Click for iPod Accessories

A Click shows your site support to my Sponsors

Accelerate Your Mac! - the source for performance news and reviews
The Source for Mac Performance News and Reviews
Review: Giga Designs G4 1.25GHz CPU Upgrade
By Mike
Published: 3/14/2003
Real World Performance Tests
Intro | Benchmarks  | Apps Tests | Game Tests | Installation | Specs/Design
Applications Performance Tests

This page lists test results in common Mac applications like iMovie2, iTunes3, Photoshop 7.0, and time to convert a very large QT movie (1.92GB) to MPEG4. It also includes a multitasking test using iTunes3 playback/visuals while downloading a large file via a high-speed internet connection. (The system is doing Audio, Video, Disk, Firewire and Network I/O simultaneously.)

Although I wish I could have included more tests, these results will give you a good idea of how performance compares with these upgrade in many common tasks. (I do not currently have a 100MHz bus G4 tower to test.)

The graphs include tests of:

  • A Mirror Drive Door (MDD) Dual 1GHz G4 system
  • A Digital Audio (DA) G4/533 Dual CPU
  • The DA system w/Powerlogix dual 1GHz upgrade
  • The DA system w/Giga Designs G4/800 (7451 with no L3 cache),
  • The DA system w/Giga Designs 1GHz (7455 CPU w/2MB L3 cache)
  • The DA system w/Giga Designs 1GHz model set to 800MHz (to compare 800MHz w/L3, although it's not a perfect comparison since the 800MHz G4 w/o L3 cache is a 7451 CPU)
  • The DA system w/Giga Designs 1.25GHz (7455A w/2MB L3 cache) set to 1.33GHz.

I wish the Mac had the wide range of benchmarks available for the PC and I wish I could afford to buy $1000 packages like FinalCut Pro to use for tests. (I can't justify that sort of outlay these days just for benchmarking.) However I try to use the common/popular applications that I have in tests that you can repeat, as well as providing information that others don't (not just a series of graphs and benchmarks), even though it takes a lot of time. If I had a team of people here with more expensive software packages and more help with running this site, there's a lot more I'd like to do. But you still get more than just numbers and graphs in the multi-page reviews here, although they are not as complete as I would like.

Quicktime to MPEG4 Conversion
I used a very large 1.92GB (GigaByte) Quicktime movie and timed how long it took to export (convert) to MPEG4 using the default settings in Quicktime 6 (Pro). (Shorter bars are faster of course.)

QT to MPEG4 Export Test Results

Notice this QT6 Pro export shows no benefit from dual CPUs. I don't currently own any expensive/high-end video apps (like Final Cut Pro) currently, so I can't provide comparisons using them.

iMovie 2 Tests
I used the same test I have since iMovie was released - stacking the 6 tutorial file clips end-to-end (no transitions) and timed how long it took to export the movie using the standard "CDROM Medium" settings. (By using the tutorial with no variables like transitions, it's something everyone can easily test with their own systems.) I used the last version of iMovie 2 with OS X 10.2.3 so that past upgrade results would be comparable. (Shorter bars/lower times are faster.)

iMovie2 QT Export Test Results

The Giga Designs 1.33GHz upgrade performance bettered a MDD Dual 1GHz and was about twice as fast as the original Dual G4/533 CPU module in this test. (As you can see from the results, iMovie2 doesn't really take advantage of dual CPUs.)

iTunes MP3 Conversion
Time to convert an Audio CD outer track (4 min, 20 second) song to MP3 (192Kbs rate quality setting) using a Lite-on 40x12x48x Firewire drive (oxford911 bridge case). The outermost song track used to try to minimize the effect of the drive's speed (although the drive speed isn't a bottleneck for this test). I used a fast Firewire CDRW drive so that the same drive could be used for all system tests and avoid using the (relatively slow) internal Superdrives. (Shorter bars/lower times are faster.)

iTunes MP3 convert

Dual CPUs are a plus for this test as you can see from the MDD and PL Dual results.

MultiTasking: iTunes3 Playback + Visuals + High-Speed Downloading in I.E.:
To test how each CPU performed while multitasking, I ran a test doing something I've personally done many times - playing a song in iTunes with the visuals on while downloading a large file. The internet connection is a cable modem (via Airport). iTunes3 Visuals was set to large size, with the show FPS and Use OpenGL enabled. (I did not select the options to cap framerate or use Faster/rougher display.) Screen mode was 1600x1200/millions colors on a Sony F400 CRT. Internet Explorer 5.2.2 was used. Downloads were from an Apple server of a large file (30MB).
The graph below shows the average FPS rates shown in iTunes3 visuals during the duration of the download (while audio was playing of course - higher numbers are faster).
This may not be the perfect multitasking test, but it's a) something that I commonly do in OS X and b) the system is doing Audio, Video, Disk, Firewire and Network I/O simultaneously.

