Any body care to share any Optimization tips for Jaguar to get the most performance out of the system. I already run X optimize and Mac Janitor and I find that run frequently, they keep the system runing fairly smooth. Anything else I should know?
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Jaguar Speed Tips
- Thread starter User X
- Start date
- Sort by reaction score
Change themes to get rid of the eye candy and performance will improve. Be careful though, cause Duality and most themes don't work right now. I used Themer and chose the 3rd included theme which is very subdued compared to Aqua.
Use TinkerTool, (the just released version) to remove the anti-aliasing.
Don't add any apps or drivers that you think have not been updated, like Logitech. They may end up taking cpu cycles for no reason and may not even work. I'd wait until September to start adding utilities and drivers.
Use TinkerTool, (the just released version) to remove the anti-aliasing.
Don't add any apps or drivers that you think have not been updated, like Logitech. They may end up taking cpu cycles for no reason and may not even work. I'd wait until September to start adding utilities and drivers.
Use shadowkiller to remove the rendered shadows 'neath every window and menu. This shows improvements in benchmarks EVERY time.
I don't know about performance, but for cleanup you can go through your kernel extensions. It's much like you would go through your OS 9 uneeded extensions.
Do you really need PC Card support on a Mac? Support for SCSI cards you'll never use? The ability for the computer to speak text?
Getting rid of these can prob free up some memory for things you use more.
You can get rid of them in the terminal with super-user privledges.
I don't know about performance, but for cleanup you can go through your kernel extensions. It's much like you would go through your OS 9 uneeded extensions.
Do you really need PC Card support on a Mac? Support for SCSI cards you'll never use? The ability for the computer to speak text?
Getting rid of these can prob free up some memory for things you use more.
You can get rid of them in the terminal with super-user privledges.
Buy a new Dual 1Ghz or 1.25 Ghz and stuff it full of RAM you will not be dissapointed in the speed. This is the fastest computer I've ever used and my PC friends agree.
"Extension" management in OS X - bad!
More info: http://www.osxfaq.com/Tutorials/extensions_hell/index.ws
I think suggesting this is a seriously bad idea. Trying to clean up an OS X system in the same way you'd do an OS 9 system probably won't give any noticable performance benefits, and can possibly do some real damage if you aren't sure which files to remove, it's NOT just limited to files in the /System/Library/Extensions folder. For example, the suggestion to remove speech "extensions" won't work - those files aren't kernel extensions anyway and are scattered all over the /System/Library folder. And you might not know if other parts of the OS might need whatever you decide to remove, which might cause crashes or even worse problems. OS X itself does the maintenance of which extensions to use depending on your hardware, so why not let it do the job? Unlike OS 9, kernel extensions do NOT take up extra memory if you don't have the hardware present in your system, they're dynamic so they only load and stay in memory if you need them.
More info: http://www.osxfaq.com/Tutorials/extensions_hell/index.ws
IMO, Shadow Killer improves performance but at the expense of readability. It's very difficult to distinguish overlapping windows from each other with this utility turned on, since OS X doesn't have a window border. I tried it for a while and removed it, it wasn't worth the small benefit in speed it gave my system. Your mileage may vary, of course.
I have to say that, as someone who has little (none!) Apple experience, but a reasonable amount of experience on Unix and deriratives (Linux especially), the notions of "optimisers" and "performance enhancers" are very strange. Whereas the naff filing systems of Microsoft have spawned a small industry (almost) of programs to optimise your disk layout, the Unix industry has spent decades of simply not needing the concept -the hard work was done at the concept/design stage. Unix systems can easily run a decade and have no-one saying their disks need optimising.
I agree with the comments that removing system components sounds both dangerous and unlikely to yield any relevant performance gains. In terms of code improvements, Unix (ie Darwin), etc, is event driven. Things don't happen unless something wants to know about it and then only when the relevant "happening" (event) occurs. So, unless Apple have been paricularly bad at their job (which I personally don't believe to be the case), I don't see there being performance overheads from having system components (e.g. speech recognition or whatever) that are not being used. There might be a small argument relating to code occupying RAM, but this is unlikely to make any difference in the real world. If some code is included but not needed, it will get paged to disc when when it first becomes an issue (ie it's occupying memory and something else could do with the memory) and then it will never get paged back in as it will never be needed.
I'd couple this with a principle of not modifying complex systems unless you're absolutely certain what you're doing. IFF Apple had managed to get things sufficiently wrong that un-used components contributed undesirable overhead, the "correct" fix is for Apple to work out - not for these components to be deleted by the average user.
More 2 cents worth.
