LePaul, I get anywhere from 90-100 fps with my setup:
1.8GHz P4 Northwood
512MB DDR Ram
DX8.1
NOTE: I run with V-Sync enabled for everthing. My monitor will do 120Hz rates at 1024x768, which is where I usually set my resolution at, depending on what I am doing.
Now, about drivers. In the days before the original Radeon came out ATI got a deserved bad reputation for having lousy drivers. With the original Radeon, they did a damn fine job, however, they released the 8500 line too early and got lambasted again, justifiably so.
NVidia's strenth is also thier weakness, in my opinion. How many threads, just on the BBS, start with, "Which NVdia drivers are best?". The following tirad of different opinions on which are best is mind boggling.
All video drivers have problems, regardless of who makes them. Accept that and it makes the decision a bit easier.
I think one of the really annoying things about NVidia comes to changing a video driver. They do not clean up after themselves very well and usually require some type of third party solution or manually purging the drivers from the system.
ATI, whether by accident or design, does a pretty nice job of cleaning up. Actually, it appears they use the same naming convention for any given product so upgrading/downgrading on the same video card just stomps the old driver.
Between the systems I have around here, I do not think twice about changing an ATI video driver, but for the NVidia based systems, I have to have a really good reason to change them due to the amount of time/effort required.
ATI does have one technical edge over NVidia, but it requires software to support it. ATI's TRUFORM technology is impressive. I wrote some simple tests to see what it would do (if you can call DX programming simple
). Just a simple sphere and each pass I upped the TRUFORM level. With each step, the number of polys doubled. What I was looking for was two thing 1) frame rate differences and 2) fill rate versus drawing power.
In frame rate it works pretty much like ATI says. The frame rate drops was too negligible to measure.
In the second part, I was curious how much processing power was needed to do the work and how it would affect fill rates. I really thought fill rates would suffer given the procssing power need to effectivel tessalate the polys. But I was very surprised to see the cards fill rate stay constant.
NOTE: The sphere was textured using a WRAP. So the card had to calculate the UV coordinates as it tessalated the polys. I noted no artifacts from this.
If software gets behind this technology, FSAA will become a moot point, as this does what FSAA does, but it does it at the correct level of the draw with significantly better results and virtually no degradation in performance.
Sorry, I kind of deviated from the intent here. I think the real questions you want to ask yourself are pretty simple.
1) Do I want the best frame rates I can get regardless of anything else? (this is not a bad thing, by the way)
2) Do I place more importance on video quality and am willing to give up some frame rates for it?
If you like (1), then NVdia is the way to go, otherwise ATI is the way to go. Simple.
