nV News Deals Shop Archive Search Files Forum Feed Articles IRC Chat GeForce.com

Search Site
Ads by Google
News Room
GeForce Technology
DirectX 11
3D Vision
3D Vision Surround
GeForce GTX 580
GeForce GTX 570
GeForce GTX 560 Ti
GeForce GTX 480
GeForce GTX 465
GeForce GTX 460
GeForce GTS 450
GeForce GTX 295
GeForce GTX 280
GeForce GTX 260
GeForce GT 240
GeForce 9800 GTX
GeForce 9800 GX2
GeForce 9600 GT
GeForce 8800 Ultra
GeForce 8800 GTX
GeForce 8800 GTS
GeForce 8800 GT
GeForce 8600 GTS
GeForce 8500 GT
GeForce 7950 GX2
GeForce 7950 GT
GeForce 7900 GTX
GeForce 7900 GS
GeForce 7800 GTX
Watercooling Project
My Book 500GB
Raptor Hard Drive
Guide To Doom 3
EVGA Stuff
EVGA Precision
GPU Voltage Tuner
OC Scanner
SLI Enhancement
Reviews and Awards
Benchmark Reviews
GeForce Italia
GPU Review
Hardware Pacers
Overclocker Cafe
PC Extreme
PC Gaming Standards
PhysX Links & Info
3DChip (German)
Creative 3D Blaster 5 FX5900 Review - Page 3 Of 3


Before I made the move to a GeForce FX, I had been using a GeForce3 and GeForce4. Obtaining acceptable performance from newer games on the GeForce3 was challenging and I often played at lower resolutions in addition to tweaking image quality in the driver control panel. My first line of attack was to modify the image setting to Performance or High Performance. I also tinkered with lowering the mipmap detail level and rarely used antialiasing since supersampling caused a large impact on performance.

The GeForce4 Ti 4200 was a great leap forward from the GeForce3, especially in antialiasing performance, which was based on multisampling. With the GeForce4 Ti 4200, it wasn't neccessary to compromise image quality as I had done on the GeForce3. I could game at 1024x768 and higher resolutions and found performance to be acceptable in many games with 2X antialiasing. Most of the fine-tuning consisted of tweaking in-game settings and I could always disable frame-rate killing features like shadows if I needed a boost in performance. However, the combination of using antialiasing and anisotropic texture filtering caused performance to lag.

Now that you know where I am coming from, the 2D image quality of the GeForce FX and the Creative FX5900 were a major improvement over the GeForce3. With the GeForce3, I generally kept the Windows resolution at 1024x768 since image quality degraded sharply at higher resolutions. But the continued improvement in 2D image qualty since the GeForce3 has allowed me to comfortably move up to higher resolutions on the GeForce FX.

Certain aspects of 3D image quality on the GeForce FX continue to be adjusted through driver revisions. More specifically, the behaviour of anisotropic texture filtering under Direct3D seems to have dropped off in quality a bit when compared to the GeForce4. However, the anisotropic filtering quality of the GeForce FX continues to surpass the competition - see this comparison.

The two images below are used to illustrate the different levels of anisotropic filtering that are available on the 3D Blaster 5 FX5900. The first image shows a scene from the OpenGL-based game Call of Duty.

Scene From Call of Duty
Scene From Call of Duty

The second image below was created by cutting out the exact same section from the scene above with different levels of anisotropic filtering enabled. Each new section was joined to the previous section, which provides us with an effective side-by-side comparsion.

GeForce FX Anisotropic Filtering Comparison
0X To 8X Anisotropic Filtering

The far left image section contains no AF (trilinear filtering), followed by 2X, 4X, and 8X AF. In most cases, 4X AF is sufficient on the GeForce FX as the difference between 4X and 8X AF is often unnoticable.


I'd like to thank Mike for helping me out with this part of the review as I needed a way to effectively show variations in image quality. This information was originally intended to appear in a follow-up to his GeForce FX 5950 Ultra preview, which he will expand upon in an upcoming review of the XFX GeForce FX 5950 Ultra.

Again, from the Call of Duty, the table below uses Javascript to display high quality PNG images that were captured with different levels of antialiasing and anisotropic filtering enabled. Each image was created by cutting out the same section from the full size 1024x768 image that was generated by Call of Duty's in-game screen capture. Above and to the far right of the image are headings for the anisotropic and antialiasing modes that are displayed by this program. The two lines are labeled as:

  • 0X AA / AF = Displays images that contain no antialiasing and either no, 2X, 4X, or 8X anisotropic filtering depending on the selection.

  • 8X AF / AA = Displays images that contain 8X anisotropic filtering and either no, 2X, 4X, or 8X antialiasing depending on the selection.