I agree with the comments that removing system components sounds both dangerous and unlikely to yield any relevant performance gains. In terms of code improvements, Unix (ie Darwin), etc, is event driven. Things don't happen unless something wants to know about it and then only when the relevant "happening" (event) occurs. So, unless Apple have been paricularly bad at their job (which I personally don't believe to be the case), I don't see there being performance overheads from having system components (e.g. speech recognition or whatever) that are not being used. There might be a small argument relating to code occupying RAM, but this is unlikely to make any difference in the real world. If some code is included but not needed, it will get paged to disc when when it first becomes an issue (ie it's occupying memory and something else could do with the memory) and then it will never get paged back in as it will never be needed.
I'd couple this with a principle of not modifying complex systems unless you're absolutely certain what you're doing. IFF Apple had managed to get things sufficiently wrong that un-used components contributed undesirable overhead, the "correct" fix is for Apple to work out - not for these components to be deleted by the average user.
More 2 cents worth.
Thanks for being another voice of reason...
Many of the users here (and in the real world) are stuck in the OS 9 mentality. In OS 9, disk optimization, and disk maintenance tools were vital to keeping a system in top form.
Now users are continuing their habits, even though the OS has made many of them unnecessary. I have had more complaints about Norton rendering a drive unusable than I care to mention. I often ask, why were you running Norton. The response is always, 'I do it once a month'.
Folks, you are now running pretty UNIX. If you don't have a problem, don't fix 'it'. Norton for X is a disaster, and the cause of countless kernel panics, and much lost data. Drive 10 is pretty bad as well. The system runs fsck on boot, and will fix many node problems with only a reboot. If you still have trouble, boot into single user mode (cmd-s at startup), and type 'fsck -y'. Repeat this command until no corrections are made, then type reboot.
In a worst case scenario, boot into 9, and run Disk Warrior. It has the best (and most trustworthy) directory repair for HFS+ filesystems.
Outside of that, the best advice I can give you is to avoid speed software. Unless you understand what you are doing, you can add overhead to your system, create problems, break things, and gain minimal speed increases. I guarantee that a $20 peice of shareware (or a $60 piece of software) is not as well tested or robust as Apple's OS on Apple's Hardware tested by Apple's QA and Dev teams. If it ain't broke....
P.S. I realize this has nada to do with the topic, but hey...
Re: "Extension" management in OS X - bad!
I suggested this, because there is no reason for things like an UltraTek 66 driver to be on my mac, when I'll never use it. Just extra space taken up. I speak only English, so anything non-English is gone too. Etc, etc.
But that's just me. I was happy with supper-buggy buggy 6c48 over stable but slow 10.1.5.
If I let my system go idle I'll never learn anything. I usually learn by breaking things, then fixing them.
I suggested this, because there is no reason for things like an UltraTek 66 driver to be on my mac, when I'll never use it. Just extra space taken up. I speak only English, so anything non-English is gone too. Etc, etc.
I'm so used to the Shadow killer look, that when I go to a machine with those heavy rendered borders, want to change it back.
But that's just me. I was happy with supper-buggy buggy 6c48 over stable but slow 10.1.5.
Shadow Killer is great. And you would be surprised how much CPU the Aqua interface eats up. Plus I hate the whiteness of it. I was using Duality with a theme called "Underling" that was very plain gray but it wacked my whole system out and I wound up doing a fresh install. Turn off Dock magnification and basically any and all eye candy options. They are a total waste.
I'm not certain whether there's a hint of sarcasm or not in this. But maybe I should have a go at positively contributing ideas on performance enhancmenet, and maybe some other readers can give me the benefit of their experience.
I'm considering spending money in two areas to improve performance: buying RAM and buying a RAID card to do RAID0 striping. The Sonnet Tempo RAID133 sounds like a reasonable choice, based upon on too much hunting around. Anyone got any experience using this - are the gains worth spending the extra money?
Re: Re: "Extension" management in OS X - bad!
Fun indeed, but potentially dangerous to recommend to others
Having looked through the G4 data sheet from Moto, I noted all the performance event counting registers. ADC offers some tools to use this to and perform observations on code, including code not specifically compiled to contain information gathering (a la gprof, etc). If you want a real challenge (and this is something I want to play with when my DP arrives next week) - try improving Darwin (source available - so improve in the recode and recompile sense) to reduce any negative observations made. For example, what happens with I/D caching when an interrupt occurs? Might there be improvements to locking/flushing/ignoring the cache during interrupts that improves performance? Unix type kernels are notorious for featuring many sections that of code that don't loop - ie they have little to be gained from being put into I cache and have a negative impact on whatever user code was previously in the I cache.
BTW, I expect/hope that Apple will already have done a lot of work in this area so I don't expect it to be easy to find any ways of improving performance - but this just adds to the challenge.
