Accelerate Your Mac! 3dfx Mac Voodoo5 5500 PCI Review

3D Applications:
Update: Lightwave 3D v6 Issues: As readers have also learned, the Voodoo5 with the current drivers (v1.01) has texture and other image problems in Lightwave 3D v6. For samples of the problems see these screenshots at www.exchange.co.uk/lightwave/video.htm (FYI - that page no longer exists) Note: I just received word the problem report there was from a PowerCenter Pro owner who also had the same problems with a dual G4/450 (which is a supported system). A reader also reported the Voodoo5 displayed no textures in smooth shaded mode in the Sept. 6th www.xlr8yourmac.com news page.

Hopefully the Lightwave problems can be addressed in a future update, but as of early September, 2000 the Voodoo5 card would not be recommended for Lightwave 3D users. I now wonder what card was used during the Macworld NY 2000 3dfx booth Lightwave v6 demo (I assumed it was a voodoo5). Granted most of the time that I watched the demo the work being done was modeling in wireframe mode.

While I waited for Lightwave 3D 6.0 to arrive, I ran informal tests in an application that supported QD3D/RAVE like my old favorite Infini-D 4.01. I set all views to to show gouraud shaded objects (using models from the Ch. 7 tutorial file). Granted this scene does not have a lot of objects (table, bottle, globe and 3d logo) but performance on the G4/500 with the Voodoo5 was very good. After moving an object by the time it snapped into position it was fully shaded again (with highlights from light sources). In my experience with Infini-D 4, the speed of the machine it's running on (CPU speed) is a major factor in interactivity.

Even though Infini-D 4.01 is a very old application now, I saw no artifacts or other image quality issues with the Voodoo5.

CineBench 2000 v1.0 Results:
(Update) On 9/14/2000 Maxon released an updated Cinebench 2000 which I used to test performance of the Voodoo5 against an ATI Rage128 PCI and Rage128 Pro AGP. (For more info see www.cinebench.com. They have details on the benchmark and how to read the results.)

Note the disappointing results with the Voodoo5's OpenGL mode performance. It was slower than Cinema's software mode. Software mode reported very slightly higher results with the Voodoo5 installed than with the Rage128.

Quickdraw 3D RAVE Benchmarks:
As noted previously, RaveBench 1.1.1 image quality tests failed with the Voodoo5 (lack of texture mapping) so results are not show with that benchmark.

Walker 1.1 Tests:
As I've done in most all my graphics cards reviews over the years, I ran tests in Lightwork's Walker 3D viewer with the 2D screen set to 1280x1024, millions colors (Walker uses a 350x350 default window size). I used the highest polygon count scene provided with Walker - the 49,000 polygon 'Corridor' scene . (Note: Walker does not seem to be available at their website anymore.)

Although the graph is sorted by the highest Max framerates, in my opinion the most important figure is the lowest framerate during the test, as that indicates how the card handles the toughest part of the scene.

Benchmarks are not the final word, especially those that test primitives, which vary in their frequency of use by 3D applications. Walker for instance does not seem to use many of the primitives tested by RaveBench.

3D Apps. Performance Summary: In the limited testing I was able to do so far, performance seems good. RAVE will likely to not be supported in OS X and since OpenGL is becoming the Mac 3D API standard, I'm not personally that concerned about the RAVE performance of the Voodoo5 (especially considering the current driver state). However due to the fact RaveBench didn't show textures with the Voodoo5 (although the Voodoo3 did) and the disappointing Walker scores I think the RAVE driver has room for improvement. On the plus side my old Infini-D 4 application didn't seem to have any problems. At this stage of the driver development it's too early to tell what RAVE performance could be with later updates. When Lightwave 3D v6 arrives for testing, I'll be able to comment more on OpenGL 3D application performance.

I know that game performance is important in a graphics card to many of you, so the next page shows results I had running several popular 3D games.


