Navigate:

Advice
Beginners
BIOS Guide
CPUs
Links
Mainboards
Memory
Network
Storage
Video/Sound Cards

Contact
Forum
SiteMap
Sponsors
WebNews
Home
. .

Prices:

Mainboards

ABIT
ASUS
Chaintech
Shuttle
Soyo
Tyan

CPU
Intel
P4 2.4C-800
P4 2.6C-800
P4 2.8C-800
P4 3.0-800
P4 3.2-800

AMD
AthlonXP
XP 1700+
XP 2000+
XP 2400+
XP 2500+
XP 2700+
XP 3000+
XP 3200+

Athlon64
Athlon64 3200+
Athlon64 FX-51

Opteron
Opteron 240
Opteron 242
Opteron 244
Opteron 246

Memory

Corsair
Crucial
Kingston
Mushkin
OCZ

What are you
shopping for?



































































































































































LOSTCIRCUITS

SHORTCUTS:
top page
specs
tech. details
setup / 2D
SS7 D3D
SS7 OpenGL
SS7 CAD
Athlon D3D
Athlon OpenGL/CAD
conclusion
 3DLabs Permedia3 Create!
Mind Over Matter
(Review by MS, October 2, 1999)

CAD applications

Over all these gaming benchmarks, it is easy to forget that the Create! was never intended to be a pure gaming card. On the contrary, its strength lies in the performance in CAD applications. There are a number of CAD benchmarks, from Caligari Truespace to Indy3D and, of course, whatever is supplied with a review kit could be very selective. But let’s take a look at what we have. And don't say that CAD and SS7 don't go together.

Caligary Truespace 4.2

Truespace is heavily optimized for Intel’s SSE instructions and thus the numbers are not overwhelming in the first place. However, here is it where the tables are turned and the Create! pulls ahead of the Viper770 whereas the Rage Fury does remarkably well, at least in D3D mode.

In OpenGL, the Ati Rage Fury did not run correctly, showing only skeletons and surfaces bar textures instead of the real environment while the Viper770 achieved slightly better frame rates, with the drawback of line white-outs trailing behind every moved object. No question, therefore, that the Create! is the winner here.

Frame rates are shown for the Create! (C), the Viper770 Ultra (V) and the Rage 128 Fury (A) running either D3D (light bars) or OpenGL (dark bars) mode. Note that the Fury does not run correctly in OpenGL. All benches were run in 16 bit color.


Lightscape Viewperf 6.1

Lightscape is a mixture of ray tracing and 3D representation benchmarks. While both Viper770 and Create! ran correctly at 32bit color depth, the Viper770 gave incorrect results in 16 bit color mode. The Rage Fury ran correctly only in 256 color mode, at greater depth only the blue channel was displayed and the results will not be considered here. Again, the Create! beats the Viper770 but not by too much when only the frame rates are considered. However, the Viper770 display consistently showed minor pixel blackouts which need to be weighed against the frame rates.

Frame rates in the different applications in Viewperf 6.1 for the Create! in 16 and 32 bit color and the the Viper770 Ultra in 32 bit color. 1: AWadus / 2: DRV 06 / 3: DX 05 / 4: Light 03. Please note that the frame rates for Light 03 are multiplied by a factor of 10 to fit them into the graph.

Lightscape

What we are looking at is a three dimensional rotation of interior design. The demo comprises 4 environments out of which one file gave a "corrupted file" error message. Options are a simple rotation, tumbling and walk through. The demo ran without problems at all resolutions (640x480, 800x600 and 1024x768). Frame rates are given for 640x 480 in 32 bit color depth.

Benchmark Create! Viper 770 Rage 128 Fury
2 Rooms 4.24 4.25 2.51
Chamber 16.71 18.20 11.06
Hall 18.20 18.34 11.33

Treemark

Last not least, there is the recently released treemark video benchmark from nVidia. Knowing where it comes from, one might expect some optimizations for the TNT(2) graphics adapters. As standard the "simple" version was run with 1272 branches and 1273 leaves (compplex: 4567 / 4568). The Fury takes off here but does not display the branches, thus creating a ghostly image of a cloud of leaves hovering above the ground and, thus, falls out of the race. Not surprisingly, the Viper770 is faster here but the performance difference is only about 4% or 9.4% respectively.

Benchmark Create! Viper 770
Treemark simple 4.92732 5.12007
Treemark complex 1.0664 1.16746

A Quick Summary

The Permedia3 Create! offers stunning visual quality, and works exceptionally well with the SS7 platform / K6-III. While D3D games are running at overall more than acceptable speed and superior detail and color fidelity, the Create! does have problems delivering more than average frame rates in the Quake family of games (including HalfLife, not shown).

On a side note, the K6-2 appears not to be the most suitable partner for the Create! at least not for CAD applications since benchmarks were running so painfully slow that we decided not to further pursue this combination.

On Celeron systems, the Create shows about the same performance as on the SS7 platform / K6-III at equal clock speed. That is, in the Intel optimized Caligari Truespace, the performance increase is in the range of 25% over the K6-III whereas, interestingly enough, the K6-III outruns the Celeron in QuakeII by about 20% independent of the resolution. A clear and well needed example for what was missed in the past by ignoring 3DNow! and AMD's write combine buffers and how good the Permedia3 Create! drivers and OpenGL ICDs in fact are. The Celeron benches were run at 3x150 MHz on the Shuttle AV61 which will be reviewed in full shortly.

Next Page:    => next =>

Click Here!

If you enjoyed reading this article and found it useful, please consider making a small donation to LostCircuits.
Thank you!

General disclaimer: This page only reflects the author's personal opinion and assumes no responsibility whatsoever regarding any of the contents or any damages that may occur explicitly or implicitly from reading the contents of this site. All names and trademarks mentioned in this review are the exclusive property of the respective parent companies.
All contents of this site are protected by international copyright laws. Reproduction of the contents even in parts is not allowed except after written permission by the author and referral to this site.
Copyright 1998 - 2008 LostCircuits