NOVEMBER 08, 2002
Apples to Apples II: Dualies and Droolies
Performance benchmarks for the current G4 lineup
by David Nagel

Everything about the current lineup of Macintosh G4s should make them at least mildly superior to the previous lineup of desktop G4s ... in theory, anyway. They have faster chips, dual processors all around, faster bus speeds and, of course, DDR memory. And those are only the obvious advantages. But does theory hold up to practice in the Mac universe? We tested the entire lineup of current G4s--from the dual 867 MHz G4 to the dual 1.25 GHz G4--against the previous generation and found some fascinating results.

And what's so bloody fascinating about these results? Well, maybe they're not that fascinating, but I hate to think I wasted about 50 hours of work without something interesting to show for it. As with our previous benchmarks, performance depended heavily on how well the software was written for the hardware. However, in general, the new lineup of G4s did indeed show some impressive gains over the previous generation, which is somewhat contrary to results reported in other publications. The new dual 1 GHz G4, for example, saw gains as high as 11 percent over the previous-generation dual 1 GHz G4 in QuickTime encoding speed, though in some tests the new model ran virtually the same as the old. [an error occurred while processing this directive] Of course, if you compare the new mid-range G4 with the old mid-range G4, there's just no comparison at all. And the new high-end 1.25 GHz G4 simply blows away the old high-end 1 GHz G4. The new dual 867 MHz model--the entry-level G4--offered some impressive results as well.

Our results, as always, are based on tests we think are fairly representative of what a creative professional in the visual arts might encounter when actually using a computer for professional work. For motion graphics and compositing, we ran several tests using Adobe After Effects 5.5, Discreet Combustion 2 and Apple Final Cut Pro 3. For print graphics, we ran tests involving Adobe Photoshop 7 and Illustrator 10. And for Web graphics and video, we ran tests on Discreet Cleaner 6 and Wildform Flix Pro 3.

While you might not use these particular applications, their functions are representative of the kinds of work creative professionals encounter everyday: video editing, effects, compositing, graphics creation and manipulation, vector and text creation and manipulation, encoding to vector and SWF formats and transcoding video from QuickTime to various other formats.

The tests
All of the tests for the current lineup of desktop G4s (DDR memory) were run in Mac OS X on machines equipped with 1.25 GB of memory, as was the previous-generation dual 1 GHz G4. Tests on the previous-generation G4 933 MHz G4 were run under Mac OS X 10.1.4. Since we no longer have access to a 933, we can't update its benchmarks under the current version of Mac OS X. The machines were otherwise stock configurations. All networking and energy saving features were disabled. And all of the tests were run using the same settings on each machine. So, essentially, all things were about as equal as they could be.

Motion graphics, effects, video and compositing
Our first category is the one that usually shows the most dramatic difference between older and newer hardware. And such was the case with this latest round of benchmarks. The applications included Final Cut Pro, After Effects and Combustion, three powerhouses in the video category. Final Cut Pro and Combustion, in particular, know how to take advantage of recent Mac hardware, especially when it comes to dual processors. After Effects seems just barely to touch the second processor, but, since all of the current Macs run the same number of processors, there's not much of an issue there for benchmarking.

In the case of Final Cut Pro, we see the current line of G4s easily exceeding the performance of the old 933 MHz G4. But even against the old dual 1 GHz G4, even the lowliest sub-gigaHertz model in the current lineup held up fairly well. The tests all involved rendering, essentially measuring the raw 2D processing power of each machine. The tests were conducted as follows:

  • Test 1: Three video filters on a single video track: Gaussian Blur, Broadcast Safe and Basic 3D, with the layer rotating on all axes.
  • Test 2: Three clips arranged in two video tracks. I used the Color Corrector on one clip and Lens Flare and Posterize on the other two shorter clips. I also applied a Cross Iris transition.
  • Test 3: A single clip with Basic Border, Whirlpool and Curl.
  • Test 4: Primarily a filter test, including Diffuse, Mirror, Color Balance, Desaturate Highlights and 4-Point Garbage Matte.
  • Test 5: Three different text tracks (Video Generator) on top of video footage, including Typewriter, Crawl and Text. All were animated, and the Text track included a highly varied Pond Ripple effect.

Final Cut Pro 3 Tests

(All results listed in seconds.)
Dual 1.25 GHz
Dual 1 GHz
Dual 867 MHz

Dual 1 GHz
Old Model Running OS X 10.2

Dual 1 GHz
Old Model Running OS X 10.1.4
Single 933 MHz
Test 1
Test 2
Test 3
Test 4
Test 5

Page 1 of 4

1 2 3 4 Next
Related sites:Animation ArtistAV VideoCreative MacDigital AnimatorsDigital Media DesignerDigital Post ProductionDigital Pro SoundDigital ProducerDigital Video EditingDigital WebcastDV FormatDVD CreationFilm and Video MagazineMedia WorkstationThe WWUG
Related forums:

[an error occurred while processing this directive]