Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

G4fanboy

macrumors 6502
Original poster
Mar 9, 2013
319
198
Andalucia Spain
After a post about Leopard slowing performance with updates I got this Bible about Tiger on the garden

http://macintoshgarden.org/forum/tolerating-osx-starting-tiger

Eyoungren has achieved a system with more headroom (even).
"
I removed the AudioIPCDriver.kext and that seems to have been a great help. Normally my Quad idleas around 6-10% of CPU. Now, it's idling around 2-4%. It's very low now.

So, at least this has been a good benefit here. Going to try it on my other Macs."

Let us use this post for the Tiger diggings we can find thru that lead.
 

coconutshrimp

macrumors newbie
May 28, 2015
9
13
I tried out the tweaks he listed and i've noticed a marked improvement on my 10.4.11 Tiger install. Mainly on the UI/Video rendering side of things.

One thing i'm still not sure of though is if running Core Video framework 1.4.1 is an improvement on 10.4.11 or something I could do without (assuming the use of Quartz Core 1.4.9). His original post is kind of vague on that.

Otherwise some fascinating, and useful stuff on there.
 
  • Like
Reactions: G4fanboy

robertdsc

macrumors regular
Jan 28, 2014
202
9
Funny, I have run Tiger 10.4.11 on every PPC Mac I have ever owned and none of them had any issues whatsoever. Tinkering with the innards in this way seems kind of counterproductive to me.

Now Leopard, on the other hand, has been a horrific mess. I downgraded the few G5s I tried to use with it to safe Tiger installs. I refuse to use Leopard at all.
 

eyoungren

macrumors Penryn
Aug 31, 2011
28,796
26,886
Funny, I have run Tiger 10.4.11 on every PPC Mac I have ever owned and none of them had any issues whatsoever. Tinkering with the innards in this way seems kind of counterproductive to me.

Now Leopard, on the other hand, has been a horrific mess. I downgraded the few G5s I tried to use with it to safe Tiger installs. I refuse to use Leopard at all.
I have had the opposite experience on my work Macs. I refuse to use Tiger unless I absolutely have to. Tiger has serious issues with connecting to Windows servers and it's capabilities as a print server are problematic.

I am forced to use DAVE on the one work Mac still running Tiger. Without it, it's impossible to connect to any server share at all.

Leopard was much more stable. I didn't have many issues with Tiger on my home Macs, but because of all it's problems at work this is the reason I do not use it at home unless I must.
 
  • Like
Reactions: AphoticD

G4fanboy

macrumors 6502
Original poster
Mar 9, 2013
319
198
Andalucia Spain
The original post talks about a bug in AppleDVD in Tiger client vs Tiger Ser ver. Erik, Maybe your were using Tiger client before the last G3?
 

eyoungren

macrumors Penryn
Aug 31, 2011
28,796
26,886
The original post talks about a bug in AppleDVD in Tiger client vs Tiger Ser ver. Erik, Maybe your were using Tiger client before the last G3?
I started using Tiger some time in 2007 I think. I didn't update to Leopard until early 2010.

I've got no real problems with Tiger on my own Macs (I'm running Tiger Server on my PM G3) but it had some serious issues at work.

Coincidentally, only now does Thursby software believe that Apple has fully implemented a proper SMB/CIFs stack in High Sierra and have discontinued their product. So, accounting for a few years with OS9 it took Apple 20 years to finally get this done - while everyone else was already there.

Leopard also affords me the luxury of using the latest apps my Macs will still run.

I won't deny that a basic setup of Tiger is faster than Leopard. And probably stll faster than my optimized versions of Leopard. But I guess this is where preference comes in. While I value speed, I do not value it as much as others here seem to. I value usability.

Being on Tiger at work was one of the issues that prevented me from using QuarkXPress 8 and Adobe CS4 does not run on Tiger. Don't believe Office 2008 works on Tiger either.

That said, I'm all for optimizing Tiger. I'm still using it on my G3 as I said. My only problem is that my G3 is headless so restarting it to affect changes is a hassle.
 
  • Like
Reactions: G4fanboy

swamprock

macrumors 65816
Aug 2, 2015
1,206
1,760
Michigan
I have had the opposite experience on my work Macs. I refuse to use Tiger unless I absolutely have to. Tiger has serious issues with connecting to Windows servers and it's capabilities as a print server are problematic.

