iXBT Labs - Computer Hardware in Detail

Platform

Video

Multimedia

Mobile

Other

Brief Account of Testing Modern 3D Accelerators in DOOM III

Part 2: middle and low-end video cards.

August 25, 2004








Contents

  1. Introduction
  2. Video cards' features
  3. Testbed configurations, benchmarks
  4. Test results: Performance
  5. Test results: Quality
  6. Conclusions



We proceed with our performance tests of video cards of different price levels in DOOM III. Last time we benchmarked the high end 3D accelerators, very expensive video cards of the RADEON X800 and GeForce 6800 series.

Today we shall present weaker and cheaper solutions. It goes without saying that if even the most powerful accelerators demonstrated not more than 50-60 fps with the Ultra High and High Quality settings, there is absolutely no sense in testing admittedly weaker cards with these settings.

That's why we divided our video cards into the low and middle end sections and tested them in DOOM III with low and medium settings correspondingly. The only exception is made for RADEON 9800 PRO and GeForce FX 5900, which can manage the High Quality settings, so there will be two data sets for these video cards: Medium and High Quality.

We have also changed the range of resolutions. Usually we use resolutions from 1024x768 to 1600x1200. In this case it's silly to use these resolutions for an obvious reason, that's why we limited ourselves to two resolutions: 800x600 and 1024x768.

What's the difference between the Ultra High, High, Medium, and Low modes? Let's have a look at the table:

Game settings Quality settings
Ultra High High Medium Low
Precompressed textures (image_usePrecompressedTextures) No (0) Yes (1) Yes (1) Yes (1)
S3TC usage (image_useCompression) No (0) Yes (1) Yes (1) Yes (1)
Object details (com_machineSpec) Highest (3) Average (2) Low (1) Lowest (0)
Ignore high quality textures (image_ignoreHighQuality) No (0) No (0) Yes (1) Yes (1)
Compress normal maps (image_useNormalCompression) No (0) No (0) Yes (2) Yes (2)
Anisotropic filtering (image_anisotropy) AF8x (8) AF8x (8) No (1) Denied (0)
Downsize bump maps (image_downSizeBump) No (0) No (0) No (0) Yes (1)
Reduce specular effects (image_downSizeSpecular) No (0) No (0) No (0) Yes (1)
Reduce quality (image_downSize) No (0) No (0) No (0) Yes (1)


I suppose that you all understand how much the image quality suffers in Low Quality. However, at the end of the article you can see screenshots demonstrating the effect.

And now it's high time to introduce our contenders.



Video Cards



NVIDIA GeForce 5900
Albatron Gigi GeForce FX 5900, 128MB, 400/850 MHz, 4x2 (8x0)


NVIDIA GeForce 5900 XT
Prolink PixelView GeForce 5900 XT Golden, 128MB, 390/700 MHz, 4x2 (8x0)


NVIDIA GeForce 5700 Ultra DDR3
Albatron Gigi GeForce FX 5700 Ultra, 128MB DDR3, 475/950 MHz, 4x1


NVIDIA GeForce 5700
Leadtek WinFast A360 TD, 128MB, 425/500 MHz, 4x1


NVIDIA GeForce 5700 LE
Galaxy GeForce FX 5700LE, 128MB, 250/400 MHz, 4x1


NVIDIA GeForce 5500
ASUS V9520/TV, 128MB, o/c up 270/400 MHz, 2x2


NVIDIA GeForce 5200
ASUS V9520/TV, 128MB, 250/400 MHz, 2x2


ATI RADEON 9800 PRO
HIS Excalibur RADEON 9800 PRO IceQ, 128MB, 380/680 MHz, 8x1


ATI RADEON 9800
Sapphire Atlantis RADEON 9800, 128MB, 325/590 MHz, 8x1


ATI RADEON 9600 XT
Sapphire Atlantis RADEON 9600 XT, 128MB, 500/600 MHz, 4x1


ATI RADEON 9600
Sapphire Atlantis RADEON 9600, 128MB, 325/400 MHz, 4x1


ATI RADEON 9600SE
Sapphire Atlantis RADEON 9600SE, 128MB 64bit, 325/400 MHz, 4x1


ATI RADEON 9550
HIS Excalibur RADEON 9550, 128MB, 250/400 MHz, 4x1


ATI RADEON 9200
Gigabyte RADEON 9200, 128MB, 250/400 MHz, 4x1




I suppose there is no point in commenting on these cards, all of them were tested in our lab.

Installation and Drivers

Testbed configurations:

  • Athlon 64 3200+ based computer
    • AMD Athlon 64 3200+ (L2=1024K) CPU
    • ASUS K8V SE Deluxe mainboard based on VIA K8T800
    • 1 GB DDR SDRAM PC3200
    • Seagate Barracuda 7200.7 80GB SATA HDD

  • Operating system – Windows XP SP1; DirectX 9.0c
  • Monitors: ViewSonic P810 (21") and Mitsubishi Diamond Pro 2070sb (21").
  • ATI drivers v6.476 (CATALYST 4.9beta); NVIDIA drivers v61.77.

VSync is disabled.

