 |
MSI nBox N5900 Ultra-VTD256: We take a look at MSI's flagship FX card. Dubbed the nBox, it packs in the goodies, and is backed up by the killer FX5900 GPU. |
| Date: |
October 8, 2003 |
| Manufacturer: |
|
| Written By: |
|
| Price: |
|
|
|
|
|
OpenGL Games
Currently there is a smaller amount of OpenGL based games, with only two major engines currently available that stress video cards. First we will look at Jedi Knight II which is based of the very common Quake III engine. This game offers very nice graphics, and has some very good gameplay. However with current high end video cards they are CPU limited in this game until we put AA/ansiotropic filtering on. But lets look at the results at 1024*768 to see driver quality with the drivers.

| Video Card |
Minimum FPS |
Average FPS |
Maximum FPS |
| MSI FX5900 Ultra Oc'd: |
126 |
157.91 |
205 |
| MSI FX5900 Ultra: |
122 |
154.33 |
189 |
| Radeon 9000 Pro AiW: |
120 |
149.45 |
186 |
| MSI FX5900: |
N/A |
144.30 |
N/A |
| Radeon 9800 Pro AiW: |
N/A |
142.20 |
N/A |
| Matrox Parhelia: |
114 |
136.77 |
164 |
We can see from these results that the newer drivers for all three companies have improved performance by quite a fair amount. ATi has gained 5% from the Cataylst 3.6 to 3.7, nVidia has gained probably near that, as the clock speed difference isn't that much of a determining factor. Lastly we have the Matrox Parhelia which has definitely improved in score since we originally tested this card. Overall all three cards are very playable with the minimum frame rate never going below 110fps. How about when we turn on the additional quality features of AA and ansiotropic filtering?

| Video Card |
Minimum FPS |
Average FPS |
Maximum FPS |
| MSI FX5900 Ultra Oc'd: |
124 |
152.58 |
187 |
| MSI FX5900 Ultra: |
106 |
149.36 |
192 |
| Radeon 9000 Pro AiW: |
25 |
33.2 |
42 |
| MSI FX5900: |
N/A |
141.20 |
N/A |
| Radeon 9800 Pro AiW: |
N/A |
140.10 |
N/A |
| Matrox Parhelia: |
77 |
92.10 |
111 |
What can we see here, performance wise? The 5900 Ultra really hasn't lost any performance going to 4X AA and 8X ansiotropic filtering, losing about 3-4% of its frame rate. However we can see that there is a larger fluctuation in min/max frame rate, as the non-overclocked Ultra almost reaches a low of 100fps, about 16fps lower than in the non AA test. The Radeon 9800 and FX5900 are similar in that they don't lose much if any performance in this test. The Radeon 9000 shows us what the term 'budget video card' means, as it drops over 100fps from its previous test. The Parhelia loses about 33% of its performance in this test, which isn't that much when this card came out but seems high with the current massive bandwidth that is on the newer cards. Does moving to 1600*1200 slow down the CPU limited FX5900 Ultra?

| Video Card |
Minimum FPS |
Average FPS |
Maximum FPS |
| MSI FX5900 Ultra Oc'd: |
123 |
153.33 |
178 |
| MSI FX5900 Ultra: |
127 |
154.08 |
198 |
| Radeon 9000 Pro AiW: |
62 |
80.42 |
94 |
| MSI FX5900: |
N/A |
136.5 |
N/A |
| Radeon 9800 Pro AiW: |
N/A |
138.7 |
N/A |
| Matrox Parhelia: |
64 |
75.96 |
89 |
In a short answer, no, raising the resolution does not stop the FX5900 from being CPU limited. The regular FX5900 starts to show its becoming more fill rate limited in this test. The Radeon 9800 Pro is still not really fill rate limited as it only a couple of fps lower than the results from 1024*768. The Radeon 9000 Pro and the Parhelia are the weakest video cards here, but will still provide over 60fps in this game, with the Radeon 9000 performing slightly better. What happens when we enable AA and ansio at this resolution?

| Video Card |
Minimum FPS |
Average FPS |
Maximum FPS |
| MSI FX5900 Ultra Oc'd: |
126 |
142.33 |
156 |
| MSI FX5900 Ultra: |
118 |
132.08 |
151 |
| MSI FX5900: |
N/A |
103.90 |
N/A |
| Radeon 9800 Pro AiW: |
N/A |
130.50 |
N/A |
| Matrox Parhelia: |
36 |
43.96 |
59 |
Here is the first time that the FX5900 have a real difference between them. The regular FX5900 simply drops, by about 30fps, compared to previous results and even compared to the Ultra card. The Ultra ties the Radeon 9800 in this test, but bests it when the 5900 Ultra is overclocked. Of the cards you see here, the only card that is not playable is the Matrox Parhelia, which has a maximum frame rate of under 60fps. With Jedi Knight on a whole these high end cards have absolutely no problem providing playable frame rates even at 1600*1200 with 4X AA and 8X ansiotropic filtering. How about Serious Sam SE, does it slow down the cards?
|