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

upgrading tiger to leopard ? think again

MacHamster68

MacHamster68

macrumors 68040
Original poster
Sep 17, 2009
3,251
5
Hello , i was really curious and did want to know if leopard is better as tiger
and it is not , it feels and looks more flashy , has some more features , but it slows things down
i have installed osx 10.4.11 only, no programs whatsoever
and after it leopard 10.5.8 without programs whatsoever and run geekbench

on my mac mini g4 1.42 1gb ram . 32mb vram 80gb harddrive

with tiger score is :872

with leopard the score is : 844

so that brings an end to the myth leopard would run better ;)

here the links to both results

tiger
http://browse.geekbench.ca/geekbench2/view/249571

leopard
http://browse.geekbench.ca/geekbench2/view/254249
 
A 28 point score difference in geekbench is very minimal. I have a heavily upgraded older G4 tower and I prefer Leopard. Tiger is a tad faster yes but not enough to make me keep using it. Leopard offers a lot more than just more eye candy over Tiger.

For me Leopard is more stable and it supports all forms of sleep (system, hard drive, display) where as Tiger does not on my system. Leopard also has much better security features.

I could sit here and list 100+reason why Leopard is better but do not have time right now. Maybe later. :)
 
Some time ago i made similar tests using Xbench (yeah, i know it's ****) and noticed that same with one of PM G5 i had. Score with 10.4.11 was much better, especially at CPU section.
Interesting is, that on Intel side 10.5 is much more efficient than 10.4 - my MP start running noticeably faster (in benchmarks it was visible too) after Tiger->Leo upgrade.
Seems to me then, that 10.5+ is Intel optimized OS, with added possibility to work on PPC ;)
BTW, with my all PPC Macs i use Tiger and certainly will not upgrade it to Leo.
 
666sheep said:
Some time ago i made similar tests using Xbench (yeah, i know it's ****) and noticed that same with one of PM G5 i had. Score with 10.4.11 was much better, especially at CPU section.
Interesting is, that on Intel side 10.5 is much more efficient than 10.4 - my MP start running noticeably faster (in benchmarks it was visible too) after Tiger->Leo upgrade.
Seems to me then, that 10.5+ is Intel optimized OS, with added possibility to work on PPC ;)
BTW, with my all PPC Macs i use Tiger and certainly will not upgrade it to Leo.
Click to expand...

xbench is the most inconsistent and sketchy benchmark ever on any os. I don't trust it's results at all as the last new version was in 06.

My mac is a true Frankenmac and has every reason to not run well on a newer OS yet Leopard runs much better all round.

The embedded technology in Leopard is much smarter and more secure. In my honest opinion (which is 16+ years of Mac experience) Leopard is the best OS I have ever used. The only plus Tiger has is that it is a tad faster (maybe 5-10%)on PowerPC hardware.
 
zen.state said:
xbench is the most inconsistent and sketchy benchmark ever on any os. I don't trust it's results at all as the last new version was in 06.
Click to expand...

Yes, i know this. I don't believe it's results when comparing graphics section, but i think, that CPU and HDD sections are at least good to compare old hardware before and after modifications (i was playing a lot with OC'ing of G4 Macs and RAID and i ran Xbench thousand times, so i know it's weaknesses). It was enough for that, and i think that it's scores are valid to compare the same hardware setup before and after "something".
Regardless of it, i used Geekbench too ofc, with similar results difference between OS. I got link for my MP only (10.4.11): http://browse.geekbench.ca/geekbench2/view/182405 With 10.5.8 i got about 5600 pts (but not submitted to results site).
 
you could be right there with leopard optimized for intel , i always thought about leopard as a stopgap for snow leopard , not a bad one , its a nice looking system and it works well , but it just does not run better on a ppc Mac
and after playing around a bit i noticed something else ...flashplayer 10 , youtube runs smoother in tiger then in leopard , with tiger i get about 79-92%processor usage while in leopard constant 100% and the fan starts spinning like hell right after clicking play while in tiger it takes a bit time before the fan starts kicking in , maybe thats where these 10-15% faster running will be noticeable most :eek::confused:
 
Optimized for intel? Well there are 2 versions.. PowerPC and Intel. Each one is optimized for each. I get equally good altivec performance in Leopard vs. Tiger. You guys see Leo as something that it really isn't.