Clicking on a selection will reveal the image with the chosen method of anisotropic filtering and antialiasing enabled. Be patient while the image is downloaded from the server as they average around 300 KB. Once all of the images have been selected and downloaded to your PC and saved in your cache, you'll be able to quickly switch between them to see the differences. The default image contains no AA and no AF.

GeForce FX Image Quality Comparison
0X AA / AF    =    0X    2X    4X    8X
8X AF / AA    =    0X    2X    4X    8X

To better see the effects of antialiasing and anisotropic filtering, center the image on the screen and position your eyes about 12-16 inches from the monitor while comparing the images.

With the 3D Blaster 5 FX5900, I've been able to enable combinations of antialiasing and anisotropic filtering modes. Even 8X antialiasing, which looks incredible, can be used at 1024x768 more often in older games. Being able to enable these features at higher resolutions makes the GeForce FX light years ahead of the GeForce3 and GeForce4.


At the stock speed of 400MHz/700MHz, the 3D Blaster 5 FX5900 is a strong performer. But at some point most of us will experiment with increasing the core and memory clock speeds. For me, overclocking consists of increasing the core and memory speeds individually in increments of 5MHz. Then I run the 3DMark03 game tests to completion while examining the display for artifacts.

After having determined the maximum clock speeds individually, I combine them and continue testing with 3DMark03. At this point, if I don't end up with a successful run, I alternate decreasing the core and memory speed in 5Hz increments until a successful run is achieved. With the 3D Blaster 5 FX5900, I ended up at 425MHz/750MHz. Let's see how this translates into added performance.

Overclocking Results

Performance increases range from 3% in GunMetal to 14% in Halo, which averages out close to the 10% increase I've been accustomed to getting on other graphics cards that I have used.


The 256-bit memory interface is a major bonus on the 3D Blaster 5 FX5900, which really helps out with antialiasing performance. The latest ForceWare drivers have improved DirectX 9 based shader performance across the board, bringing the level of playability closer to where it should be when compared to the competition. Kudos to NVIDIA for the increase in Halo playability in particular.

The cooling fan wasn't very loud and any noise is easily overshadowed by case fans. The overclocking potential of the 3D Blaster 5 FX5900 was not dependent on the power supply as all attempts at overclocking ended up with similar results. The graphics card was tested with a 400W generic PSU, a 450W PowerMagic PSU, and a 525W PowerMagic PSU.

Making use of a DVI-Analog converter has given me an opportunity to experiment with using two monitors under the control of NVIDIA's nView. This was my first experience with a dual monitor setup and I was impressed with the capability of nView, which NVIDIA has turned into a quality feature.

Including Gun Metal was a nice touch. Most reviews rate Gun Metal in the 80-90% category and I found the game offered solid gameplay. Gun Metal was developed to expose the strengths of the GeForce FX architecture and is enjoyable, replayable, and impressive.

On the downside, OpenGL performance has not increased much over the GeForce FX 5600 Ultra when AA and AF are not used. Likewise, expect minimal increases in performance in CAD/CAM applications.

I was disappointed with the limited overclocking results, although the type of memory used doesn't offer much headroom for improvement. The physical size of the graphics card may be an issue if you had trouble with getting a GeForce3 to fit in your system.

The 3D Blaster 5 FX5900 is a great deal and is complimented by stable and feature-rich drivers from NVIDIA. It also has the looks with a red PCB and a gold plated heatsink. At the time this article was published, we found the 3D Blaster 5 FX5900 in stock at Komplett UK.

Price List @ Komplett UK (inc. VAT)

Kudos go out to Creative for distributing a great bang-for-the-buck graphics card package. A word of thanks also goes out to Eoin Layden at Creative Labs Europe for his support and keeping in touch with nV News.

Back to nV News

Last Updated on December 18, 2003

Table of Contents


nV News - Copyright © 1998-2014.
Search Products

Ads by Casale
Metro: Last Light
PlanetSide 2
Miscellaneous Links
AutoDesk 123 Design
Build Your Gaming PC
FPS vs. Frame Time
Free Games And MMOs
GeForce SLI Technology
HPC For Dummies
PC Game Release Dates
Play Classic PC Games
Steam Hardware Survey
Video Game Designers
TechTerms Dictionary
GPU Applications
AMD GPU Clock Tool
AMD System Monitor
EVGA OC Scanner
EVGA Precision
EVGA Voltage Tuner
Gainward ExperTool
GPU Voltage Tuner
GPU Caps Viewer
GPU PerfStudio
GPU Shark
MSI Afterburner
NVClock (Linux)
NVIDIA Inspector
SLI Profile Tool
The Compressonator
3DCenter Filter Test
3DMark 11
3DMark Vantage
PhysX Applications
Cell Factor Revolution
Cryostatis Tech Demo
Cube Wall Demo
PhysX FluidMark
Fluid Physics
NV PhysX Tweaker
PhysX Downloads
PhysX at YouTube
Add-In Partners
BFG Technologies
Point of View