Become a MacRumors Supporter for $25/year with no ads, private forums, and more!

AquaTrimcelerator - The Unofficial 10.4.12

z970mp

macrumors 68040
Original poster
Jun 2, 2017
3,135
3,260
The Matrix
I'll try this out soon, but first, what breaks with iChat? I use it for Bonjour and Jabber with my Mac OS X Server around the house.

It will fail to launch.

All references to VideoConference.framework in the script should be removed if this is undesired.
 
  • Like
Reactions: Slix

Raging Dufus

macrumors 6502
Aug 2, 2018
449
743
Kansas USA
So, I gave TenFourTrim a shot on my 12" PowerBook G4. Ultimately, I decided it wasn't for me, for two reasons:
  1. A purely aesthetic consideration: Although I used the version that retains Dashboard, it apparently had removed the font that Dashboard used (whichever font that was, I don't know). I didn't like my widgets' appearance with the new font, which at first glance looks like some variety of Times.
  2. A functional consideration: Initial benchmarking showed ~25% drop in AltiVec performance.
Here are the Xbench scores taken about 15 minutes after completion of TenFourTrim:

G4 Post-TFT.jpg

As you can see, the overall effect of TFT was positive, though negligible, after so short a period of time. I did take into account @z970mp's suggestion that using your Mac for some hours/days will build system caches to further improve performance; however, I don't see how that could improve AltiVec performance.

I should note I also tested TFT on my 1.0 GHz TiBook, and overall changes were very similar to what you see above. However, the drop in AltiVec performance on the TiBook was less pronounced: ~10%. I can only attribute this to the different CPU's: the TiBook's 7455 vs. the AlBook's 7447A. Again, this is all very preliminary, after only one round of benchmarking. It's all I plan to do though, so there's the info for anyone who wants to pick up that ball and run with it.

I can see how TFT could benefit a G3 system. G3's don't have AltiVec, so would suffer nothing by applying TFT, and performance gains in the UI would be noticeable and would accumulate over time. Memory and graphics performance were only negligibly impacted in my tests. Changes in disk performance on my TiBook with a 7200 RPM spinner were likewise negligible (you should probably disregard the disk results above from my 12", which has a SSD). I can see how the build-up of system and app caches over time could benefit these as well.

Here's my personal takeaway, based solely on the very limited experience described above: If I had a G3 system, I'd use TFT. On a low-end G4 system, I might use TFT; assuming the impact on AltiVec was small, a low-end G4 might stand to gain more than it loses. If I had a G5, I'd give it a try; I'm guessing the pure power of a G5 would negate any impact on AltiVec.

But for my mid- to high-end G4's, I won't be using TFT. I'd prefer to apply TFT's changes a la carte, only using the ones that don't inhibit my G4's performance (whichever those are, and I don't know...needs further research).

Not to take anything away from @z970mp's work, which is fantastic and I think achieved his goal of optimizing Tiger for G3's. Kudos!
 
  • Like
Reactions: Jubadub and z970mp

z970mp

macrumors 68040
Original poster
Jun 2, 2017
3,135
3,260
The Matrix
So, I gave TenFourTrim a shot on my 12" PowerBook G4. Ultimately, I decided it wasn't for me, for two reasons:
  1. A purely aesthetic consideration: Although I used the version that retains Dashboard, it apparently had removed the font that Dashboard used (whichever font that was, I don't know). I didn't like my widgets' appearance with the new font, which at first glance looks like some variety of Times.
  2. A functional consideration: Initial benchmarking showed ~25% drop in AltiVec performance.
Here are the Xbench scores taken about 15 minutes after completion of TenFourTrim:

View attachment 851772

As you can see, the overall effect of TFT was positive, though negligible, after so short a period of time. I did take into account @z970mp's suggestion that using your Mac for some hours/days will build system caches to further improve performance; however, I don't see how that could improve AltiVec performance.

