Accelerate Your Mac! -Accelerate Your Mac's Reader review of the Newer Tech G3 333-400 CPU Upgrades Review

Click for iPhone accessories!

A Click on this Banner shows your site support to my Sponsors

Newertech G3 Upgrades Review
(333-400MHz CPU Cards and ZIF Upgrade)
Bryan William Jones
Review date: 2/15/99

Applications performance tests and comments

Intro | Benchmarks | Appl. Tests | Software/Controls | Documentation | Specifications | Summary

Applications/General use comments

Application performance included tests in Adobe Photoshop, Bryce 2, JMP (a statistics program by SAS Institute), and EASI/PACE 6.1 (this is a remote sensing/image processing package from PCIGeomatics)

Adobe Photoshop Tests:

All operations were performed on a 44.4 MB image in Adobe Photoshop 5.02 with 200 MB of memory allocated to the application. Times reported are for a complete series of operations encompassing image transitions, RGB conversions, and a standardized sequence of filters and mask operations.

The tests in Photoshop reveal that for the 400 MHz G3 card the savings in time for a 44.4 MB image is about 74 seconds per image. Or about a 12% increase in performance for Photoshop over the stock 9600/300. This time savings can easily add up to an hour or more per day for users that spend large amounts of time in Photoshop. As such they will have to look at the 400 MHz card, especially if they are currently using anything slower than a 9600/300.

Bryce 2 Tests:

Bryce 2 is a 3D design tool that allows one to easily perform rendering of 3D environments, however, it requires lots of horsepower to render results in times that don't keep you waiting through lunch. In keeping with Mike's previous tests of CPU upgrade cards with Bryce 2, sample scenes in the KAI folder "Magical Night I", and "Alexandria II" were used to gauge performance in Bryce 2. Again Bryce 2 was left with the default memory allocation of 16980k, and antialiasing was on. All times are reported in minutes.

In Bryce 2 rendering tests, the 400 MHz card proved itself to be from 37-40% faster than the stock 9600/300, with the G3/366 MHz card being 29-30% faster, and the G3/333 MHz card being about 20% faster.

JMP Tests:

JMP is a statistics program designed to make data exploration through statistics easier to visualize. Times reported are for a correlation matrix calculation of a large spreadsheet database. (A side note about the performance of computers in the last several years: I first had to make this calculation from this database when I had my stock MacIIsi. The exact benchmark for this computer is hard to tell as the IIsi finished the calculation sometime on the third day! The Daystar upgrade card I purchased for the MacIIsi specifically because of this project cut this to sometime in the second day. Compare those times to the following results for the latest in desktop computing technology.) Times reported are in seconds.

The first thing to note from this is that if you are running a 604e or less, and spend lots of time with statistics, or using spreadsheet applications like Excel, you absolutely need a G3. The increase in integer performance from the G3 with its backside cache is truly impressive. Even Newer's slowest card of the bunch, the G3/333 delivered a 54% improvement in performance in calculations done in JMP. The 366 MHz G3 card was 59% faster, and the 400 Mhz card was a whopping 64% faster than my stock 604e running at 300 MHz. From personal experience, these results are similar to any spreadsheet application like Microsoft's Excel, or statistics packages like SPSS, and Data Desk.

EASI/PACE Tests:

The specific test performed in this program was an image classification. This is a computationally intensive mathematical procedure originally developed by the Department of Defense (DOD) and NASA to classify pixels in a multilayered or multispectral image into categories based on hyperdimensional statistical calculations. (Read: for interpreting satellite images from space) For this review in the interest of time the operations were performed on a small multispectral image as larger images can exponentially expand the amount of computational time required. However, the relative performance of the upgrades can be readily inferred and extrapolated to larger image sizes.

These tests indicate an increase in performance over that of the 300 MHz 604e chip of 15% for the G3/333, 28% for the G3/366, and 36% for the G3/400 MHz G3 chips. As an aside, I have run this same multispectral image using the same algorithms on a 333Mhz Pentium II running Windows NT, and guess what? It took about 16 seconds or about 142% longer than the G3/333 to perform the same task.
The speed increase in EASI/PACE, Photoshop, the statistics applications, and another wonderful data visualization program, IDL from Research Systems will save me hours of work. All of which I will probably squander on the next application reviewed with these cards, Unreal. But before I go there, here is the place for my soapbox. These tests reinforce my hope that Apple will again start to look more seriously at the Macintosh in the science and engineering markets. I would much rather use the Macintosh (I can't begin to tell you how much time I have wasted not only trying to get Windows NT, or 9X to work with either new software or hardware, but also having to refocus my thoughts after having them scattered due to the Wintel "hammer" not working), but over the last several years, I have seen Apple steadily loose ground in terms of market share in these areas even though the combination of the MacOS and the RISC based PowerPC processor demonstrates time and time again its superiority in terms of performance and ease of use over that which the Microsoft/Intel consortium offers. [Note - see the latest Photoshop 5 Performance page for results that show a dual PII 450/NT 4 machine trails a most faster Macs in overall performance-Mike] Hopefully, once OS X comes out, Apple will have an OS that when combined with the PowerPC will give Apple the power, extendibility, and security of UNIX that scientists and engineers require with the sensibility and ease of use of the MacOS that makes people more productive. (I really hope that they implement distributed processing in OS X, or OS X server at some point.)