I am forced to use DAVE on the one work Mac still running Tiger. Without it, it's impossible to connect to any server share at all.

Leopard was much more stable. I didn't have many issues with Tiger on my home Macs, but because of all it's problems at work this is the reason I do not use it at home unless I must.

Have you looked into Sharepoints instead of DAVE for connecting to server shares?

Office 2008 runs on Tiger just fine, as long as it's a G4 running 10.4.9 or higher.

EDIT: never mind. It runs on Tiger to give the OS similar network sharing that Leopard gives.
 

eyoungren

macrumors Penryn
Aug 31, 2011
28,796
26,886
Have you looked into Sharepoints instead of DAVE for connecting to server shares?

EDIT: never mind. It runs on Tiger to give the OS similar network sharing that Leopard gives.
I actually tried Sharepoints at one time (on Tiger), but couldn't figure out how to make it work.

I guess my biggest disappointment was all the stuff I had to do on the server side in order to allow Tiger Macs to connect. Even then…if a server share dropped or I made more than two file copy requests Finder beachballed. If I was lucky it just restarted. If not, I had to reboot the Mac.

Then there was the dealing with AppleDouble files (the "." files). BlueHarvest helped, but moving to Leopard was what really stopped the litter on the server.
 

swamprock

macrumors 65816
Aug 2, 2015
1,206
1,760
Michigan
I actually tried Sharepoints at one time (on Tiger), but couldn't figure out how to make it work.

I guess my biggest disappointment was all the stuff I had to do on the server side in order to allow Tiger Macs to connect. Even then…if a server share dropped or I made more than two file copy requests Finder beachballed. If I was lucky it just restarted. If not, I had to reboot the Mac.

Then there was the dealing with AppleDouble files (the "." files). BlueHarvest helped, but moving to Leopard was what really stopped the litter on the server.

All good and fair reasoning.

Regarding the beachballing, unlockupd will (mostly) fix that, even on normal systems that have to access network shares from Tiger- http://bafford.com/software/unlockupd/
 

eyoungren

macrumors Penryn
Aug 31, 2011
28,796
26,886
All good and fair reasoning.

Regarding the beachballing, unlockupd will (mostly) fix that, even on normal systems that have to access network shares from Tiger- http://bafford.com/software/unlockupd/
Thanks for that. Will see about installing it on my server at home tonight. Haven't had any issues but any bug patches are good.

This would perfectly explain my problem so many years ago though. Should have known it was some sort of bug!
 
Last edited:
  • Like
Reactions: swamprock

G4fanboy

macrumors 6502
Original poster
Mar 9, 2013
319
198
Andalucia Spain
Erik, what CAN do Photoshop CS4 that cant be done on Tiger´s PS CS2 or Mac OS 9 Photoshop 7?

The same Question for X'Press vs Old School Freehand/Corel

I am a musician, not a Image Professional as you. I know what Mac OS 9 Protools vs Mac OS X Protools differences are. 90-95% can be done with the older version.
[doublepost=1522780819][/doublepost]"Tiger CoreVideo.framework v1.4.0 comes with installing QuickTime v7.1.6 and/or OSX 10.4.9
Tiger CoreVideo.framework v1.4.1 (The Best) comes only with installing QuickTime v7.2.0
Tiger CoreVideo.framework v1.4.2 comes with all QuickTime for Tiger newer than v7.2"
 

eyoungren

macrumors Penryn
Aug 31, 2011
28,796
26,886
Erik, what CAN do Photoshop CS4 that cant be done on Tiger´s PS CS2 or Mac OS 9 Photoshop 7?

The same Question for X'Press vs Old School Freehand/Corel

I am a musician, not a Image Professional as you. I know what Mac OS 9 Protools vs Mac OS X Protools differences are. 90-95% can be done with the older version.
[doublepost=1522780819][/doublepost]"Tiger CoreVideo.framework v1.4.0 comes with installing QuickTime v7.1.6 and/or OSX 10.4.9
Tiger CoreVideo.framework v1.4.1 (The Best) comes only with installing QuickTime v7.2.0
Tiger CoreVideo.framework v1.4.2 comes with all QuickTime for Tiger newer than v7.2"
Standards.

I need to be able to open the things our customers send us and I need to be able to send our printer what they require of us. The older a product gets the less likely I am to be consistent with those standards.
 