I should note I also tested TFT on my 1.0 GHz TiBook, and overall changes were very similar to what you see above. However, the drop in AltiVec performance on the TiBook was less pronounced: ~10%. I can only attribute this to the different CPU's: the TiBook's 7455 vs. the AlBook's 7447A. Again, this is all very preliminary, after only one round of benchmarking. It's all I plan to do though, so there's the info for anyone who wants to pick up that ball and run with it.

I can see how TFT could benefit a G3 system. G3's don't have AltiVec, so would suffer nothing by applying TFT, and performance gains in the UI would be noticeable and would accumulate over time. Memory and graphics performance were only negligibly impacted in my tests. Changes in disk performance on my TiBook with a 7200 RPM spinner were likewise negligible (you should probably disregard the disk results above from my 12", which has a SSD). I can see how the build-up of system and app caches over time could benefit these as well.

Here's my personal takeaway, based solely on the very limited experience described above: If I had a G3 system, I'd use TFT. On a low-end G4 system, I might use TFT; assuming the impact on AltiVec was small, a low-end G4 might stand to gain more than it loses. If I had a G5, I'd give it a try; I'm guessing the pure power of a G5 would negate any impact on AltiVec.

But for my mid- to high-end G4's, I won't be using TFT. I'd prefer to apply TFT's changes a la carte, only using the ones that don't inhibit my G4's performance (whichever those are, and I don't know...needs further research).

Not to take anything away from @z970mp's work, which is fantastic and I think achieved his goal of optimizing Tiger for G3's. Kudos!

Interesting. I don't believe I've seen those drops in CPU performance on my tests. Perhaps it has varying results for different platforms and configurations?

Remember that it slims the system's components to the minimum stable threshold, not necessarily the recommended or best balanced threshold, where there's just enough remaining to keep a stable, functional system. In that sense, it would indeed be best for G3s and other low end machines just powerful enough to run Tiger.

Perhaps the greatest takeaway from all this is that your mileage will greatly vary...
 

Raging Dufus

macrumors 6502
Aug 2, 2018
449
743
Kansas USA
Interesting. I don't believe I've seen those drops in CPU performance on my tests. Perhaps it has varying results for different platforms and configurations?

Remember that it slims the system's components to the minimum stable threshold, not necessarily the recommended or best balanced threshold, where there's just enough remaining to keep a stable, functional system. In that sense, it would indeed be best for G3s and other low end machines just powerful enough to run Tiger.

Perhaps the greatest takeaway from all this is that your mileage will greatly vary...

Agreed. Tiger covers a far greater variety of hardware than any other version of Mac OS, so it would be difficult indeed to craft a solution that gave everybody a boost. I think you've hit the mark for G3's though.

One thing I should note for others: If you're going to do benchmarking after applying TFT, make sure you give Spotlight a chance to do its thing first. These changes will cause Spotlight to re-index your drive(s) at the first opportunity, you'll need to wait til it's finished before doing any testing. I didn't take this into account the first time I benchmarked, and was shocked at the appalling score!
 
  • Like
Reactions: z970mp

mmphosis

macrumors regular
Jan 3, 2017
116
123
I thought TenFourTrim would turn off SpotLight which is one of the first things I do.
Code:
cd /System/Library/
mv LaunchAgents/com.apple.Spotlight.plist LaunchDaemons/com.apple.metadata.mds.plist /Users/myuser/Desktop/
..and, reboot.
 
  • Like
Reactions: Raging Dufus

Raging Dufus

macrumors 6502
Aug 2, 2018
449
743
Kansas USA
I thought TenFourTrim would turn off SpotLight which is one of the first things I do.
Code:
cd /System/Library/
mv LaunchAgents/com.apple.Spotlight.plist LaunchDaemons/com.apple.metadata.mds.plist /Users/myuser/Desktop/
..and, reboot.

I agree with you, as far as Tiger goes. Spotlight on Tiger isn't all that useful, and there are better and less resource-intensive options like EasyFind.

