Click for Data Doubler kits!
Click for Data Doubler kits!


One Click on the Banner above shows support to my
Sponsors

PowerBook G3 and Solo Image


PowerBook G3/250 vs Gateway Solo PII/266 9100XL
Application Performance Tests
By Mike
Review Date: July 1998


This is part two of the review which lists application performance test results.
Part one listed initial impressions from use and a comparison of features.
In the tables below, scores listed in bold indicate best performance.

The Gateway Solo had 128MB of RAM and Windows 98.
The Maintenance Scheduler was running in the background
on the Solo, along with several system tray icons for
display settings, IE, etc. No other applications were running.
The Solo 8GB disk was approximately 50% full and not optimized.

The PowerBook G3/250 had 160MB, OS 8.1,
disk cache was set to 1MB, Virtual memory and AppleTalk were off.
The PB 4GB disk was also appx. 50% full and not optimized.

AC adapters were used for all tests, and energy settings were set
for maximum performance (never sleep, no processor cycling, etc.).


Power-On to OS Ready

I measured the time from pressing the power-on button to completely loading the OS (ready for use). The PowerBook had 160MB of RAM vs. 128MB for the Solo (more RAM to test at power-on) but the Solo was set to automatically scan the root directory for virii at boot.

Power-Up Times

PowerBook
G3/250

Gateway Solo
9100XL PII/266

1:16.50

1:09.07


Bryce 3D


The Bryce 3D test consisted of rendering the Lava Lake sample image.
Anti-aliasing was set to normal, Spatial Optimization set to low and report render time enabled.
Graphics mode on each machine was set to 1024x768, thousands colors.

Bryce 3D Lava Lake
Render Time

PowerBook
G3/250

Gateway Solo
9100XL PII/266

5:16

6:40


Ray Dream Studio 5


The Ray Dream test consisted of rendering the Complete.RD4 sample image.
Tests were run at the 'still-screen res' and 'still-high res' render settings.
Graphics mode on each machine was set to 1024x768, thousands colors.

Ray Dream Studio 5 Complete.RD4
Render Time

Render Mode

PowerBook
G3/250

Gateway Solo
9100XL PII/266

Screen Res
640x480, 72dpi

2:32.85

2:59.31

High Res
2048x1366, 300dpi

9:55.25

11:03.00


Adobe Premiere 4.2

The Premiere test consisted of making a Quicktime movie of the complete sample project which included transitions, titles and audio. Graphics mode on each machine was set to 1024x768, thousands colors, MMX plug-in used on Solo.


Test 1: 640x480, 24fps, video compressor, 75% quality setting with 22KHz 16-bit stereo audio.


Test 2: 320x240, 15fps, video compressor, 75% quality with 22KHz, 8-bit mono audio.

Premiere 4.2 Make Movie Tests

Render Mode

PowerBook
G3/250

Gateway Solo
9100XL PII/266

640x480, stereo

6:25.60

7:35.43

320x240, mono

50.08

48.22


Photoshop 5.0

The Photoshop tests consisted of a series of operations on the 'Photo5' sample file. Graphics mode on each machine was set to 1024x768, millions colors. Times included any image refresh that occurred at the end of the operation (as in normal use you'd not be able to continue working until after this occurred).

Photoshop 5.0 Tests

Render Mode

PowerBook
G3/250
(46MB alloc)

PowerBook
G3/250
(96MB alloc)

Gateway Solo
9100XL PII/266

Load Application

12.34

12.35

18.28

Load 4.41mb file

7.66

7.61

4.19

CYMK convert

3.69

3.66

1.96

300 DPI resample

1:32.81

55.81

1:10.57

Flatten image

44.22

37.09

1:18.84

Unsharp Mask
(50%/3 pixels)

20.84

8.19

7.96

Colored Pencil
(4/8/25)

1:39.90

1:31.47

2:09.72

Rotate Image
(3 degrees)

1:09.84

54.62

1:13.46

Save File*
(64MB)

2:24.38

1:41.44

3:51.88

* 4 layer, 300dpi file.


Microsoft Word and Excel Tests


Office 97 was used on the Solo, Office 98 on the Macintosh (latest versions available at time of tests).
Graphics mode on each machine was set to 1024x768, thousands colors.

Word Tests:

The Word test consisted of loading and scrolling through a 1213 page document with a mix of font sizes. The word 'Macintosh' was placed at the end of the document and a 'find' operation was timed as well as scrolling tests using the page-down key (held down). Fast save was disabled and after making one character change I measured the time required to save the file. The final test was the time to calculate the statistics on the document. For reference the file contained 4,929,439 characters including spaces. Fast saves were disabled.

MS Word Tests
(1213 Page Doc)

Test

PowerBook
G3/250

Gateway Solo
9100XL PII/266

Load Appl

5.07

4.03

Load File

Less than 1 sec.

Less than 1 sec.

Find

5.00

2.02

Save (6.9MB)

18.91

20.05

Scroll (Pg Dwn)

4:55.01

1:59.05

Statistics

18.22

7.07

Excel Tests:

The Excel test was a recalculation of a huge 2.2MB spreadsheet file containing a mix of integer and FP functions. Memory allocation was increased from 10MB to 30MB with no differences in the recalc. times. The spreadsheet was changed to 'precision as displayed' in an attempt to provide more comparison data by trimming the precision of the numbers used. The results were the most shocking of the entire test series and demonstrate that Windows Excel is obviously highly optimized for the OS. I repeated the recalculation test 3 times to verify that the results did not change due to the large performance difference I saw.

Note that changing precision did not affect the Solo's recalc time, but did dramatically reduce the time on the PowerBook. This seems to indicate that they are using two different default precision settings (number of decimal places on real numbers).

Excel 2.2MB Sheet

Test

PowerBook
G3/250

Gateway Solo
9100XL PII/266

Load Appl

3.44

3.22

Recalc

1:01.44

8.47

Recalc
(Prec.as Displ)

38.35

8.45




Summary:

In the final analysis, the PowerBook is a better performer in all graphics applications, even considering the dynamic memory allocation of Windows 98 and its large disk cache. Business applications, especially Excel showed an unbelievable difference that I can only guess is due to Microsoft's years of optimizing them for Windows and their intimate knowledge of the operating system and its compilers (they developed both).
The good news is that later Mac OS versions will feature more native code and better performance they say.
(Update - See the April 25th, 2005 main site news page. An OS X user still noted much better performance with PC excel vs Mac/Office Excel - even when running PC Excel under Virtual PC.)
If you missed part one of this review you'll want to read it to get a better feel for the design features and ease of use comparisons of the two machines, including a page with photos comparing them from various angles.


Back to XLR8YOURMAC.COM
Your Source for the best in CPU/SCSI/VIDEO card reviews, daily news, and more!

Copyright, 1998, all rights reserved.
No part of this site's content is to be reproduced in any form without permission.
All brand or product names mentioned here are properties of their respective companies.
Disclaimer: Users must read and are bound by the Site Terms & Conditions of Use.