Click for SMOKING FAST SATA 6G SSDs!
Click for SMOKING FAST SATA 6G SSDs!


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: XLR8's MACh Speed G4zTM
G4 CPU Upgrade for AIO/Beige/B&W G3s
By Mike
Published: 11/22/99
*Real World* Performance Tests
Intro | Benchmarks  | Appl. Tests | Software Controls  | Installation | Specs/Design | Summary
Applications and Game Performance
This page lists test results in common Mac applications like Photoshop 5.5, SoundJam MP, Word98, Bryce 2, Infini-D 4.01, After Effects 3.1, Premiere 4.2.1 and many popular 3D games like Quake1, Quake2, Q3test and Unreal.

Note: As I mentioned on the intro page, although not shown in this review, I later tested G4s vs G3s using iMovie (timing now long it takes to complete the export/output to a Quicktime Movie). A G4 is almost twice as fast at this task as the same speed G3, one of the most impressive gains from a G4 I have seen. If you spend a lot of time doing video compression or output of DV to Quicktime movies, a G4 CPU can nearly cut times in half compared to a G3 cpu of the same speed, in the same system.


SoundJam MP 1.11 MP3 Encoding Tests
Casaday & Green's SoundJam MP 1.11 features Altivec support that shows a sample of what gains can be had from use of the G4's unique instruction set. Since the Apple G4/AGP system had a DVD drive and the B&W G3 system had a 32X CDROM drive, I also tested times to convert a song to MP3 format from the hard drive (removing the CDrom drive as a factor). The hard drive file was an AIFF format of the same 5 minute, 46 second song file used in the test of CD to MP3 encoding. In these tests, a G4 CPU was from 48 to 56% faster than a G3 CPU of the same speed.



Photoshop 5.5 Tests:
I also ran a complete PS5Bench (21 filter test) series using Photoshop v5.5 with the Altivec extensions (active for G4 CPUs only). PSBench settings are 1024x768, millions colors, VM off, Interpolation set to bicubic (better) and Photoshop should be allocated enough RAM to avoid any swap file activity from the 10MB test image filter actions. (I allocated 140MB to Photoshop 5.5 for this review.)

Unlike my past PSBench tests, to remove apparent disk activity I saw even with a 10MB image file and 140MB of RAM allocated to Photoshop, I set Photoshop 5.5's 'History' settings from the default 20 to 1 and unchecked the 'automatically create snapshot' option. This dramatically lowered the filter times and removed all signs of disk activity during the filter tests (each filter is run 3 times during the test). Photoshop 5.5 showed much higher PSBench times without these changes than I saw with Photoshop 5.02 from previous tests, since v5.5 seemed to hit the scratch disk much more quickly than I'd seen with previous versions, at least with the default history settings.

Note: All G4 CPU tests had the 4 OS 9 Altivec extensions active as well as the Adobe first release Altivec Core (v5.5) and Lighting Effects (v5.5.1) Filter. As many readers know, there is a bug in the current Altivec Core extension that should be fixed in a future release.

The Altivec Gain column notes how many times faster the G4 CPU w/Altivec extensions were than the same speed G3 (G4/400 vs G3/400 in the same machine). The total time shows only a 15% gain, due to the fact many of the filters were not affected by Altivec including those that took by far the longest times to complete. Notice the almost 3.5x faster Lighting Effects filter performance however. At the same CPU clock speed a 3.5 times increase in performance is very impressive - which shows the performance potential of Altivec instructions.

[Update: I've added results with an AMD Athlon 600MHz CPU based system running Windows 98. Note: Unlike the Pentium III and G4, there is no support in PS 5.5 for AMD's Altivec/PIII-like enhanced instruction set, but it still does well on brute force alone (triple superscalar FPU engine, highly pipelined design and 600MHz clock speed).]