Note that we used a beta version of CATALYST 4.9, which contains ICD OpenGL (alpha version) rewritten from scratch. The driver is still in the development stage, so the current ATI test results can be considered preliminary.

Test results: performance comparison

We used the following test applications:

  • DOOM III (id Software/Activision) – OpenGL, test settings are specified in the tests, demo SAM3

If you want to get the demo-benchmarks, which we use, contact me at my e-mail. I'm warning you that the demo-benchmark in DOOM III is 150-200MB(!), 25MB compressed.

Concerning the benchmark, we did not use Demo1 installed with the game but recorded our own demo instead - SAM3 (timedemo sam3 console command is used for the test) This is done to avoid possible optimizations for the existing game benchmark.

Important information: The maximum FPS in the game is limited to 60 frames per second, but when we run timedemo this limitation is lifted.

AF8x mode in the diagrams means Anisotropic 8x Quality



Middle end accelerators








Low end accelerators








Summary table of performance differences






Test results: quality comparison

I shall not provide the quality differences between the cards today (you can see them in the previous article).

But to show the visual difference between the four above-mentioned quality settings in the game is quite reasonable.

Quality Example 1
Ultra High



High



Medium



Low



Quality
Example 2
Ultra High



High



Medium



Low



Quality
Example 3
Ultra High



High



Medium



Low



Quality
Example 4
Ultra High



High



Medium



Low





We can see that the Low settings heavily damage the quality. But have a look at the game speed even in this mode...

Conclusions

Thus, on the whole:

There is obviously no sense in comparing the competing video cards (ATI/NVIDIA) with each other, as they are on the level.

The only exception is the defeat of RADEON 9200 by its competitor GeForce FX 5200. But then RADEON 9550 is very good.

The question is why the top ATI products catastrophically lose to those from NVIDIA, while their cheap solutions are on the level? What's the matter? I guess we partially answered this question earlier when we assumed that the key to NVIDIA success was the proprietary technology for shadow processing, mind that we didn't disable shadows even in Low Quality. Thus, this effect is due to UltraShadow and fast stencil belonging only to the FX 5900 series and higher. Disarmed FX 5700/5500/5200 couldn't shoot ahead. Plus there is a possibility that NVIDIA shader optimization is implemented only in top video cards.

Now about the speed itself: we can see that even in the Medium/Low mode at 800x600 the gameplay demonstrated by some video cards is below critical. It means that the owners of RADEON 9200 (including 9600SE) and FX 5200/5700LE will have to upgrade their video cards if they want to play DOOM III. There is no other way out. You can still decrease the quality level, but watching blurred walls in 2004 is a shame.

Conclusion: The owners of low end (and some middle end) video cards will have to part with their cards and buy faster accelerators. The example of DOOM III is illustrative: the future will present more and more games requiring more considerable performance capacity from video cards.

Andrey Vorobiev (anvakams@ixbt.com)

10 August, 2004

Write a comment below. No registration needed!


Article navigation:



blog comments powered by Disqus

  Most Popular Reviews More    RSS  

AMD Phenom II X4 955, Phenom II X4 960T, Phenom II X6 1075T, and Intel Pentium G2120, Core i3-3220, Core i5-3330 Processors

Comparing old, cheap solutions from AMD with new, budget offerings from Intel.
February 1, 2013 · Processor Roundups

Inno3D GeForce GTX 670 iChill, Inno3D GeForce GTX 660 Ti Graphics Cards

A couple of mid-range adapters with original cooling systems.
January 30, 2013 · Video cards: NVIDIA GPUs

Creative Sound Blaster X-Fi Surround 5.1

An external X-Fi solution in tests.
September 9, 2008 · Sound Cards

AMD FX-8350 Processor

The first worthwhile Piledriver CPU.
September 11, 2012 · Processors: AMD

Consumed Power, Energy Consumption: Ivy Bridge vs. Sandy Bridge

Trying out the new method.
September 18, 2012 · Processors: Intel
  Latest Reviews More    RSS  

i3DSpeed, September 2013

Retested all graphics cards with the new drivers.
Oct 18, 2013 · 3Digests

i3DSpeed, August 2013

Added new benchmarks: BioShock Infinite and Metro: Last Light.
Sep 06, 2013 · 3Digests

i3DSpeed, July 2013

Added the test results of NVIDIA GeForce GTX 760 and AMD Radeon HD 7730.
Aug 05, 2013 · 3Digests

Gainward GeForce GTX 650 Ti BOOST 2GB Golden Sample Graphics Card

An excellent hybrid of GeForce GTX 650 Ti and GeForce GTX 660.
Jun 24, 2013 · Video cards: NVIDIA GPUs

i3DSpeed, May 2013

Added the test results of NVIDIA GeForce GTX 770/780.
Jun 03, 2013 · 3Digests
  Latest News More    RSS  

Platform  ·  Video  ·  Multimedia  ·  Mobile  ·  Other  ||  About us & Privacy policy  ·  Twitter  ·  Facebook


17

Copyright © Byrds Research & Publishing, Ltd., 1997–2011. All rights reserved.