AphoticD

macrumors 68020
Feb 17, 2017
2,282
3,459
On my Pismo G3 400Mhz, running 10.4.11, I experimented with disabling the AudioIPCDriver.kext and found CPU idle dropped from 1-2% to 0-1% (Longest Battery Life). Audio still works fine.

It would be interesting to know what this extension is doing exactly.

I have MenuMeters reporting the CPU %. It’s welcoming to see it idle on zero! I’ll try it on a few other PowerPCs next.

I’ll play with the various graphics frameworks too. It would be nice to see a more fluid Tiger UI on the Pismo, but I don’t expect Quartz Extreme performance from it.
 
  • Like
Reactions: eyoungren

eyoungren

macrumors Penryn
Aug 31, 2011
28,796
26,886
On my Pismo G3 400Mhz, running 10.4.11, I experimented with disabling the AudioIPCDriver.kext and found CPU idle dropped from 1-2% to 0-1% (Longest Battery Life). Audio still works fine.

It would be interesting to know what this extension is doing exactly.

I have MenuMeters reporting the CPU %. It’s welcoming to see it idle on zero! I’ll try it on a few other PowerPCs next.

I’ll play with the various graphics frameworks too. It would be nice to see a more fluid Tiger UI on the Pismo, but I don’t expect Quartz Extreme performance from it.
Yeah, this is what I noted both on my Quad running 10.5.8 and my 17" PB running 10.5.8.

It's also reduced on my 17" MBP running 10.6.8.
 

G4fanboy

macrumors 6502
Original poster
Mar 9, 2013
319
198
Andalucia Spain
"AudioIPCDriver.kext - this extension first appeared in OSX 10.4.4 and slows the computer down a lot, and I'm yet to find a use for having it. This might be a trick Apple used in order to make Intel look superior to PowerPC. 10.4.4 is the first OSX for Intel and at the same time, this here useless extension which severely handicaps PPC (and Intel a little) also appears. I suspect erasing /System/Library/Extensions/AudioIPCDriver.kext (+rebooting) is the same thing as what some others are calling "disabling BeamSync" on another mac internet site. I don't know the specifics of BeamSync, suspecting I don't need to."

From the 1st post

Could it be the same Beamsync trick?

From post 33:
"So ignore what I said about Tiger v10.4.9 and I return to my original opinion that with Tiger, in is run with either v10.4.8 (the most stable and fastest without CoreVideo) or v10.4.11 (in which there is a crash-on-DVD worm bug in 10.4.9 thru 10.4.11 Client PPC). 10.4.9-.10 are trouble, more trouble. Now I have Server PPC 10.4.11 I am happy with, Server is a lot slower than Client, for like killer gaming. I'll continue hunting for the OS module causing the crash-on-DVD. Then I can report on the most awesome 10.4.11 volume set of changes."
[doublepost=1522802368][/doublepost]https://www.tonymacx86.com/threads/audio-driver-889a-audioipcdriver.6166/

Could it be sound related?
 

AphoticD

macrumors 68020
Feb 17, 2017
2,282
3,459
To contribute, here's a general UI speed tweak I perform on my PowerPC Macs;
  1. Install Onyx (v1.8.6 for Tiger, v2.0.6 for Leopard)
  2. Go to the Parameters section.
  3. In the General tab, set Speed of display of sheets: to Fast.
  4. In the Finder tab, uncheck Graphic Effects: Application Zooming, Window Zooming and Animate Info Window.
  5. In the Dock tab, set Minimize Using: to Scale Effect.
  6. Uncheck Animate / Graphic Effects: Opening applications.
Although these will only shave off micro-seconds, it just feels more responsive IMO.
 

AphoticD

macrumors 68020
Feb 17, 2017
2,282
3,459
After some further digging around, I came across an Ars poster with very insightful information on what they refer to as "Memory tax" - https://arstechnica.com/civis/viewtopic.php?f=19&t=1329095

This reads just like our friend SkyCapt. I think it's a safe assumption to make that "whichkraft" is the same person. Definitely worth a read.

I'm downloading the Mac OS X 10.5.5 Combo Update to extract the older CoreVideo.framework and will try disabling AudioIPCDriver.kext on a Leopard Mac next.
 