If you guys want to use a 5+ year old OS that has not had an update since 07 go for it.

In my experiences Leopard is better for Intel and PowerPC systems. Many Intel people that upgraded to Snow Leopard are back on Leo. My girlfriend is one of them.
 
there is only one version if you buy the leopard retail dvd which installs on both ppc and intel only difference on a intel mac it installs bootcamp too
 
MacHamster68 said:
there is only one version if you buy the leopard retail dvd which installs on both ppc and intel only difference on a intel mac it installs bootcamp too
Click to expand...

Yes I understand both versions are on one disk. My point is that it is not simply an Intel optimized OS that was ported to PowerPC. My altivec benchmarks are just as high in Leopard as any OS X before it. That means the PowerPC code IS optimized for PowerPC.

Almost all Mac apps are written in xcode which simultaneously creates Intel and PowerPC optimized code. You guys can rely on blind facts but I will rely on real ones.
 
zen.state said:
Yes I understand both versions are on one disk. My point is that it is not simply an Intel optimized OS that was ported to PowerPC. My altivec benchmarks are just as high in Leopard as any OS X before it. That means the PowerPC code IS optimized for PowerPC.

Almost all Mac apps are written in xcode which simultaneously creates Intel and PowerPC optimized code. You guys can rely on blind facts but I will rely on real ones.
Click to expand...

btw i quiet like your sawtooth , nice paintjob ..well done looks really really nice
 
flopticalcube said:
The one area that Leopard outshines Tiger is in machines that have multiple cores or CPUs. This is probably why Intels see a big improvement but many PPC machines do not.
Click to expand...

I have a single G4 7448 @ 1.8GHz. The 7448 is the newest G4 chip and was just introduced in late 05. Very efficient CPU. I used to have a dual 1.42 MDD and my sawtooth with single 1.8 is pretty much the same speed all round in Leopard.

The 7448 is by far the fastest G4 chip ever. My single 7448 is faster than a dual 7447 1.8 in many ways.

When you think about it the 7448 chips are newer than all the G5 chips. Not the same kind of chip but my point is that maybe it's why my mac runs Leo so well. It's seriously silky smooth on Leo.

For the record.. the 7447 is what is in the mac mini and all the later consumer G4 desktops and the later powerbook/ibook. Decent CPU but it was well trumped power-wise by the 7448. The sad thing is that the G4 didn't get good till Apple stopped using it.
 
zen.state said:
Yes I understand both versions are on one disk. My point is that it is not simply an Intel optimized OS that was ported to PowerPC. My altivec benchmarks are just as high in Leopard as any OS X before it. That means the PowerPC code IS optimized for PowerPC.

Almost all Mac apps are written in xcode which simultaneously creates Intel and PowerPC optimized code. You guys can rely on blind facts but I will rely on real ones.
Click to expand...

I meant, that IMO PPC optimization is not that good as Intel in 10.5. That's why i put ";)" at the end of my phrase. I also noticed better performance on multithread tasks with Intel than PPC (10.5). PPC seems to do better with multithread tasks with 10.4. Depending on testing method YMMV ofc :)

What do you use to test altivec performance?

BTW, i second MacHamster's admiration of your Sawtooth. Looks very nice indeed. Could you post some pics of the internals?
 
666sheep said:
I meant, that IMO PPC optimization is not that good as Intel in 10.5. That's why i put ";)" at the end of my phrase. I also noticed better performance on multithread tasks with Intel than PPC (10.5). PPC seems to do better with multithread tasks with 10.4. Depending on testing method YMMV ofc :)

What do you use to test altivec performance?

BTW, i second MacHamster's admiration of your Sawtooth. Looks very nice indeed. Could you post some pics of the internals?
Click to expand...

I use Altivec fractal carbon. Very reliable benchmark as the results are very consistent.

When I get around to it I will take a photo of the innards. Here are a couple photos of the plastics after painting and before they went back on the tower.