Fun indeed, but potentially dangerous to recommend to others
Having looked through the G4 data sheet from Moto, I noted all the performance event counting registers. ADC offers some tools to use this to and perform observations on code, including code not specifically compiled to contain information gathering (a la gprof, etc). If you want a real challenge (and this is something I want to play with when my DP arrives next week) - try improving Darwin (source available - so improve in the recode and recompile sense) to reduce any negative observations made. For example, what happens with I/D caching when an interrupt occurs? Might there be improvements to locking/flushing/ignoring the cache during interrupts that improves performance? Unix type kernels are notorious for featuring many sections that of code that don't loop - ie they have little to be gained from being put into I cache and have a negative impact on whatever user code was previously in the I cache.
BTW, I expect/hope that Apple will already have done a lot of work in this area so I don't expect it to be easy to find any ways of improving performance - but this just adds to the challenge.
I've heard the Sonnet has some problems in 10.2. I have a SIIG (same as the Acard/Miglia) which is great!
Sadly, I don't have 9 anymore, and there don't seem to be any real good disk benchmarkers for X, but it feels about 50% as fast as my U160 12 drive RAID (which is plenty fast!).
Reinstall it
I noticed a double to trippel speed increase from 10.1 to 10.2 after i had to reinstall the system totally. Everything from this point on was blazing fast. Even if i started to install more and more. However i agree you/me have to add 1 GB to 1.5 now. It showes that with more graphic app running it slowes down a bit. Not much but you see it.
Dave
I noticed a double to trippel speed increase from 10.1 to 10.2 after i had to reinstall the system totally. Everything from this point on was blazing fast. Even if i started to install more and more. However i agree you/me have to add 1 GB to 1.5 now. It showes that with more graphic app running it slowes down a bit. Not much but you see it.
Dave
Re: Jaguar Speed Tips
Bad idea. X optimize is redundant in X.2, the people working on Chimera already took out a piece of code that did something similar because they were worried it might interact strangely with the X.2 code.
Bad idea. X optimize is redundant in X.2, the people working on Chimera already took out a piece of code that did something similar because they were worried it might interact strangely with the X.2 code.
Buying more RAM is always a good idea.
I've had some bad experinces with the Acard133, which is the same as the Tempo133. My card gave me intermittent errors all the time that sometimes led to having to reformat my drive. Eventually, my card completely died, and now I have to replace it. It's <1 year old.
RAID 0 leaves no room for error. It should be called AID, because there is no redundancy. If one of the drives or the controller fails, like mine did repetedly for me, then you're screwed.
The ATA RAID cards are 33MHz/32-bit PCI cards. So the max transfer rate that can go through the card is 133MB/sec. With 4 ATA drives, (the max the controller can handle) you probably wouldn't ever saturate the card with data.
I know there is a harware RAID card made by Sonnet/Acard. This would be a much better solution to the default software RAID provided by Apple's Disk Utility, since I heard the software implemenation is poor. I knmow they have X support, but I don't know about 10.2. You'd have to check the respective manufacture's websites.
I haven't heard too much about RAID in 10.2 yet, so you'd be trailblazing your way on that accord.
The only drives that have the 133 spec are made by Maxtor. I set up one of these drives, and the performance difference was minimal.
I would read up on RAID before you do anything. I was thinking about setting up some U160 RAID w/X, but decided against it. You can decide for yourself by doing some research. I would suggest reading up on RAID at Storage Review.
The general gist that I got was that the negatives outweigh the positives for RAID 0, especially for desktop applications. I would really recommend reading through the wholde RAID article if you're so inclined. It really taught me a LOT.
It arrives tomorrow morning now (online shopping is dangerous if you like impulse buying!), along with shiny doors.
That's bad. Not the sort of reliability anyone wants.
I don't know if it was intentional, but your use of the word "trailblazing" helps firm my opinions up quite a lot! I'll wait a few months and let other people experiment with this first!
Thanks.
I'd be interested to know if Shadowkiller actually really improves performance with Quartz Extreme enabled.
I upgraded to 10.2 and let Shadowkiller run but speedwise I noticed no difference between having it run and not having it run in 10.2.
Maybe someone with Quartz Extreme could run the Let 1k Windows Bloom with Shadowkiller running and without it running to see if it's really just a placebo for those of us with QE.
I am running a 700MHz iBook (16mb vram)
I upgraded to 10.2 and let Shadowkiller run but speedwise I noticed no difference between having it run and not having it run in 10.2.
Maybe someone with Quartz Extreme could run the Let 1k Windows Bloom with Shadowkiller running and without it running to see if it's really just a placebo for those of us with QE.
I am running a 700MHz iBook (16mb vram)