Filter Altivec Gain
(G4/400 vs G3/400)
XLR8 G4/400Z
(in B&W G3)
B&W G3/400 Apple G4/450 AGP XLR8 G4/450Z
(in B&W G3)
Athlon 600MHz
Win98
256MB RAM
Rotate 90° CW 2.67x 0.3 0.8 0.3 0.3 0.3
Rotate 9° CW 1.15x 2.6 3 2.3 2.3 2.2
Rotate .9° CW 1.17x 2.4 2.8 2.2 2.2 2.1
1 pix Gaus. Blur 2.00x 0.7 1.4 0.5 0.7 0.6
3.7 pix Gaus. Blur 1.57x 2.1 3.3 1.6 1.9 2.2
85 pix Gaus. Blur 1.65x 2.6 4.3 1.9 2.4 4.2
Unsharp Mask
50%/1pix/0 level
1.70x 1 1.7 0.8 0.9 0.7
Unsharp Mask
50%/3.7pix/0 level
1.54x 2.4 3.7 1.9 2.2 2.3
Unsharp Mask 50%/10pix/5 level 1.48x 2.5 3.7 1.9 2.3 2.6
Despeckle 2.08x 1.2 2.5 1.1 0.9 2.0
RGB-CYMK 1.02x 5 5.1 4.7 4.5 4.2
Reduce 60% 2.45x 0.53 1.3 0.5 0.5 1.1
Lens Flare 1.00x 4.4 4.4 3.8 4.0 3.3
Color Halftone 1.00x 3.9 3.9 3.4 3.5 5.0
NTSC Colors 1.00x 4.2 4.2 4.0 3.9 4.7
Accented Edges 1.02x 11.1 11.3 10.3 9.9 12.5
Pointillize 1.01x 15.8 15.9 14.5 14.1 13.8
Water Colors 1.01x 23.3 23.6 21.7 20.8 25.7
Polar Coordinates 1.38x 3.9 5.4 3.3 3.5 7.2
Radial Blur 1.01x 35.9 36.2 34.6 32.1 35.8
Lighting Effects 3.44x 2.7 9.3 2.3 2.5 6.2
Total Time (1.15x) 128.53 147.8 117.6 115.4 138.7
System Altivec Gain
(G4/400 vs G3/400)
XLR8 G4/400Z
(in B&W G3)
B&W G3/400 Apple G4/450 AGP XLR8 G4/450Z
(in B&W G3)
Athlon 600MHz
Win98
256MB RAM

Interesting is that the fastest system was the XLR8 G4/400 upgrade running at 450MHz, slightly faster than the Apple G4/450 AGP system. Although 150MHz faster in clock speed, the Athlon trails most systems in this test. If Photoshop had 3DNow extension support, the Athlon would have done far better (at least on filters that were enhanced).


Word 98 Tests
I ran several tests in Microsoft Word 98 using the MacBench 4.0 doc file (203 pages, 50.9K words) which contains a good mix of font sizes, etc. The G4 gain was very small in these tests as shown in the table below. I didn't have Office 98 installed on the Apple G4/450 system so therefore it the table shows only a MHz/MHz comparison of the G4/400 vs G3/400 in the same B& system. All times are in seconds.
Word 98 Test XLR8 G4/400Z B&W G3/400 G4 Gain
Scroll Down 11.6 12.1 1.04x
Scroll Up 11.5 11.8 1.03x
Word Count 1.87 2.02 1.08x
Summarize 56.85 56.94 Nil

I included this to satisfy some reader's curiosity, but generally you don't buy a CPU upgrade to speed up a word processor (which most often is waiting on you during typical use).


Infini-D 4.01 Tests
Infini-D 4.01 does not use Altivec extensions, but is a common application I have used for comparisons of CPU/FPU performance. I used the same 'Chapter 7 completed' tutorial scene file from my past reviews. Rendering quality was set to Ray Trace, medium anti-aliasing, shadows on, patch detail low. I didn't change the default QT movie output file options. The graph below shows times to complete the 150 frame movie rendering with the stock B&W G3/400, the same system with the XLR8 G4 upgrade at 400 and 450MHz and the Apple G4/450 AGP (Sawtooth) system, all running OS 9. Infini-D was allocated 40MB of RAM for the tests. OS 9 and its 4 Altivec extensions were used for all G4 tests. An interesting note was that the G4/450 AGP Sawtooth system finished this same test more than 2 minutes faster with OS 8.6.


Just as in my August tests with a G4/400 upgrade in a Genesis system, in this test a G4 CPU was a bit slower than a G3 of the same speed. This makes no sense to me but repeated testing show it to be a fact. If Infini-D 4 used Altivec I'm sure the results would have been far different.

Bryce 2 Tests
I also ran tests in Bryce 2, using the samples scenes in the KAI folder called "Magical Knight I" and "Alexandria II". Bryce 2 was left at the default memory allocation of 16980k, antialiasing was on, screen resolution was set to 1024x768, thousands colors.