2809405163_d78e0aa758.jpg


2789911964_ec518786a9.jpg
 
The dark colour looks black to many but it's actually a very dark brown. My father owns a private business where he gets government contracts to modify smaller busses into handi-cap accessible ones. He paints them white and blue over the yellow and black. this means he has a giant paint booth with propper ventilation and a 5000 dollar paint gun.

The brown is actually a ford factory colour from the 70's that he had the mixture numbers for. Called "black coffee brown". The white is just a standard high gloss white.

So with the colour name and my love of the drink I simply call the mac "coffee". haha what a geek huh?
 
And for the record.. my mac on Leopard gets 932 on geekbench and 57 on xbench. Under Tiger it was about 1019 for geekbench and 63 for xbench.

I am fine with the slight drop in performance as both my mac and myself respond better to Leopard. Personally I think Tiger is more pretty but I prefer the much improved socket layer tech for security sake in Leo. Time machine is a big plus also for me.
 
I finally got around to taking a photo of the insides. My girlfriend had my camera at the time so I had to use my crappy phone cam. Sorry about the glare and bad quality. I also took one of the back. The nobs are a PCI slot fan controller for the 2 heatsink fans. The Newertech upgrades are great but the fans spin at full speed and vibrate like crazy on the heatsink. So much that I worried about it shortening the life of the CPU.. So what I did was hook up a controller and using rubber mounts.

As you can see I need to clean up the wiring a bit. I also plan to modify an ATX PSU so I can add a Radeon 9800 and a couple more drives.

im000547.jpg


im000560.jpg
 
For the real geeks like me here is a long form list of my full setup specs:

System

- Apple PowerMac G4 Sawtooth (AGP Graphics)

CPU

-Freescale G4 1.8GHz 7448 PowerPC w/1MB on chip L2

Memory

-2GB Crucial PC133 SDRAM (4x 512MB) w/heat spreaders

Storage/Optical

FirmTek Seritek/1S2 2 channel PCI SATA controller
-Western Digital 1TB Caviar Black SATA HD 64MB buffer (boot)
-Hitachi 1TB Deskstar 7K1000.B SATA HD 16MB buffer
Antec MX-100 USB 2.0 HD Enclosure
-Hitachi 1TB Deskstar 7K1000.C SATA HD 32MB buffer
Antec MX-100 USB 2.0 HD Enclosure for Time Machine
-Hitachi 500GB Deskstar P7K500 SATA HD 16MB buffer
On-board ATA/33
-Pioneer DVR-112D PATA 16x DL DVD-RW
-iOmega 100MB Zip Drive

Video

-Nvidia Geforce 6200 256MB DDR2 AGP low profile video card (flashed x86)
-Dell Ultrasharp 2407WFP-HC 24" LCD (1920x1200 DVI PVA panel)

USB

-NEC Chipset USB 2.0 PCI card w/ 6 ports (5 ext. 1 int.)
-Creative Xmod USB sound controller w/volume knob
-Elgato EyeTV 250 USB TV tuner w/65 channels
-iPod Shuffle 1GB 2nd gen. (silver)
-Canon iP2000 Ink-jet printer
-Apple Aluminum keyboard & Logitech Trackman Wheel trackball

Sound

-Logitech X-230 2:1 speakers
-2x Sony MDR-V300 headphones

Case

-Coffee-Brown & White painted case plastics (car paint)
-Noctua NF-P12 120mm fan w/anti-vibration mounts (system)
-Antec Tricool DBB 80mm fan (power supply)
-PCI fan controller for 2x 50mm CPU heatsink fans

Internet

-15Mbit Down 1Mbit Up cable internet
-Linksys 4-Port 10/100 router
-Built-in 10/100 ethernet

Power

-OptiUPS 600 Watt voltage stabilizer
-MinuteMan 1800 Watt 2100 Joules surge protector

Furniture

-Ikea Jerker desk with 2 shelves
-Ikea Markus chair
 
Nice, very nice setup :) Is this Newertech CPU overclockable? Can you change multiplier on it or is it auto adjusting (like Sonnets)?
Sawtooths work very well with 120 MHz bus, i've done few of them (removing resistors on logic board is easiest way). Only uncomfortable thing is, that you must set proper frequencies in OF to see proper CPU and FSB freq in ASP. But improvement is noticeable. Think about that, especially if you're real geek ;)

Some Cube users (with upgraded CPUs) reported that FSB 100 is choking their machines, especially with Leo (to be not completely off-topic;)).
 
