UT2003 GPU Shootout

Yes, we do adopt to what we are used to, I think what Reverend is getting at though is that for 3d realtime graphics, just like for that car analogy, there can be a level where enough is enough for the present and foreseeable future. He doesn't want 30 fps back, but instead doesn't see a neccessity for going towards 1000fps, which is where I agree!

In the early days of automobiles, 10 to 20 mp/h were top speeds and while they were feeling comfortable at the time, there was still an obvious need for an increase. The sensible limit of that however was reached soemtime around the area where most purchasable cars had a top speed of 60-100 mp/h. Of course there are cars that can go faster than 100-150 mp/h, but by the limits of todays streets, laws, material science and physics more doesn't make much sense unless you talk about professional racing, which leaves about 99% of the population happy with ~100-150 mp/h. The same can be said for framerates, while 30 was good at the time it needed improvement, and the sensible limit of that was reached say at 60-120fps (depending on gametype), which will definitely be enough for 99% of the users (those without inhumanly good eyesight to really percieve more than 120 fps). Everything above that could be considered cool, but kind of a waste of power (power that could otherwise be used to achieve better looking graphics).

So, instead of focusing just on how fast our cars, eh ... graphics card's top speed is, we are now at a level of performance where things like stability, image quality, features and average framerate become just as or even more important than raw top speed. I'd rather have an average of 60fps while never dipping below 45 or above 90, than having an average of 60fps with a max of 190 and a minimum of 20 - obviously my prefered scenario would be a rather bad looking case in many of today's reviews!

People and websites still focusing much more on max top speed instead of the full picture, which today just doesn't cut it anymore IMHO ...

 

For me it's enough ! Perhaps it's not fast enough for multiplayer-duels, but against the computer it should be ok ! And then i can also play at 800x600 or reduce details til i have a new graficscard (hopefully PowerVR-based )!

CU ActionNews

 

I like to play mainly FPS and an almost constant 60 fps is more than enough for me. I agree with Reverend, and we need to focus on quality now.

For current FPS 32bits games a good DDR3Dcard/1GHzCPU will play very well for me, but with UT2003 and U2 it is difficult to predict what will be good enough because we know almost nothing about it. Only Anand has the benchmark.

Unofficial UT2003 FAQ: http://cleaned.beyondunreal.com/UT2FAQ/UT2FAQ.htm

I hope my Gf3Ti200/P3-S will be enough to play smooth at maximum quality

 

I hope your are joking, assuming some credibility to Anand's numbers, on his P42.5 the Gf3Ti200 on High Detail/1024x768 only gets 42.9fps average on Antalus, which is the 'outdoor' stress level. There are more demanding outdoor levels with falling snow etc as well.

 

Aren't weather conditions adjustable?

 

Randell,
No joking, I can play at 800x600x32 high detail (66fps) because I have a small trinitron monitor.
IIRC he used a Athlon. Also we know nothing about the CPU scallability and my little P3-S is really good

 

OK 800x600 it is then

Ail - weather options, I dont know, I havent found them, unless they are inherent in some other option.

 

Ailuros,

most likely no. I believe they are spawned from their new particle system that is placed inside a map. As soon as I get my copy I will check for you.

 

It's not as if we haven't seen UT2K3 pics before. There's already an official website, with pics, fer cryin' out loud.

 

Yeh but an NDA lasts untill it expires.

 

If you consider Anandtech's first use of UT2003 using driver build 848, the Radeon 8500 beat the geforce3 Ti500 in every benchmark except 1600x1280x32. In that test, the nvidia product beat ATI by a monstrous 0.9 FPS. This conclusion would have been contradicted by tests generated by using the driver build 856. In those runs, the geforce3 Ti500 beat the Radeon 8500 in every resolution. Beta software is not something that creates reliable results. This benchmark does not give an individual a good idea of what kind of performance we should anticipate. Anandtech's use of UT2003 beta software tends to emphasize their pro-nvidia bias.

AnandTech questioned the Radeon 8500's success in the driver build 848 supposedly as a result of ATI having not yet to addressed a fog concern. AnandTech hypothesized that when the fix was applied the Radeon 8500's performance would drop. This calculation was exceedingly imprudent because no one really knew why the difficulty existed. His conjecture directed culpability towards ATI. If the UT2003 engine was created on an nvidia platform, Epic would be at fault. Since there was no proof of that, the objective option would have been to point out the fog matter and leave it at that. But the problem is that AnandTech goes even further as to make an inquiry of the integrity of the Radeon 8500 performance numbers. This is pure bias.

I trust that ATI has already corrected the texture matter and UT2003 and has yet to be made public. No one seems to know what changes nvidia did to their drivers in order to optimize for UT2003. I can’t imagine that they didn’t alter anything.

Later on after a driver revision Anandtech did the benchmarks with driver build 856, the Radeon 8500 performance improved 16% at 1024x768x32, 24% at 1280x1024x32, and 14% at 1600x1200x32 in that order. The fog matter had been rectified and there was no slump in performance. We can therefore conclude that AnandTech was wrong with his original “assumptionâ€￾ that ATI was at fault. Further glance at the driver builds used we can see the driver build 848 evaluation was done on Jan 24. The Build 856 review was put up two weeks after on Feb 6. The problem here is that when testing for driver build 848 was being done, the drivers from ATI were unofficial 7.65 which contained version 6015. So while AnandTech got the most recent drivers from nvidia, he did not for ATI. Essentially, he was using the drivers from Oct release Radeon 8500. Even nvidia was correcting troubles the first few months the geforce3 Ti500 was on the market. If AnandTech obtained the latest ATI drivers, the fog issue would never have been an issue at all. The Radeon 8500 would have soundly beaten the geforce3 Ti500 in all tests of driver build 848.