Inno3D Home Page Inno3D Home Page

FAQ News Search Archive Forum Articles Tweaks Technology Files Prices SETI
Visit NVIDIA's home page.
Favorite Pics
Click to Enlarge
OCZ Tech Titan 3
1.0GHz Pentium III
eVGA MX Shootout
nForce Preview
VisionTek GeForce3
2001 Spring Lineup
GeForce3 Preview
eVGA TwinView Plus
VisionTek GF2 Ultra
OCZ GeForce2 Pro
Winfast GeForce2 MX
GeForce2 vs Quake3
Linksys Cable Router
GF2 FSAA Shootout
GeForce2 MX Preview
Benchmarking Guide
Quake 3 Tune-Up
Power DVD Review
Live! Experiences
Memory from

FastCounter by bCentral

 Visitors Are Online
Powered by
3D Chipset
Gamers Ammo
Reactor Critical
GeForce FAQ
Dayton's Misc.
G-Force X Sweden
Maximum Reboot
Media Xplosion
nV Italia
Riva Station
nV News Home Page

VisionTek GeForce2 Ultra - Benchmarks and Gaming

Skip To:

Test System

Alienware - Ultimate Gaming PCAlienware Area 51:Aurora (Modified)

  • AMD Duron 1000m (OCZ)
  • Abit KT7 KT133 motherboard (OCZ)
  • 256MB PC133 Virtual Channel CAS2 SDRAM
  • 15GB ATA66 7200rpm HD
  • Sound Blaster Live! X-Gamer
  • Linksys 10/100 NIC
  • MidiLand S4 8200 5.1 Sound System
  • Martian Red Case and Keyboard

As usual, I leave all my games in the settings that I like and use when gaming. I do not turn off sound, I do not lower geometry settings. I like my eye candy...

We are going to compare default to overclocked scores in this review.


We'll start the benchmark analysis with MadOnion's 3D Mark 2000.

While 3D Mark continues to be the most widely used Direct3D measuring stick, many still dislike it's T&L "bias". If you have read one of my reviews before, I have to say it is not bias too use features that are available on a video card that may not be available on a competitor's card. I call it smart.

3D Mark 2000 1024x768 16bpp

Default Clock Speeds - 250MHz Core/460MHz Memory

Overclocked Speeds - 285MHz Core/500MHz Memory

I attribute the fillrate increase when overclocking to the increase in memory bandwidth. The polygon increase benefits from the core clock increase as well as the memory clock increse. Despite a memory bandwidth limitation, an increase in core clock will affect performance in situations where simple textures are used or color diversity is not as dominant, easing the memory load and allowing the GPU to get ahead of the memory wait times. This does not happen too often, though...

3D Mark 2000 1024x768 32bpp

Default Clock Speeds - 250MHz Core/460MHz Memory

Overclocked Speeds - 285MHz Core/500MHz Memory

Again, we see around a 10% increase from overclocking in the raw data aspects of 3D Mark. 32-bit color scales just like 16-bit when you have as much horsepower as the Ultra.

Shading the Pixels

One of the more impressive features of the GeForce2 architecture has to be the NVIDIA Shading Rasterizer (NSR). Until the Ultra, the performance penalty of using pixel-shaders was severe enough to force a gamer out of the comfortable resolutions of 1024x768x32 on the GeForce2 GTS.

To show off the NSR, I usually use the Evolva Bumpmapped Benchmark.

Evolva Performance

Take this...


Add this...


You get this...


I will not lie, the hit on framerate is significant. However, the Ultra gives you the headroom to hammer fillrate limits and maintain solid framerates for gaming. Here are results of the Evolva Rolling demo at 1024x768x32bpp.

Default Clock Standard Bump-Mapped
Minimum 32 30
Average 120 66
Maximum 238 138
Minimum 39 33
Average 127 72
Maximum 238 155

72 frames per second is extremely playable in the less dizzying third-person adventure game. 1280x1024x32 is definitely in the Ultra's range of usable resolutions when bump-mapping, especially when overclocking. Visual quality of bumped 10x7x32 far surpasses non-bumped 16x12x32. I look forward with much anticipation to effects using DirectX 8 on the NV20, but by the performance levels demostrated here, the Ultra will not find itself obsolete any time soon.

More Tables and Pictures

You didn't think I was going to skip over my favorite game ever?

I even ran a few MDK2 benchmarks. On to the timedemos...

Next Page: More Benchmarks

Skip To:

Last Updated on January 21, 2001

All trademarks used are properties of their respective owners.