AphoticD

macrumors 68020
Feb 17, 2017
2,282
3,459
I'm downloading the Mac OS X 10.5.5 Combo Update to extract the older CoreVideo.framework and will try disabling AudioIPCDriver.kext on a Leopard Mac next.

Ok, so I downloaded 10.5.1 Update, 10.5.2 Combo all through to 10.5.5 Combo and poked around with Pacifist. None of them had the CoreVideo.framework. It turns out that CoreVideo.framework v1.5.0 was from 10.5 (fresh), so I created another partition and did a clean install of 10.5, then updated this to 10.5.5 with the combo.

(CoreVideo.framework 1.6.0 is installed during 10.5.6 and remains unchanged to 10.5.8).

After all of that, my findings were the opposite. CoreVideo.framework 1.6.0 delivers higher Quartz Extreme frame rates than v1.5.0 in both the WindowManager and the Dock (when magnification is enabled).

Here are some FPS readings (Using the Xcode 3.1.4 > Performance > Quartz Debug > Frame gauge)

PowerBook G4 Late 2005 DLSD 15” 1.67Ghz / 2GB DDR2 / ATI Mobility Radeon 9700 128MB

10.5.8 with CoreVideo 1.5.0
BeamSync: Auto
2D Dock Magnification:~28 - 32 fps
WindowManager movement:~34 - 40fps

10.5.8 with CoreVideo 1.5.0
BeamSync: DISABLED
2D Dock Magnification:~36 - 42 fps
WindowManager movement:~55 - 58fps

10.5.8 with CoreVideo 1.6.0
BeamSync: Auto
2D Dock Magnification:~40 - 45 fps
WindowManager movement:~58 - 60fps

10.5.8 with CoreVideo 1.6.0
BeamSync: DISABLED
2D Dock Magnification:~45 - 55 fps
WindowManager movement:~60 - 62fps


The short answer is: CoreVideo.framework v1.6.0 introduced in 10.5.6 IMPROVES the UI frame rates *substantially* over v1.5.0 when BeamSync is set to Automatic (default).

There were also reasonable improvements found in v1.6.0 when BeamSync was disabled, however as expected, window tearing is evident when moving things around quickly and although the frame rate is higher, it makes the UI appear a little rougher. The Dock magnification didn’t appear to be affected by tearing in the same way window movement is.

In these tests, I had already disabled AudioIPCDriver.kext and set the Dock to be 2D (Black) via Onyx.

Dock-FPS-BeamSyncDisabled.jpg



So now I’m confused, wasn’t the exercise to improve performance by trying older (pre-bloat) components?

Kudos to the CoreVideo / QuartzCore team for their optimizations. It’s funny, when you disable the 3D, reflective Dock in Leopard, the whole UI responsiveness improves. So I’ll add that one to the official list of Onyx based speed tune ups.

For the more curious: CoreVideo.framework was introduced in QuickTime 7 (requiring Panther 10.3.9 or Tiger 10.4) as a means of synchronizing OpenGL buffer flushing (drawing) with the display hardware vertical refresh rate. In my (limited) Cocoa programming experience, I have wrangled the CVDisplayLink into submission for several subclassed NSOpenGLView rendering contexts. The CVDisplayLink service is at the heart of CoreVideo.framework and works as a finely tuned redraw timer. My understanding is that this was introduced with QT7 because at the time, the QuickTime team had rewritten the video rendering engine, dropping the Carbon cornerstone QuickDraw in favour of a ground-up Cocoa engine (with many Carbon dependencies I’m sure). The new rendering engine used GPU accelerated OpenGL based video playback. The major advantage being the GPU Hardware was able to take the video rendering load off the CPU almost entirely, allowing the CPU to handle the codec/decoding. Various GPUs also introduced onboard decoders, such as MPEG2 and h264, further reducing the CPU load. QT7 was the first version to support h264 playback.
 

dbdjre0143

macrumors 6502
Nov 11, 2017
361
382
West Virginia
After some further digging around, I came across an Ars poster with very insightful information on what they refer to as "Memory tax" - https://arstechnica.com/civis/viewtopic.php?f=19&t=1329095

Wow, I just read through that thread and I feel like I'm trying to decipher some deeply concealed state secret more than speed up my computer. Sounds like there might be something there but it would take some more coffee to figure out exactly what, and that's coming from a software engineer with a Computer Science degree :p.
 