Unreal TimeDemo Tests

Finally since games are so popular (and end up convincing many people to purchase upgrade hardware to increase game performance) I also ran tests in Unreal to determine the increase in frame rates or frames rendered per second. Faster frame rates (up to a point) lead to better game performance and a more immersive environment. As mentioned before, my system contains a VoodooII card from Microconversions which is the most important factor in Unreal game performance on systems that have enough horsepower to feed pixels to the video card. The performance of the standard 9600 with the 604e chip running at 300 MHz was very good in Unreal with only occasional pauses in the rendering. With any of the G3 cards frame rates did not increase dramatically, but those irritating pauses disappeared completely. All tests in Unreal were performed in 800X600 with the HQ settings from the Unreal Time Demo Utility found in the 3D Game News section of this site.

As can be seen from this graph, the frame rates only increased about 6% from the 604e chip to the 400 MHz G3 card, but this lack of significant frame rate increases is most likely due to the VoodooII card's screaming fast ability to push pixels. [For comparisons of Unreal framerates with various video cards and CPUs, see the TimeDemo results page.-Mike]
The G3 desktop machine equipped with the Voodoo I accelerator card from Techworks reported scores of 16.28 frames per second in the stock configuration, and 17.62 frames per second with the Newer 366 MHz card. About an 8% increase in performance.
If you are really into games and you have a slower system, start with a voodoo card and then get a Newer G3/333. The combination can be had for around $1100-1200. This is still hundreds of dollars cheaper and faster than a new 300 MHz machine that does not have the benefit of Voodoo graphics. [The new Rage128 in the B&W G3's run Unreal very well however - see the Yosemite Game Performance page for comparisons. See our Upgrade or New system? article by Darin Ames for more on upgrade vs. new system purchase decisions.-Mike]

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 Newertech G3 Upgrade Review Pages
Intro | Benchmarks | Appl. Tests | Software/Controls | Documentation | Specifications | Summary
- or -
Back to WWW.XLR8YOURMAC.COM

Copyright © Michael Breeden, 1999.
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.