Leopard is a different matter. Spotlight was greatly improved with Leopard and better integrated into the system. For those interested, this article explains the differences well: https://tidbits.com/2007/11/01/spotlight-strikes-back-in-leopard-it-works-great/
 
  • Like
Reactions: mmphosis and Slix

Jubadub

macrumors regular
Nov 1, 2017
198
232
I can see how TFT could benefit a G3 system. G3's don't have AltiVec, so would suffer nothing by applying TFT, and performance gains in the UI would be noticeable and would accumulate over time. Memory and graphics performance were only negligibly impacted in my tests. Changes in disk performance on my TiBook with a 7200 RPM spinner were likewise negligible (you should probably disregard the disk results above from my 12", which has a SSD). I can see how the build-up of system and app caches over time could benefit these as well.
Sounds like the scripts could be renamed from "TenFourTrim" to "Ten4G3". As in, "10.4, for G3", intended for G3 and earlier.
 
  • Like
Reactions: 1042686

Raging Dufus

macrumors 6502
Aug 2, 2018
449
743
Kansas USA
Sounds like the scripts could be renamed from "TenFourTrim" to "Ten4G3". As in, "10.4, for G3", intended for G3 and earlier.

I dunno, maybe. You can't read too much into my one round of testing, there's just too many variables that remain to be explored. Is the AltiVec hit on all 7447A's or just the one in my 12" PB? Why did it have less effect on the 7455 in my TiBook? Will the effect increase/decrease with processor speed? What about 7400's, 7410's, 7457's, etc., etc.?

And if you look at my results, the serious hit was only to AltiVec. Overall CPU performance only dropped an insignificant amount, and given the speed gains in the UI, you'd never notice it unless using an app that was optimized for AltiVec. Given that Tiger officially supported a ton of G3's for its entire run, there shouldn't be a problem with any program that doesn't specify a G4 as minimum.

It's just too early to tell what TFT could be. More testing needs to be done, I hope more people will post results here.
 
  • Like
Reactions: z970mp

Jubadub

macrumors regular
Nov 1, 2017
198
232
And if you look at my results, the serious hit was only to AltiVec. Overall CPU performance only dropped an insignificant amount, and given the speed gains in the UI, you'd never notice it unless using an app that was optimized for AltiVec. Given that Tiger officially supported a ton of G3's for its entire run, there shouldn't be a problem with any program that doesn't specify a G4 as minimum.
I believe many Tiger-compatible games are G4/G5-only, and that's a concern with AltiVec in mind, which benefit from it. Also browsers like TenFourFox are optimized to use AltiVec as much as possible, with a specialized version for each processor. Those are 2 major hits and big "nono"s in my opinion. Surely there's more, but that's what comes immediately to mind.

But, of course, as you say, more real-world testing is required for a more informed conclusion. A lot could have affected your benchmarks and whatnot, so it could be anything. No G5 benchmarking yet, which would also be interesting to see.

In any case, whether or not performance loss is confirmed to be a fact from TFT in some scenarios, the goal of TFT is definitely on the right track, which is to remove as much "less useful" stuff as possible to maximize gain in performance and space (both storage and RAM), and it's good someone was pursuing that goal. So even if it doesn't work now, it can be fixed, and work later.
 

Raging Dufus

macrumors 6502
Aug 2, 2018
449
743
Kansas USA
I believe many Tiger-compatible games are G4/G5-only, and that's a concern with AltiVec in mind, which benefit from it.

That's true, games are a good example.

Also browsers like TenFourFox are optimized to use AltiVec as much as possible, with a specialized version for each processor.

In TenFourFox's case, the cure would be to use the G3 version, which to my knowledge isn't really optimized for G3's, it just lacks the optimizations included in the two G4 versions and the G5 version. Many people prefer it over either G4 version, even on G4's.
 
