fastest MacOS Web browser by a factor of 2.
Chimera is, of course, based on Gecko, the Mozilla rendering engine. It's mainly the work of Mozilla uber-hacker Dave Hyatt.
Gerv
It tells you the answer...
You'd have a hard case arguing that OSX doesn't have room for improvement speedwise, but it's this horrible thing that some people like to pretend that it is. Some of the blame goes to Apple, some goes to the application writers. Mac IE renders some stuff painfully slow. I don't know why. Like the article said, things like slashdot comments feel like they're taking all day. In reality, it's only 5 seconds, but we all know what sort of attention spans people have nowadays. There's a pretty new browser called Chimera that is early in development, and still has a limited feature set, but it renders things almost instantly, including slashdot comments. So there isn't some inherent problem within the OS that makes it impossible for your applications to function reasonably.
Not to sound too much like an apple apologist, but they've done quite a bit to get OSX to where it is so far, and the more I use it, the more I appreciate where it's advanced over OS9. I don't mind waiting a bit for things to improve. Just like I don't really mind anymore waiting 5 seconds for IE to throw together the comment threads. Most of us could benefit from learning a little patience.
Although I would surmize that it's apple's fault that they get judged so harshly. Seeing as steve jobs claims that every time someone in their company makes a sketch on a post-it note, they've created a new revolution in the world, people are justified in being extremely critical.
One time I threw a brick at a duck.
What makes you think IE on the Mac is just a patchwork port running in some sort of mythical compatability layer? While it may be true that older versions of IE on the Mac (versions 4.5 and earlier) were based on the Windows versions, version 5 was built from scratch for the Mac. I would encourage you to do a little research before posting next time.
tim
I use OmniWeb. Primarily. It's render outclasses anything else on the platform. It's very fast too in comparaison to IE.
Blaming Apple for IE's sluggish performance is a bit easy. Coming from the IE project manager, it's downright insulting.
For browsing outside a proxy, I sometime uses the new Chimera browser. It's a Cocoa (Objective-C) -based browser that's based on Fizilla. Fizilla is a Mac OS X version of Mozzila.
Chimera is astonishingly fast. It's render is better than Netscape 6.2, but like OmniWeb, it's JavaScript support is still lacking somewhat. Fortunately, javascript support isn't an issue for me, unless I require online banquing, where I'll use Netscape 6.2 (despite it's utter ugliness).
The problem is in the Tasmin rendering engine used by IE for Mac. But blaming Apple seemed to be the easiest thing for them to do.
There are certainly performance problems in OS X's UI, but let's give blame where blame is due.
-jon
Remember Amalek.
No, it doesn't. It uses PDF. NextStep/OpenStep used display PostScript. PDF is a different animal.
www.lucernesys.comHorizon: Calendar-based personal finance
Outside of Apple Corp. MBU (Mac Business Unit) @ Microsoft is the largest Apple development shop in the world, and Apple has noted this more than once. The MBU is largely independent of the rest of MSFT (which is why Office.X is so much nicer/faster/more stable than Office 2000). I haven't seen a performance problem with IE on a G3-500 running both 9.2 and OSX.
this is getting old and so are you
blog
- Win IE & Mac IE have completely different codebases. If there's overlap it is only in snippets of code shared between the development teams.
- Aside from knowing the product history this is easily demonstrated by looking at the errata for each browser. They have very different feature sets / CSS implementations / rendering issues / etc.
- Win IE 5.x is a "Carbon" application; this means it is running using a set of libraries based on the old MacOS. However it is not running in the old MacOS itself (a "Classic" application). Indeed in spite of being a Carbon application the IE 5.x for MacOS X cannot run on MacOS (though there are IE 5.x for MacOS.)
- This is in line with MS Office v.X which hasn't been code-synched with it's Wintel cousin for years, is also Carbon-based, and also does not run on MacOS.
- So, in point of fact, you've got every one of yours wrong.
Mac IE is not a port of Win IE, is not running in an emulation layer, and has no excuse not to be faster.On the other hand Mac IE is more standards-compliant overall then it's Wintel cousin in spite of some glaring CSS deficiencies & other asst'd bugs. It has a notably better design in some areas, incorporates some nice features like the left-hand bar, and a much better cache (as in not-broken.)
Of course Win IE has it's own set of bugs and deficiencies so overall they're about equal with the Mac IE being somewhat more "right" & the Win IE getting more support from sites.
For the future I expect that Carbon applications like Mac IE will be eventually replaced (or superseded.) Though they've been pushed farther then Apple originally wanted (gotten more features, more support, etc.) they're still not as effective at taking advantage of MacOS X as Cocoa applications are. On the other hand they're a relatively easy port and work nearly as well so they're the obvious step for developers with large code bases and little familiarity with Objective-C & Apple's Next-derived OO development environment.
I don't read ACs: If a post isn't worth so much as a nom de plume to its author then I wont bother either.
The main issue is the Internet Explorer still runs off of the "Classic Event Model" where it constantly polls for new events. The newer Carbon event model supports those old methods, because EVERYONE used them in the old system. Think how much CPU that takes when all those old programs (even though they are "carbon compliant") are constantly jumping up and down asking if they've gotten an event.
o n/ CarbonPortingTools/carbonportingtools.html
The new "Carbon Event Model" allows you to associate events with handlers, and when an event fires that you'd like to pay attention to, your call-back gets fired. Much more effecient.
The cocoa event model is even more robust.