Newertech G3 Upgrades Review (333-400MHz CPU Cards and ZIF Upgrade) Bryan William Jones Review date: 2/15/99
Applications performance tests and comments
Intro \| Benchmarks \| Appl. Tests \| Software/Controls \| Documentation \| Specifications \| Summary
Applications/General use comments
Application performance included tests in Adobe Photoshop, Bryce 2, JMP (a statistics program by SAS Institute), and EASI/PACE 6.1 (this is a remote sensing/image processing package from PCIGeomatics)
Adobe Photoshop Tests:
All operations were performed on a 44.4 MB image in Adobe Photoshop 5.02 with 200 MB of memory allocated to the application. Times reported are for a complete series of operations encompassing image transitions, RGB conversions, and a standardized sequence of filters and mask operations. The tests in Photoshop reveal that for the 400 MHz G3 card the savings in time for a 44.4 MB image is about 74 seconds per image. Or about a 12% increase in performance for Photoshop over the stock 9600/300. This time savings can easily add up to an hour or more per day for users that spend large amounts of time in Photoshop. As such they will have to look at the 400 MHz card, especially if they are currently using anything slower than a 9600/300.
Bryce 2 Tests:
Bryce 2 is a 3D design tool that allows one to easily perform rendering of 3D environments, however, it requires lots of horsepower to render results in times that don't keep you waiting through lunch. In keeping with Mike's previous tests of CPU upgrade cards with Bryce 2, sample scenes in the KAI folder "Magical Night I", and "Alexandria II" were used to gauge performance in Bryce 2. Again Bryce 2 was left with the default memory allocation of 16980k, and antialiasing was on. All times are reported in minutes. In Bryce 2 rendering tests, the 400 MHz card proved itself to be from 37-40% faster than the stock 9600/300, with the G3/366 MHz card being 29-30% faster, and the G3/333 MHz card being about 20% faster.
JMP Tests:
JMP is a statistics program designed to make data exploration through statistics easier to visualize. Times reported are for a correlation matrix calculation of a large spreadsheet database. (A side note about the performance of computers in the last several years: I first had to make this calculation from this database when I had my stock MacIIsi. The exact benchmark for this computer is hard to tell as the IIsi finished the calculation sometime on the third day! The Daystar upgrade card I purchased for the MacIIsi specifically because of this project cut this to sometime in the second day. Compare those times to the following results for the latest in desktop computing technology.) Times reported are in seconds. The first thing to note from this is that if you are running a 604e or less, and spend lots of time with statistics, or using spreadsheet applications like Excel, you absolutely need a G3. The increase in integer performance from the G3 with its backside cache is truly impressive. Even Newer's slowest card of the bunch, the G3/333 delivered a 54% improvement in performance in calculations done in JMP. The 366 MHz G3 card was 59% faster, and the 400 Mhz card was a whopping 64% faster than my stock 604e running at 300 MHz. From personal experience, these results are similar to any spreadsheet application like Microsoft's Excel, or statistics packages like SPSS, and Data Desk.
EASI/PACE Tests:
The specific test performed in this program was an image classification. This is a computationally intensive mathematical procedure originally developed by the Department of Defense (DOD) and NASA to classify pixels in a multilayered or multispectral image into categories based on hyperdimensional statistical calculations. (Read: for interpreting satellite images from space) For this review in the interest of time the operations were performed on a small multispectral image as larger images can exponentially expand the amount of computational time required. However, the relative performance of the upgrades can be readily inferred and extrapolated to larger image sizes. These tests indicate an increase in performance over that of the 300 MHz 604e chip of 15% for the G3/333, 28% for the G3/366, and 36% for the G3/400 MHz G3 chips. As an aside, I have run this same multispectral image using the same algorithms on a 333Mhz Pentium II running Windows NT, and guess what? It took about 16 seconds or about 142% longer than the G3/333 to perform the same task. The speed increase in EASI/PACE, Photoshop, the statistics applications, and another wonderful data visualization program, IDL from Research Systems will save me hours of work. All of which I will probably squander on the next application reviewed with these cards, Unreal. But before I go there, here is the place for my soapbox. These tests reinforce my hope that Apple will again start to look more seriously at the Macintosh in the science and engineering markets. I would much rather use the Macintosh (I can't begin to tell you how much time I have wasted not only trying to get Windows NT, or 9X to work with either new software or hardware, but also having to refocus my thoughts after having them scattered due to the Wintel "hammer" not working), but over the last several years, I have seen Apple steadily loose ground in terms of market share in these areas even though the combination of the MacOS and the RISC based PowerPC processor demonstrates time and time again its superiority in terms of performance and ease of use over that which the Microsoft/Intel consortium offers. [Note - see the latest Photoshop 5 Performance page for results that show a dual PII 450/NT 4 machine trails a most faster Macs in overall performance-Mike] Hopefully, once OS X comes out, Apple will have an OS that when combined with the PowerPC will give Apple the power, extendibility, and security of UNIX that scientists and engineers require with the sensibility and ease of use of the MacOS that makes people more productive. (I really hope that they implement distributed processing in OS X, or OS X server at some point.)
Unreal TimeDemo Tests
Finally since games are so popular (and end up convincing many people to purchase upgrade hardware to increase game performance) I also ran tests in Unreal to determine the increase in frame rates or frames rendered per second. Faster frame rates (up to a point) lead to better game performance and a more immersive environment. As mentioned before, my system contains a VoodooII card from Microconversions which is the most important factor in Unreal game performance on systems that have enough horsepower to feed pixels to the video card. The performance of the standard 9600 with the 604e chip running at 300 MHz was very good in Unreal with only occasional pauses in the rendering. With any of the G3 cards frame rates did not increase dramatically, but those irritating pauses disappeared completely. All tests in Unreal were performed in 800X600 with the HQ settings from the Unreal Time Demo Utility found in the 3D Game News section of this site. As can be seen from this graph, the frame rates only increased about 6% from the 604e chip to the 400 MHz G3 card, but this lack of significant frame rate increases is most likely due to the VoodooII card's screaming fast ability to push pixels. [For comparisons of Unreal framerates with various video cards and CPUs, see the TimeDemo results page.-Mike] The G3 desktop machine equipped with the Voodoo I accelerator card from Techworks reported scores of 16.28 frames per second in the stock configuration, and 17.62 frames per second with the Newer 366 MHz card. About an 8% increase in performance. If you are really into games and you have a slower system, start with a voodoo card and then get a Newer G3/333. The combination can be had for around $1100-1200. This is still hundreds of dollars cheaper and faster than a new 300 MHz machine that does not have the benefit of Voodoo graphics. [The new Rage128 in the B&W G3's run Unreal very well however - see the Yosemite Game Performance page for comparisons. See our Upgrade or New system? article by Darin Ames for more on upgrade vs. new system purchase decisions.-Mike]

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 Newertech G3 Upgrade Review Pages Intro \| Benchmarks \| Appl. Tests \| Software/Controls \| Documentation \| Specifications \| Summary - or - Back to WWW.XLR8YOURMAC.COM
Copyright © Michael Breeden, 1999. 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.