AphoticD

macrumors 68020
Feb 17, 2017
2,282
3,459
Skycapt speaks about Tiger and the best tiger posible. He talks about other Leopard file for the DVD malware bug

I understand we started with Tiger here, so I was following the advice from the Ars poster "whichkraft" regarding Leopard:
OSX 10.5 Leopard won't let you remove CoreVideo. Up to 10.5.5 it's just QT and AudioIPCDriver.kext to deal with, but then, 10.5.6 thru 1.5.8 are a lot slower than 10.5.5 because it's CoreVideo's fault again. Big speed improvement when replacing 10.5.8's CoreVideo.framework with the version stolen from out of OSX 10.5.5

As I discovered, on my specific hardware, the result was the newer CoreVideo.framework in 10.5.6 - 10.5.8 outperforms the older version quite a bit. I will have to try CoreVideo 1.4.1 on Tiger next, but personally, I've never really found Tiger to lag or skip on a G4 with a Quartz Extreme capable GPU, much at all. I know Panther and Jaguar feel quicker, but in my own testing with Quartz/CoreGraphics rendering and testing the advances made in Quartz Extreme performance (which is GPU based, OpenGL window compositing), I have found that Tiger performs both Quartz drawing and Quartz Extreme rendering the FASTEST. Outperforming both Panther and Jaguar, even at the (implied) "crippled" state of 10.4.11.

The poster on Ars copped a lot of criticism, but they believe in what they have discovered. So, I won't claim it to be untrue, except for my findings with Leopard so far.
 
  • Like
Reactions: G4fanboy

AphoticD

macrumors 68020
Feb 17, 2017
2,282
3,459
I tried out the tweaks he listed and i've noticed a marked improvement on my 10.4.11 Tiger install. Mainly on the UI/Video rendering side of things.

One thing i'm still not sure of though is if running Core Video framework 1.4.1 is an improvement on 10.4.11 or something I could do without (assuming the use of Quartz Core 1.4.9). His original post is kind of vague on that.

Otherwise some fascinating, and useful stuff on there.

Did CoreVideo 1.4.1 on 10.4.11 make for any performance improvement? Are you able to check FPS readings with Quartz Debug?

What’s interesting is that SkyCapt / whichkraft claims performance increases when CoreVideo.framework is removed/disabled. When a Cocoa programmer elects to use CoreVideo for managing screen redraws and CV is not available, a manually implemented fallback of using a 1/60 (or even 1/110) sec Interval timer must be written in to manually trigger a view to redraw. In Tiger and above, CVDisplayLink was defined as the preferred “frame for time” redraw method. In the Cocoa documentation, it’s stated that the NSTimer method should only be implemented when writing for Panther compatibility and is less performant than the CoreVideo technique because there is no standard method to synchronize view redraws to the hardware. Effectively this will drop frames which fall out of sync as well as occasionally rendering more frames than can be displayed.

The poster mentions CoreVideo.framework appeared on the scene at 10.4.8. I believe this is due to “CoreVideo” being bundled into QuartzCore.framework in 10.4 - 10.4.7.

Although I am of course a big advocate of squeezing more performance out of older hardware, I also believe in proving results beyond “it looks better”. The poster hasn’t proven any results of his findings across the board.

Surely if these software mods “double” the performance as claimed, a
YouTube video would be easy enough to shoot as evidence to prove this.

I am genuinely interested in making performance increases work for us. I’d be happy to wrap a series of “tune ups” into a stand-alone Cocoa app for a one-click (or automatic on Login) solution to boost our aging Macs.

My short list of Onyx animation tweaks can be easily wrapped into a simple app, as can BeamSync enable/disable... if it’s proven that a specific older framework outperforms (and retains compatibility) over the last revisions in 10.4.11 and 10.5.8, then these could be easily packaged into a “tune up” installer app too.

But there’s no point putting time into something like this if all we have to go by is based on a single user’s perception of image quality (without any evidence).
 

Dronecatcher

macrumors 603
Jun 17, 2014
5,209
7,783
Lincolnshire, UK
I often feel some of these tweaks are a perceptual placebo...almost self-congratulatory - you've made the effort so it's got to work. Without some kind of benchmark it's all speculation.
The Onyx tweaks vertainly do speed up the GUI as does ShadowKiller for sure but I've never found beamsync and Quartz Extreme tweaks to make a difference.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.