  • Like
Reactions: Jubadub

Jubadub

macrumors regular
Nov 1, 2017
198
232
Many people prefer it over either G4 version, even on G4's.
Really? That's odd... That really shouldn't be a thing. Indeed, the G3 version is actually just a "global" version, so it works on any PPC architecture (probably even pre-G3, if Tiger is installed with XPostFacto). But the G4/G4e/G5 versions should come on top of the G3 version in every aspect, if run on the corresponding architecture. I really find it very unbelievable those reports aren't somehow mistaken, but that got me a bit curious, because it's not like I tried it myself. I guess I'll grab the G3 version for my Mac mini "G4e" and Quad G5, then calmly observe over time how they fare side-by-side. Though that seems difficult to benchmark with good precision.
 

Raging Dufus

macrumors 6502
Aug 2, 2018
449
743
Kansas USA
Really? That's odd... That really shouldn't be a thing. Indeed, the G3 version is actually just a "global" version, so it works on any PPC architecture (probably even pre-G3, if Tiger is installed with XPostFacto). But the G4/G4e/G5 versions should come on top of the G3 version in every aspect, if run on the corresponding architecture. I really find it very unbelievable those reports aren't somehow mistaken, but that got me a bit curious, because it's not like I tried it myself. I guess I'll grab the G3 version for my Mac mini "G4e" and Quad G5, then calmly observe over time how they fare side-by-side. Though that seems difficult to benchmark with good precision.

LowEndMac article from when TenFourFox was introduced:


From the referenced article:

"The G3 version will run on any supported Macintosh, but it will not take advantage of the additional features of G4 or G5 processors, especially the AltiVec velocity engine. The G4/G5 version will not run correctly on a G3 Mac."
I don't think any of that has changed. I know, too, that for a while before @wicknix picked up the ball w/Arctic Fox and before the Intel version of TFF was revived last year, one of the best options Snow Leopard users had was to run TFF/G3 in Rosetta. I never tried it, but it's my understanding that only the G3 build would run under Rosetta, simply because it lacked the G4/G5 optimizations.
 

AL1630

macrumors 6502
Apr 24, 2016
374
335
Idaho, USA
I tested it out on my iMac G3 (400MHz, 512MB RAM), and unfortunately I didn't seem to have any major change. Disabling spotlight seems to have made a difference though. The specific configuration you run the tweaks on must be a major factor in how big of a difference it makes. Attached are my xbench results.
 

Attachments

  • Before.txt
    1.7 KB · Views: 75
  • After.txt
    1.7 KB · Views: 75
  • After Tweaks + No Spotlight.txt
    1.7 KB · Views: 74

swamprock

macrumors 6502a
Aug 2, 2015
986
1,321
Michigan
I have already disabled Spotlight and Dashboard on my 800mhz iBook G3 using Disable Tiger Features, so running the script just removed fonts, apps, preferences, and frameworks that are no longer useful. I didn't notice any change in responsiveness, but I think that Spotlight and Dashboard are probably the biggest resource hogs anyway and they were already disabled.
 

z970mp

macrumors 68040
Original poster
Jun 2, 2017
3,135
3,260
The Matrix
I must be the only one who's seen a definite positive improvement, on multiple machines no less...
 

Raging Dufus

macrumors 6502
Aug 2, 2018
449
743
Kansas USA
I had occasion to wipe my DLSD PB's drive, so I decided to install Tiger and give TenFourTrim a go. See Xbench results below:

After a fresh install of Tiger and applying all updates to 10.4.11. No optimizations of any kind. For disk results, bear in mind this PowerBook has a SSD, however it was freshly reformatted making garbage collection a non-issue.

Before applying TFT on the left; 10 minutes after TFT on the right.

