Viper Lair
Latest Stuff

Enermax CS-656TA
Ultra X-Connect 500W PSU
AMD Athlon 64 3000+ CPU
AOpen CRW5232 CDRW Burner
Cooler Master Centurion 5 Case
Kingston 1GB USB 2.0 DataTraveller
Cooler Master Stacker Case
WACC Dual Radiator and P/A Reservoir
Kingston HyperX PC4300
Latest Stuff
Search for lowest prices:


Price Search:    for    

MSI FX5950 Ultra-VTD256 MSI FX5950 Ultra-VTD256: Looking for the baddest nVidia GPU on the block? We take a look at MSI's monster, packed with 256MB of ram.
Date: February 20, 2004
Written By:

Test Setup

Abit AN7, Barton 2500+ @ 3200+, 512MB of Corsair TWINX PC3200 Ram, 80GB Western Digital, WindowsXP SP1, ATi Radeon 9700 AiW Pro with Cat 3.10's

Abit AN7, Barton 2500+ @ 3200+, 512MB of Corsair TWINX PC3200 Ram, 80GB Western Digital, WindowsXP SP1, MSI FX5950 with Detonator 53.03

Test Software will be:
Unreal Tournament 2003
Max Payne2
Splinter Cell
Return to Castle Wolfenstein
Call Of Duty

The comparison video card will be the ATI Radeon 9700 Pro AiW. Benchmarks will be shown with and without Anti-Aliasing and Anisotropic Filtering. Using an older card like the 9700 Pro should help to show the difference in these two generations, which is where the most likelihood of upgrade comes from.

Due to the fact that the ATI 9700 Pro is not designed to do some AntiAliasing testing that the FX5900 can, we'll be presenting those benchmarks at the end of each game test. Of course, same can be said, cards reversed (no 16x Anisotropic Filtering for example), and we will not be presenting those ATI numbers since you can read about that in our All-In-Wonder review.

I did think about using Halo as an additional test but to be honest the included Timedemo is of no use to us as it features no actual gameplay but rather engine generated flyby cut scenes. Also Halo has issues with Anti Aliasing the same as Splinter Cell, so one of these two is enough.

Unreal Tournament 2003

We used the and selected the high quality batch run at all resolutions. We'll be presenting Antalus, and Citadel benchmarks, with AA/AF on and off.





Considering that the 9700 Pro is over a year older than the FX5950 Ultra it doesn't perform that badly. Without the likes of FSAA and AF the FX5950 Ultra holds the edge which is as it should be but when you enhance the graphics to just 2xFSAA and 8xAF the cards become a lot closer in performance. After 4xAA is where things become confusing as the Radeon 9700 Pro fills it's memory buffer as can be seen by the same FPS across the 2xAA to 6xAA range. The FX5950 Ultra however has no problems here and puts out playable average numbers even at 4xAA/8xAF. 6xAA is a struggle and 8x as we have seen before from the FX Series cards the memory buffer appears to reaches its limit.





The Citadel tests show the same trends we saw with the Antalus runs, that without any graphics enhancments the FX5950 Ultra is happy to kick out some BIG numbers, but once you apply things like 4xAA/8xAF the performance gap between the 9700 Pro and FX5950 Ultra closes.

Max Payne 2

Max Payne 2 is a nice test of DirectX 9. Using FRAPs we were able to get some frame rate samples. The results for Max Payne 2 were derived from 3 runs (averaged) at each setting, in the first Hospital Scene.



Max Payne 2 with Fraps shows us much the same thing as UT2K3, that the FX5950 Ultra can only hold a real edge when you don't enhance the graphics, especially at the higher resolution of 1600x1200. Once again the 9700 Pro apparently fills it's memory buffer after 4xAA 1600x1200 whilst the FX5950 Ultra can carry on, albiet with low numbers.


Shop for MSI Video Cards.
Copyright 2001-2004 Viper Lair. All Rights Reserved.