Multitasking iTunes Visuals

Granted this isn't a perfectly repeatable test (due to varying net traffic/server load, etc.) but is still interesting (note the single vs dual CPU results).


PhotoShop 7.0 (OS X) Tests
I used the 50MB image file (advanced) version of the PSBench 21 filter action script. (PSBench was created as a cross-platform test years ago for Photoshop performance that anyone can download to use themselves. Not an unknown/unspecified mix of filters as often seen in some reviews/marketing, which could be chosen to include filters that performed better on a specific processor or platform. PSBench is something you can actually download and try yourself.)

In addition to a graph showing total times for the filter series, as in my other reviews, I've also listed each filter, its description and the time each cpu took to complete it. (Each filter is run 3 times, the avg. is displayed in the table below. PhotoShop's timing function is used, not a stopwatch. For all PSbench tests, I used a History setting of 1 (minimum) and unchecked create first snapshot. PShop allocated appx 450MB of ram.)

PS 7 comparison

As you can see from the table of individual filter results below, not all the filters benefit from Dual CPUs - note the Giga Designs 1.33GHz had a slightly faster total time than a dual 1GHz CPU upgrade in the same Digital Audio system (same OS, same settings, etc.).

Filter-by-Filter Performance Comparisons:
The table below shows the times in seconds (avg. of 3 runs) to perform each of the 21 filter operations on the 50MB image. Bold numbers indicate the fastest performance of all systems/upgrades tested. (Note: Averages of the 3 filter run times are shown, however the first filter test almost always shows a lot of variation - for instance 0.9, 1.2 and 0.6 seconds with the 1.33GHz upgrade. Later filter tests were much more consistent in run/run times.)

PS 7 Filter comparison

You can download the latest PSBench action script (I used the Advanced 50MB version) at the PSBench home page.
Note: I've seen some reviewers complain about PSBench (the time it takes to run is their real primary gripe I think). Keep in mind that the script repeats each filter test 3 times for a better average time per filter (since the first filter tests especially have variation in run/run times, although little variation is seen in most filters after the first). Like any test, an average of multiple runs is actually preferable (scientifically) to just a single run. Since August 2002, I've used the 50MB version - older reviews back to 2000 used the 10MB PSBench5 script since that is what I first started using and wanted the exact same script used for comparison purposes. (And note the same 21 filters are still used in PSBench7.) PSBench creates its own image for testing, allowing everyone to use the exact same test image without a large download. (PSBench's script is a tiny download - less than 5KB.)
I'll address another reviewer's complaints about PBench:

  • As for a comment that an artist doesn't restore the image before each filter - PSBench does this so that each filter is run on the exact same image each time. It's done that way to assure the test image is exactly the same for every filter test. (I know most of you realize that.) Yes, that takes more review/test time. The benefit to having each filter work on the exact same image is in case (which has happened) where there may be a PhotoShop version or platform specific bug in a routine or specific filter that would be compounded into the image used by another filter later in the series. (This has already happened in previous Mac and PC versions of Photoshop.)

  • As far as the (valid) complaint about having to record each filter time taken, rather than only a total time for a series of filters - consider this. If all you saw was the total time for all filters (and no individual filter performance results), spotting bugs like the one that PhotoShop 6.0 had with a specific filter would be impossible. (PS 6.0 had a bug where one filter's performance on the PC was many times slower than than it should be - if all that was reported was a total time, that filter bug would never have been spotted. People that compared only total times thought the PC performance was far below what it really was.) Again more reviewer time, but it's a 'higher resolution' of test data.

  • By providing both total time and individual filter times, you have more detailed performance results and can spot which filters are more Dual Processor or Altivec aware for instance. (For reviews that show G3 vs G4 or single vs dual CPU tests.) With only a total time - you can't 'see' that information in detail. It's more work for reviewers, but more information for end users. If all you want to see is total time, it's there, but so is specific filter performance.

These are reasons I started using PSBench in 2000 and still do, because it's a fair cross-platform benchmark that you can download yourself to compare to results here or at their homepage. Yes it takes more time to run and average the results than just a single series of filters all run at once. However I still take the time to show you the total time for all 21 filters and the filter-by-filter results, so you can compare which filters you use most, see how each compares, which show a benefit from dual vs single CPUs, etc.. It's not a perfect benchmark, nothing really is. (If a professional that makes their living with Photoshop has a better action script, I'd be glad to try it.)


The next page covers performance tests with several popular 3d Mac games.


Index of Giga Design CPU Upgrade Review Pages

Intro | Benchmarks  | Apps Tests | Game Tests | Installation | Specs/Design

- or -
Back to WWW.XLR8YOURMAC.COM


Copyright © 2003.

No part of this sites content or images are to be reproduced or distributed in any form without written permission.
All brand or product names mentioned here are properties of their respective companies.

Users of the web site must read and are bound by the terms and conditions of use.