The problem lies in that programers were able to compile a "carbon compliant" application, without moving to these new event models. THIS IS GOOD. Imagine how PISSED off a developer was if they were told, "Yea, you have to move all your event code over to this new system, cause it's better." No. A developer would rather have a product up and running on OS X natively, and then move over.
Anyway, it's not that Apple has "buggered" up the system someway, the applications have exploited the API's that Apple has made available, but it was a necissary evil.
http://developer.apple.com/techpubs/macosx/Carb
Has information about the carbon event model, and high performance computing.
- Sighuh?
You can drastically speed up your OS X machine's network speed by modifying some sysctl variables. Toss the following lines into a script somewhere:
/usr/sbin/sysctl -w net.inet.tcp.sendspace=65536
/usr/sbin/sysctl -w net.inet.tcp.recvspace=65536
/usr/sbin/sysctl -w kern.ipc.maxsockbuf=524288
/usr/sbin/sysctl -w net.inet.tcp.delayed_ack=0
/usr/sbin/sysctl -w net.inet.udp.recvspace=73728
It literally doubles my web browsing and file transfer speeds. This will probably be of value only to folks with broadband or ethernet connections. It wouldn't do much for obsolete modem users.
Recently, I'd been having some performance issues with Mac OS X on my titanium Powerbook 500. (256 meg of ram on 10.1.4)
/var/run/cron.pid, but don't quote me...)
The problem was that EVERYTHING gave me spinning beach ball. File operations, minimizing Finder windows, you name it...Even scrolling in MOzilla and IE were affected. Then I read on MacAddict that OS X needs to be left running all night so that various "cleanup" tasks can run.
Anybody who has OS X should consider leaving there machine up all night so these run... It will resolve a great many problems that you're having, and allow us to go back to bashing MS and Oracle instead of Apple...
Unix people familiar with cron should have no problem with editing the cleanups to run at a more reasonable hour than 3am, 4am, and 5am (like one when your machine will be running)... (I think the file to edit is
Alternately, if you're a regular mac user and don't feel like mucking about with the terminal, hit Version Tracker and pick up MacJanitor. It's a friendly GUI that lets to schedule your daily, weekly, and monthly jobs, or trip them manually on demand.
Since I'd used the machine, it had never been awake all night (I close the lid when I go to bed, usually before 3am...) so cron had never done anything to optimize my machine.
Now? All better. Faster than I remember 10.1.1 being...
Who did what now?
And, at least in Wired News tests, OS X didn't mimic 9.2's habit of locking up completely, requiring the Mac's power cord and/or battery to be removed in order to reboot it -- hardly a satisfying user experience.
-as quoted from the article
Last I checked, the reset button worked just as well for desktop macs as it does for a regular PC. And for laptops, a simple control-command-power press will reboot everytime, no matter how badly crashed.
T Money
World Domination with a plastic spoon since 1984
I'm a hardcore geek and bought an iBook earlier this year. Personally OS X makes sense for me. I can use all my favorite UNIX development tools and when I get a damn MS Office document from marketing I can actually open it in MS Office.
An no, ksh and vim aren't slow in OS X. Not to overshadow your point, because I think it's a good one... For geeks this is a perfect system too.
I'm running a quicksilver 733Hz G4, OSX 10.1.3, and right next to it a Dell Dimension 4100, 1Ghz P3. I'm on a Pacbell DSL link.
I loaded www.cnn.com and www.apple.com under both IE and moz (9.9) under both machines.
For cnn.com, IE5 and moz on the Dell were about the same, around 2s. (Moz was the fastest to get the banner ad up, maybe IE5 was fractionally quicker overall. Very hard to tell. IE5 had the worst outlier though -- one time it took 5s.)
Moz 9.9 OSX was around 2.5-3s, and IE5 on the Mac was slowest -- 3-4s.
All browsers loaded the Apple page pretty much instantaneously. I couldn't tell the difference.
Lesson #1: use Mozilla under OSX; it's been getting faster with each point release, while IE5's remained static. IE5 can be sluggish at times.
Lesson #2: there really isn't that much of a difference between the machines. I do a fair bit of surfing on both, and they're literally side-by-side, hooked up to the same monitor. Up until now they'd always seemed about the same speed, surfing-wise, to me. So I was taken aback by the article -- and after testing, I guess the OSX browsers are a *little* slower, but not so's you'd notice much.
Mind you, I do have plenty of memory. Perhaps the iMacs were hitting the VM a little hard? Or, the pixmaps for all those pretty alpha-blended graphics probably add up. I believe there's an option to store them compressed in memory to speed things up on low memory machines, probably mentioned on one of the numerous OSX hint sites.
A.
The first thing to do is learn how to build Mozilla under Mac OS X... it's rather a bitch, but well documented here. (Actually, it's a bitch to build under Linux too, and that's the easiest platform to build on.) Then apply the ATSUI rendering patch attached to bug 121540 (sorry, can't link to bugzilla from slashdot) and rebuild. Voila! As far as I can see, it looks like it may already be in the nightlies though, and possibly even in 1.0RC1... I haven't looked yet :-). If you find that it's not, I recommend grabbing the build from stevek's iDisk. It's a lot easier than building it if you're not already building mozilla. (I was, for other reasons. Quartz was a nice extra perk.)
As for leaving out all the composer and mail junk, I don't know of a way to do that. However, current builds of chimera are fast, have quartz rendering compiled in, and are browser only. As a bonus, it's got a nice native cocoa interface that gets better and better with every build. It's still got some bugs, but I find it pretty usable.
Hope this helps!
.sig: file not found