Review: 3dfx Mac Voodoo5 5500 PCI Graphics Card Review date: 7/26/2000 (updated 9/6/2000 for Lightwave v6 issues updated 9/16/2000 for Cinebench 2000 results)
3D Benchmarks and Application Tests Results
Intro \| 2D Performance \| 3D Performance \| Game Performance \| Movie Playback \| Controls/Design \| Summary
3D Benchmark Performance
Until my copy of Lightwave 3D arrives, evaluation of 3D performance was based on tests with benchmarks and simple testing with my old copy of Infini-D 4. I could not provide RaveBench comparisons as the Voodoo5 didn't show texture mapping in the tests. (For an explanation of RaveBench's tests, see my Illustrated Guide to RaveBench published in 1998.) Since several RAVE game titles apparently run fine, I can only suspect RaveBench has some issue with the Voodoo5. Lightwork's Walker benchmarks did run fine and results are shown below. Although most owners would never run RAVE mode in Unreal with a 3dfx card, attempts to do so had severe image problems even though I had set 'non-ATI' to True in the Rave section of the Unreal.ini file. Since glide mode looks better than Rave Unreal in my opinion, it's no big loss but just a sign that not all RAVE applications may be compatible. If in doubt, contact 3dfx to verify your specific application. 3D Applications: Update: Lightwave 3D v6 Issues: As readers have also learned, the Voodoo5 with the current drivers (v1.01) has texture and other image problems in Lightwave 3D v6. For samples of the problems see these screenshots at www.exchange.co.uk/lightwave/video.htm (FYI - that page no longer exists) Note: I just received word the problem report there was from a PowerCenter Pro owner who also had the same problems with a dual G4/450 (which is a supported system). A reader also reported the Voodoo5 displayed no textures in smooth shaded mode in the Sept. 6th www.xlr8yourmac.com news page. Hopefully the Lightwave problems can be addressed in a future update, but as of early September, 2000 the Voodoo5 card would not be recommended for Lightwave 3D users. I now wonder what card was used during the Macworld NY 2000 3dfx booth Lightwave v6 demo (I assumed it was a voodoo5). Granted most of the time that I watched the demo the work being done was modeling in wireframe mode. While I waited for Lightwave 3D 6.0 to arrive, I ran informal tests in an application that supported QD3D/RAVE like my old favorite Infini-D 4.01. I set all views to to show gouraud shaded objects (using models from the Ch. 7 tutorial file). Granted this scene does not have a lot of objects (table, bottle, globe and 3d logo) but performance on the G4/500 with the Voodoo5 was very good. After moving an object by the time it snapped into position it was fully shaded again (with highlights from light sources). In my experience with Infini-D 4, the speed of the machine it's running on (CPU speed) is a major factor in interactivity. Even though Infini-D 4.01 is a very old application now, I saw no artifacts or other image quality issues with the Voodoo5. CineBench 2000 v1.0 Results: (Update) On 9/14/2000 Maxon released an updated Cinebench 2000 which I used to test performance of the Voodoo5 against an ATI Rage128 PCI and Rage128 Pro AGP. (For more info see www.cinebench.com. They have details on the benchmark and how to read the results.) All tests were run at 1024x768, millions color mode. Longer bars are faster. Note the disappointing results with the Voodoo5's OpenGL mode performance. It was slower than Cinema's software mode. Software mode reported very slightly higher results with the Voodoo5 installed than with the Rage128. Quickdraw 3D RAVE Benchmarks: As noted previously, RaveBench 1.1.1 image quality tests failed with the Voodoo5 (lack of texture mapping) so results are not show with that benchmark. Walker 1.1 Tests: As I've done in most all my graphics cards reviews over the years, I ran tests in Lightwork's Walker 3D viewer with the 2D screen set to 1280x1024, millions colors (Walker uses a 350x350 default window size). I used the highest polygon count scene provided with Walker - the 49,000 polygon 'Corridor' scene . (Note: Walker does not seem to be available at their website anymore.) Although the graph is sorted by the highest Max framerates, in my opinion the most important figure is the lowest framerate during the test, as that indicates how the card handles the toughest part of the scene. Benchmarks are not the final word, especially those that test primitives, which vary in their frequency of use by 3D applications. Walker for instance does not seem to use many of the primitives tested by RaveBench. 3D Apps. Performance Summary: In the limited testing I was able to do so far, performance seems good. RAVE will likely to not be supported in OS X and since OpenGL is becoming the Mac 3D API standard, I'm not personally that concerned about the RAVE performance of the Voodoo5 (especially considering the current driver state). However due to the fact RaveBench didn't show textures with the Voodoo5 (although the Voodoo3 did) and the disappointing Walker scores I think the RAVE driver has room for improvement. On the plus side my old Infini-D 4 application didn't seem to have any problems. At this stage of the driver development it's too early to tell what RAVE performance could be with later updates. When Lightwave 3D v6 arrives for testing, I'll be able to comment more on OpenGL 3D application performance. I know that game performance is important in a graphics card to many of you, so the next page shows results I had running several popular 3D games.

Index of 3dfx Voodoo5 5500 PCI Review Intro \| 2D Performance \| 3D Performance \| Game Performance \| Movie Playback \| Controls/Design \| Summary - or - Back to www.XLR8YOURMAC.com
Copyright © 2000, all rights reserved. All brand or product names mentioned here are properties of their respective companies. Site terms and conditions of use.