Leopard after a Month--How Fast is It?

[12/3] [Updated] Leopard after a Month--How Fast is It?

After using Leopard for five weeks we are still having some significant issues, even after the update to 10.5.1. In fact, the 10.5.1 update created one of our major issues--our 15" 2.4GHz MacBook Pro Core 2 Duo will not sleep when we use it with an aluminum Apple Bluetooth wireless keyboard.

The other significant Leopard 10.5.1 issues are:

• Dreamweaver CS3 nearly always has an unexpected quit after sleep;
• occasionally no video on wake if we use our FirmTek eSATA ExpressCard (driver/firmware v5.3.1) in our MacBook Pro;
• usually no video on wake if our 24" 2.8GHz aluminum iMac is put to sleep;
• occasionally data entry fields do not work in Safari and Oracle Calendar on our MacBook Pro and Mac mini Core Duo when we use either the original or aluminum Apple Bluetooth wireless keyboards.

These issues are all on Intel-based Macs. We have no problems at all with a Dual 2GHz PowerMac G5, a 12" 1.4GHz(upgraded) PowerBook G4 and a 1.4GHz(upgraded) PowerMac G4 Cube. With Apple having moved to Intel processors, one might think that the reverse would be the case.

Leopard's Time Machine uses a huge amount of disk space. Our external drive's 465GB capacity was filled to capacity after just 21 days of Time Machine backups. Time Machine is now dropping the older days as new days come and go.

With all the issues, we have been restarting our MacBook Pro a lot and have noticed that Leopard takes quite a long time to start up. So how fast is it? We applied many of our speed tests to the MacBook Pro and aluminum iMac. The tests we ran consisted of four benchmarking applications and four or five so-called "real world" tests.

Here are the results for the MacBook Pro:

Speed of 15" 2.4GHz MacBook Pro Core 2 Duo running Leopard vs Tiger

Test	10.5.1	10.4.9	Performance Change (%)
BENCHMARKS
Altivec Fractal 1.3 (GFLOPS)	9.6	10.0	-4
Cinebench 9.5 Rendering (2x)	707	738	-4
-OpenGL Hardware Lighting	4027	3903	+3
Let1kWindowsBloom 1.0+	11	8	-38
Xbench 1.3 - CPU	137	122	+11
- Threading	187	236	-21
- Memory	159	135	+18
- Quartz Graphics	201	175	+15
- OpenGL Graphics	129	137	-6
-User Interface Graphics	303	432	-30
"REAL WORLD" TESTS
Startup (sec)	56	31	-80
Shutdown (sec)	8	10	+20
Dup. 500 MB Folder (sec)	62	61	-2
Quake III Arena (frames/sec)	398	407	-2
Halo 2.0.2 (frames/sec)	100	86	+16

And those for the aluminum iMac:

Speed of 24" 2.8GHz Aluminum iMac Core 2 Duo running Leopard vs Tiger

Test	10.5.1	10.4.10	Performance Change (%)
BENCHMARKS
Altivec Fractal 1.3 (GFLOPS)	11.4	11.5	-1
Cinebench 9.5 Rendering (2x)	747	876	-15
-OpenGL Hardware Lighting	5110	5418	-6
Let1kWindowsBloom 1.0+	8	6	-33
Xbench 1.3 - CPU	162	145	+12
- Threading	217	281	-23
- Memory	168	160	+5
- Quartz Graphics	228	200	+14
- OpenGL Graphics	206	154	+34
-User Interface Graphics	380	652	-42
"REAL WORLD" TESTS
Startup (sec)	25	25	0
Shutdown (sec)	4	6	+33
Dup. 500 MB Folder (sec)	19	24	+21
Doom 3 (frames/sec)	115	107	+7

There is some consistency in the results. The major decreases in performance, i.e. greater than 20%, in the canned benchmarks are the same: Let1kwindowsbloom, Xbench Threading and Xbench User Interface graphics. In the real world tests, both computers show a major improved performance, i.e. greater then 20%, in shutdown times and a minor improvement in newer 3D game performances, Halo and Doom3.

The major difference is the huge increase in the MacBook Pro's startup time but no change for the iMac. [Update: The startup time on our 1.66GHz Mac mini Core Duo also increased from 25 seconds with Tiger 10.4.11 to 53 seconds with Leopard 10.5.1.] This result, along with the general slower performance on both Macs was surprising to us. We went so far as to do a clean re-install of Leopard on our MacBook Pro (originally an Archive and Install) to see if that would resolve its very slow startup time and stability issues--it didn't.

In summary, we are not happy with the general performance of Leopard 10.5.1 relative to Tiger, especially the MacBook Pro's [and Mac mini's] very poor startup time. Our expectation was that Leopard's code re-write would actually significantly improve performance. We are even less happy with the stability issues involving sleep and data fields/Bluetooth keyboards. However, we are encouraged with the continued improvement in 3D graphics and hope to see Adobe application updates and a Leopard 10.5.2 update from Apple very soon. [Bill Fox and Dana Baggett]