Again a G3 CPU of the same speed does just as well (a second or two faster) as a G4 CPU. Bryce2 does not take advantage of Altivec extensions of course (but neither does most applications you're using today). I saw the same slightly better G3 performance in tests of a G4 upgrade in my Genesis last August. Perhaps there is something in the Bryce2 and Infini-D 4 code that is better suited to a G3 rather than a G4.

After Effects 3.1 Tests
The following is a chart showing the time to render a special effects movie (appx. 8MB file size) in After Effects 3.1. Resolution was set to 1024x768, thousands colors as was common on all but the Photoshop 5.5 tests.

Note: As shown in my MAChCarrier G3/500 review, under OS 8.6 a Genesis system with that upgrade running 500/250/50 completed this same test in under 3 minutes.


Game Performance Tests:
Tests were done in Rave Quake, Quake2, Q3test 1.08 and Unreal v224b7. Since your video card is a major factor in the speed and quality of 3D games, the first gamer's purchase should be a fast 3D card (if you don't already own one) such as a 3Dfx Voodoo 2, Voodoo3 or perhaps a Rage128 or Proformance3. Also important and the first upgrade you should perform is plenty of RAM (Unreal and Quake 3 both need 90MB+ now). If you're well equipped with RAM (128MB or more) and have a G3 CPU already, a faster 3D video card is a far better investment for improved game performance and visuals as well.

Both the B&W G3 and Apple G4/450 AGP (Sawtooth) had OS 9 and OpenGL 1.1.2 installed for all tests. The G4 CPUs had OS 9's Altivec extensions active.

The AGP card/slot of the Sawtooth is not that much faster than the 66MHz PCI slot of the B&W G3. At 1024x768, the Rage128 chip itself is the bottleneck to higher performance, not the CPU.


Quake 2: OpenGL mode. All tests used the latest OpenGL 1.1.2 (Final Release). Graphics settings in the game were :

  • 16-bit color
  • Max Texture Quality
  • No Sync Every Frame
  • No CD Audio
  • Full Screen

All scores in this test were using a Rage128 video card (B&W G3 comparisons used the Rage128 Orion card in 66MHz PCI slot, Apple G4/450 system used OEM Rage128 AGP card). OS 9 used for all tests (Altivec extensions were present). To see the difference a faster video card can provide, see the Voodoo3 3000 AGP and PCI card scores from tests in the G4/450 system at my Mac 3D Card Roundup. Quake 2 1024x768 framerates are typically twice as high with a Voodoo3 as the Rage128. The new Rage128 PRO in the latest Apple G4 models promises up to 50% better 3D performance which should help, but not catch the Voodoo3.


Q3Test 1.08: I used Q3Test v1.08 for all tests. Graphics settings in the game were :

  • 16-bit color
  • Lightmap lighting
  • Geometric detail high
  • Texture quality 16-bit
  • Texture detail 1 notch from max
  • Texture filter - bilinear

The tables below show Q3Test performance with each of the system/upgrade combos in frames-per-second.

For results with faster video cards like the 3Dfx Voodoo3 - see my Mac 3D Card Roundup.

Unreal 224B7: The table below summarizes results from the 3rd cycle Timedemo of the 'Castle flyby' scene. (All tests used the same detail settings - see my FPS Database entry page to download Unreal.ini files for RAVE and 3dfx cards.)

Higher performance cards like the 3Dfx Voodoo3 would result in higher framerates, especially at 1024x768 modes.


Summary: Overall applications and game performance was very good even in comparison to an Apple G4/450 Sawtooth system. For non-Altivec enhanced applications, don't expect any real gain from a G4 CPU compared to a G3 of the same speed however. For those applications that do support Altivec, significant gains are possible.

For 3D games, the fill rate limits of the Rage128 chip usually hampers higher resolution performance. If you want the best game performance, a 3Dfx Voodoo3 card is currently a better choice (delivering up to twice the framerates at 1024x768 mode as any other Mac card to date). There are compatibility issues however so check my www.mac3dfx.com site for details before deciding if the Voodoo3 card is for you .

The next page describes the software cache control supplied with the card. Or you may use the links below to jump to a specific page.

Index of XLR8 MACh Speed G4 400MHz Review

Intro | Benchmarks  | Appl. Tests | Software Controls  | Installation | Specs/Design | Summary

- or -
Back to WWW.XLR8YOURMAC.COM


Copyright © Mike, 1999.
All Rights Reserved.
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.