 |
|
Review: page 3 of 3 |
| Nvidia XGL V. 4 Prev 1 2 3 |
||
 |
||
|
|
The cards have shown their own strengths and weaknesses fairly easily, and should distinguish themselves according to your particular taste in DCC. If your interest is in general 2D graphics computing with only a flavor of 3D, then the benchmarks definitely shouldn’t be taken for gospel; the end importance lies much more with the user experience, pricing and compatibility at that point. If you’re working largely on a single monitor with high tessellation models without the need for interactive lighting and texturing, the Nvidias would be more your flavor, while effects animators and compositors may opt for the ATI card for its strength in particles, textures and lights, and its stronger performance with dual displays.
It’s very apparent that the largest difference between the FireGL 8800 and the Nvidia XGL cards is their handling of dual displays. The FireGL essentially keeps its performance intact while both the Nvidias suffer a sometimes deep drop in performance—specifically a 43-45% drop in wireframe performance, 15-17% in shaded, 5-8% lighted, and 21-13% in hardware particle performance. So, while the Nvidias best the FireGL in most of the Maya tests in single monitor mode, they drop to a lower performance in dual monitor mode in real-life performance. [an error occurred while processing this directive] In running straight benchmarks (as opposed to gay benchmarks, not that there’s anything wrong with that), the Nvidias beat out the FireGL in all three tests, but it’s important to keep in mind that these tests only run on one monitor, and do not use a second display. I wonder how much things would change otherwise. The first test used, from SPEC, is the 3ds max viewset of their Viewperf 7.0 software (http://www.specbench.org/gpc/opc.static/3dsm01.html). Then I ran a gamer’s benchmark from MadOnion.com called 3DMark 2001, which tests the ability of the card’s gaming ability (http://gamershq.madonion.com/products/3dmark2001). Lastly was eTesting Labs Inc.'s 3D WinBench™ 2000 Version 1.1, which tests the whole 3D subsystem of a machine from DirectDraw to hardware drivers and bus width (http://www.etestinglabs.com/benchmarks/3dwinbench/3dwinbench.asp). The higher the numbers, the better the performance.
The Nvidia cards both did measurably better in the straight benchmarks pretty much across the board. I have to say, though, that I prefer to use my own experiences to gauge whether a card is good or not. And in using these cards, my assessment pretty much follows along with the Maya benchmarks. The Nvidias are quite strong in pushing models around, but since my own animation focus lies in effects, dynamics and particles as opposed to high poly count models and such, I decided the FireGL was a better choice for me, due to its far superior particles performance. Well, that’s what I originally thought. To further complicate the issue, one evening I got the call to go to kill some Zerg online with my friends. I dusted off and reinstalled "Starcraft"—the best game, ever—to play it on Battle.net. I was shocked to find it would not work with my ATI card. I was a crestfallen polar bear at the locked door of a closed ice cream shop in the middle of Nevada in July. On a hunch, and in a desperate bid to kill something online, I installed the 900XGL back in my system to try "Starcraft" out again and wouldn’t you know it, it works. The issue lies with the fact that the game needs 256-color mode to operate and the ATI card would not support that, and understandably so, since it’s strictly a workstation card and not a gaming card. The Nvidias are also workstation cards, but since they come from a different background entirely, it seems their lower end support didn’t get left out. If ATI had drivers that would support 256, then perhaps things would be different. Just the same, though, upon deciding that the Nvidia’s poor particle performance and overall performance hit in dual display mode in comparison with the ATI would not hamper me as much as would my lack of killing Zerg, I decided to keep the 900XGL in my workstation system instead. In my case, the slightly inferior performing-for-what-I-do-in-Maya graphics board won out due to its versatility (and the fact it’s a good card). And I finally got to kill some Zerg (before being disconnected from Battle.net for no reason for the umpteenth time). Dariush Derakhshani describes himself as follows: "31. Nicely bald. Fear of commitment. Mostly malaisey. Hearty appetite for irony and digestive biscuits with chocolate. Contributing author for Maya: Secrets of the Pros and Maya Savvy. Animator. Teacher. A few awards and a couple degrees. Stout little man making reality at Sight Effects in Venice, CA. Has flat feet and needs a better bio. Can be found prancing about at www.painfulurination.com. You can reach him at koosh3d@earthlink.net." Prev 1 2 3  Related Forums: Related Sites: • Animation Artist • Digital Animators • Digital Game Developer • Digital Post Production • Digital Producer • Hollywood Industry [an error occurred while processing this directive] |
|
||||||||||||||||||||||||||||||||||||