666sheep said:
Nice, very nice setup :) Is this Newertech CPU overclockable? Can you change multiplier on it or is it auto adjusting (like Sonnets)?
Sawtooths work very well with 120 MHz bus, i've done few of them (removing resistors on logic board is easiest way). Only uncomfortable thing is, that you must set proper frequencies in OF to see proper CPU and FSB freq in ASP. But improvement is noticeable. Think about that, especially if you're real geek ;)

Some Cube users (with upgraded CPUs) reported that FSB 100 is choking their machines, especially with Leo (to be not completely off-topic;)).
Click to expand...

The Newertech upgrades are overclockable as the jumpers are manual. My warrantee runs out next month so I will try it out then. Guess I will start with 1.9GHz and if that runs well I will try 2.0. I don't mind modifying CPU or GPU speed but I am not one to go mess with the bus. Other than the CPU the bus is the heart of the system and I prefer stability and longevity there.

About the 100MHz bus and choking.. I personally have little issue with that. The only time it really seems to slow for me is when I have 7+ apps running and maxed cpu. As I mentioned earlier in this thread though I think the 7448 chip makes a big difference. The 1MB of on chip full speed L2 seems to help make up for the slower bus. It obviously cannot fully compensate for it but it helps a lot trust me. The 7447 chips/upgrades only have 512K L2 and don't have near the power of the 7448 in many aspects. Thats why a 1.8 GHz 7447 sells for at least 150 less than mine did. Sadly Apple stopped using the G4 when the 7447 was the current chip. I am 99% sure that all the available cube upgrades are 7447 not 7448 chips. Also the cube can only hold 1.5 GB RAM vs. 2GB in the sawtooth. 512 on it's own isn't much but when you're living in the 1.5-2GB range an extra 512MB can help a lot for overall system smoothness. Leopard runs silky smooth on my current config. Before I got the 7448 chip.. not so much. And this might sound crazy but my memory actually performs 3-4% better with the heat spreaders. Most people only use spreaders when overclocking ram but I find they can actually slightly improve performance and lifespan. I ALWAYS use spreaders. Cool memory is happy memory. :)

For me power is only half of what makes a computer good. The other half is how you use that power and with what apps and what situations. I am not really sure what it is about PowerPC chips but I love them. I will use them as long as I possibly can. I have more than enough in the bank to go pay cash for a mac pro and 30" apple display but I don't need it.

I actually have 2 spare sawtooths. One has a G4 1GHz upgrade and the other has the stock 400mhz. They sit in the closet unplugged for future parts when needed. I even plan to spend the 800 OWC charges for the dual 1.8GHz 7448.
 
zen.state said:
And for the record.. my mac on Leopard gets 932 on geekbench and 57 on xbench. Under Tiger it was about 1019 for geekbench and 63 for xbench.
Click to expand...

Hey, i found your Xbench score :)

Regardless of Xbench's inconsistency and sketchiness, it's nice to see that 7448 CPU is that good, especially in comparison with MDD CPU.
Here's my old OC'ed MDD (in rebuild now) vs your Sawtooth: http://db.xbench.com/merge.xhtml?doc1=442130&doc2=359219
 
666sheep said:
Hey, i found your Xbench score :)

Regardless of Xbench's inconsistency and sketchiness, it's nice to see that 7448 CPU is that good, especially in comparison with MDD CPU.
Here's my old OC'ed MDD (in rebuild now) vs your Sawtooth: http://db.xbench.com/merge.xhtml?doc1=442130&doc2=359219
Click to expand...

Thanks for the link! The MDD beats it on I/O but thats what you get with a 67MHz faster bus. I am amazed at how close the user interface score is at 116 MDD and 82 Sawtooth. That is pretty close considering I have a Geforce 6200 in AGP 2x vs. the MDD with a 9800 and AGP 4x.
 
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom
Back