1.67 GHz Pre & Post.jpg

I then used the PowerBook normally for a while, downloading & installing various apps, creating documents, watching videos, listening to music, etc. etc. Below are the results after about 4 hours of such use:

1.67 GHz Post4.jpg
Notes to @z970mp:
  1. The above were performed using the initial version of TFT, which was supposed to disable Dashboard.
  2. I watched the process in Terminal, and although it said it was disabling Dashboard, it didn't actually do so. Dashboard still functions after applying TFT, albeit with a different font:
    • Dashboard Post-TFT.jpg
  3. I'm not sure Xbench is the best metric for measuring TFT's results.
 

z970mp

macrumors 68040
Original poster
Jun 2, 2017
3,135
3,260
The Matrix
Notes to @z970mp:
  1. The above were performed using the initial version of TFT, which was supposed to disable Dashboard.
  2. I watched the process in Terminal, and although it said it was disabling Dashboard, it didn't actually do so. Dashboard still functions after applying TFT, albeit with a different font:
  3. I'm not sure Xbench is the best metric for measuring TFT's results.

I've noticed this too, though I was hoping the same wouldn't apply to other machines. It's strange because it was using the command that was supposed to disable Dashboard, and that apparently worked across "all" versions of OS X.

I don't know how else it could be removed via command, so maybe I'll just replace it with the Minus Dashboard version...

As to the font, I'll see what I can do.
 
  • Like
Reactions: Raging Dufus

Jubadub

macrumors regular
Nov 1, 2017
198
232
I tried it on an iMac G3 and I would say the results are positive. Not major but it does make a difference in opening some apps.
By some apps, do you mean TenFourFox? It is known to open faster on lower-end configurations (such as lack of an SSD) the less fonts are installed, and the same applies to some other programs. If so, you can likely achieve the same effect by removing fonts manually, without removing or altering anything "extra". Although it is still possible TFT might have brought some other undocumented benefit to you, even if it isn't documented yet.
 

ScreenSavers

macrumors 68000
Feb 26, 2016
1,788
1,096
Bloomingdale, GA
By some apps, do you mean TenFourFox? It is known to open faster on lower-end configurations (such as lack of an SSD) the less fonts are installed, and the same applies to some other programs. If so, you can likely achieve the same effect by removing fonts manually, without removing or altering anything "extra". Although it is still possible TFT might have brought some other undocumented benefit to you, even if it isn't documented yet.

Tenfourfox is hardly useable on a G3 either way.
 

swamprock

macrumors 6502a
Aug 2, 2015
986
1,321
Michigan
Tenfourfox is hardly useable on a G3 either way.

I will say that the removal of the fonts by the script has sped up TFF G3 a noticeable bit. It's still a dog compared to Camino and Arctic Fox, but it's a tad more useful now.
[doublepost=1565674444][/doublepost]
I must be the only one who's seen a definite positive improvement, on multiple machines no less...

I like the fact that it removes old, outdated cruft. It's definitely not a useless utility by any means. I'm sure that had I not already disabled Spotlight and Dashboard that it would be more of a noticeable speed improvement on my end. To that end, I ran it on my new Mini G4 Tiger install and had to disable Dashboard separately (as has been covered earlier). I have no before or after benchmarks, but the machine runs admirably, so there's that...
 
  • Like
Reactions: z970mp

ScreenSavers

macrumors 68000
Feb 26, 2016
1,788
1,096
Bloomingdale, GA
Usually the first thing I do with a tiger install is:

“defaults write com.apple.dashboard mcx-disabled -boolean YES && killall Dock”

That seems to take care of the dashboard, which is especially annoying in 10.4 because it has an indicator in the dock that shows it’s constantly running
 

z970mp

macrumors 68040
Original poster
Jun 2, 2017
3,135
3,260
The Matrix
Usually the first thing I do with a tiger install is:

“defaults write com.apple.dashboard mcx-disabled -boolean YES && killall Dock”

That seems to take care of the dashboard, which is especially annoying in 10.4 because it has an indicator in the dock that shows it’s constantly running

Even with the killall Dock, it hasn't worked.

Is killall Dock required for immediate effect, or should it simply take effect the next time the Dock is relaunched?
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.