Hands-On Report: Mac OS X 10.4.3--How Fast Is It?

Last week Apple released Mac OS X 10.4.3 Update to Tiger and the server edition too. Apple released the previous update Mac OS X 10.4 at the end of July. As frequent readers know, we have benchmarked virtually every release of Mac OS X from the original 10.0 release through 10.4.2 to keep track of the improvements in its speed, a major complaint initially about Mac OS X.

Mac OS X runs very well these days and speed enhancements have largely focussed on graphics. The improvements in 10.4.3 included Core Graphics, Core Image and updated ATI and NVIDIA graphics drivers so attention to graphics continues. Apple's new-to-Tiger Quartz 2D Extreme technology (in addition to and not to be confused with Quartz Extreme) is not only not implemented by default but Apple stated that it is now permanantly disabled in 10.4.3. This is curious because we have shown that when enabled it provided a huge performance boost to the User Interface actions and Quartz Graphics benchmarks in Xbench.

So how fast is it? Did 10.4.3 speed anything up or slow anything down? To determine this we ran our usual suite of speed tests on our two test-bed Macs. The speed tests are composed of a series of "canned" benchmark applications and several so-called "real world" tests that anyone can repeat on their own Mac for comparison by following the steps that we perform.

Our long-term test Mac (since Mac OS X 10.0) has been a mostly-stock, year 2000-vintage 450 MHz Power Mac G4 Cube with 704MB RAM. With Mac OS X 10.3, Apple began including improvements that are aimed mainly at the G5 CPU and graphics cards with 32MB of video RAM or more (the Cube's has only 16MB) so we also began using our dual 2.5GHz Power Mac G5 with 1.5GB RAM and an Nvidia GeForce 6800 Ultra DDL graphics card, an upgrade, in the speed tests.

Here are the results for the G4 Cube using Mac OS X 10.4.3 and Mac OS X 10.4.2, 10.4 (Tiger) plus the last update of each two most recent "cat" versions, i.e. 10.3.9 (Panther) and 10.2.8 (Jaguar). Other versions back to 10.0 can be found in our previous articles:

450MHz G4 Cube Speed Tests (average* scores or secs).
Test Cube 450 MHz
Mac OS X Version:
10.2.8
10.3.9
10.4
10.4.2
10.4.3
Benchmarks:
Altivec Fractal (GFLOPS)
1.5
1.5
1.5
1.5
1.5
Cinebench 2003 - Rendering
49
48
48
48
49
- OpenGL Hardware Lighting
92
91
90
91
92
Let1kWindowsBloom 1.0 (sec)
61
53
30
30
31
Xbench - CPU
52
54
54
54
53
- Threading
34
42
42
42
42
- Memory
64
76
77
79
80
- Quartz Graphics
63
74
73
77
77
- OpenGL Graphics
84
80
88
88
88
- User Interface Graphics
60
102
150
148
147
- Hard Disk
66
54
56
54
51
"Real World" Tests:
Startup (sec)
54
52
53
49
52
Shutdown (sec)
7
15
16
15
16
Launch Classic Mode (sec)
-
25
36
30
31
Dup. 700 MB Folder 3600 Files (sec)
131
152
128
130
133
Scroll 142p Acrobat Reader File (sec)
57
55
55
54
55
Encode MP3 in iTunes (sec)
105
105
104
104
104
Export QT FQDV Movie in iMovie (sec)
32
36
38
39
38
Quake III Arena 1.32 (frames/sec)
22
34
32
32
34
*Average of three runs. Green is much faster. Red is much slower.

For this now low-end Mac, there are no changes--no improvements and no losses in performance when compared with 10.4.2.

Here are the Power Mac G5's results comparing 10.4.3 with 10.4.2, 10.4.1, 10.4 and 10.3.9:

Dual 2.5GHz Power Mac G5 Speed Tests (average* scores or secs).
Test Dual 2.5 GHz Power Mac G5
Mac OS X Version:
10.3.9
10.4+
10.4.1+
10.4.2+
10.4.3
Benchmarks:
Altivec Fractal (GFLOPS)
12.9
12.9
12.9
12.9
12.7
Cinebench 2003 - Rendering
647
649
646
648
645
- OpenGL Hardware Lighting
1601
1797
1801
1602
1601
Let1kWindowsBloom 1.0 (sec)
11
6
7
7
7
Xbench - CPU
241
242
241
241
243
- Threading
262
268
268
265
266
- Memory
333
385
376
371
378
- Quartz Graphics
348
525
521
507
257
- OpenGL Graphics
202
233
234
243
249
- User Interface Graphics
415
521
517
521
352
- Hard Disk
106
105
107
98
115++
"Real World" Tests:
Startup (sec)
40
35
37
38
38
Shutdown (sec)
10
8
11
10
13
Launch Classic Mode (sec)
10
11
10
11
11
Dup. 700 MB Folder 3600 Files (sec)
66
58
62
64
44++
Scroll 142p Acrobat Reader File (sec)
11
12
12
12
12
Encode MP3 in iTunes (sec)
54
52
52
52
52
Export QT FQDV Movie in iMovie (sec)
20
22
24
25
20
Quake III Arena 1.32 (frames/sec)
411
405
401
403
400
Halo 1.5.1/1.5.2 (frames/sec)
63**
67**
67**
66***
68***
Call of Duty 1.4 (frames/sec)
85
86
86
84
83
Doom3 1.3 (frames/sec)
-
-
-
53
50
*Average of 3 runs. **Halo 1.5.1 with Advanced Shaders. ***Halo 1.5.2 with Advanced Shaders. +With Quartz 2D Extreme enabled. ++New hard drive. Green is much faster. Red is much slower.

For the next-to-latest top-end Power Mac G5, the quartz graphics and user interface graphics performance of Mac OS X 10.4.3 in the Xbench tests decreased without Quartz 2D Extreme enabled by the Quartz Debug developer tools application. Despite this, there is not much effect, if any, on the "Real World" tests. The performance of Mac OS X 10.4.3 three 3D games decreased and one incresed--all slightly. Rendering a QuickTime movie improved to equal the performance in Panther 10.3.9.

Mac OS X 10.4.3, like 10.4.2 and 10.4.1, has mainly security, compatibility and stability improvements. We found no real speed improvements beyond previous versions of Mac OS X in either test computer.

The unusual history of Quartz 2D Extreme technology has apparently come to an end. During the development of Tiger after it was announced, Quartz 2D Extreme was touted as a faster graphics technology than Quartz Extreme. But when Tiger was released, Quartz 2D Extreme was available but not enabled by default. It had to be enabled with the developer tool Quartz Debug and, as shown in the table above, it made a huge difference in performance on some tests. The Read Me file with 10.4.3 claims that it has been permanently disabled even though one can still check "Enable Quartz 2D Extreme" in Quartz Debug. But when we did so the results of the Xbench tests are odd, giving a very slow score of 61 for the User Interface Graphics test. Well, there's still Mac OS X 10.5 "Leopard"....

We have used Mac OS X 10.4.3 for nearly a week on a number of Macs and PowerBooks without running into any new or significant problems. [Bill Fox]



Saw something? Send a tip

The archive ran on reader tips. What did you see, where, and do you want the credit?

Read by the editor. Never published without your say.

More in Reviews · This month in the archive