Final Words
Both of our custom benchmarks show ATI cards leading without anisotropic filtering and antialiasing enabled, with NVIDIA taking over when the options are enabled. We didn't see much improvement from the new SM3.0 path in our benchmarks either. Of course, it just so happened that we chose a level that didn't really benefit from the new features the first time we recorded a demo. And, with the mangoriver benchmark, we were looking for a level to benchmark that didn't follow the style of benchmarks that NVIDIA provided us with in order to add perspective.Even some of the benchmarks with which NVIDIA supplied us showed that the new rendering path in FarCry isn't a magic bullet that increases performance across the board through the entire game.
Image quality of both SM2.0 paths are on par with eachother, and the SM3.0 path on NVIDIA hardware shows negligable differences. The very slight variations are most likely just small fluctuations between the mathematical output of a single pass and a multipass lighting shader. The difference is honestly so tiny that you can't call either rendering lower quality from a visual standpoint. We will still try to learn what exactly causes the differences we noticed from CryTek.
The main point that the performance numbers make is not that SM3.0 has a speed advantage over SM2.0 (as even the opposite may be true), but that single pass per-pixel lighting models can significantly reduce the impact of adding an ever increasing number of lights to a scene.
It remains to be seen whether or not SM3.0 offer a significant reduction in complexity for developers attempting to implement this advanced functionality in their engines, as that will be where the battle surrounding SM3.0 will be won or lost.
UPDATE: CryTek has pointed out that the new lighting implimentation is essentially the same but uses branching in the pixel shader to accomplish what needed to be done in multiple shaders under the PS2.0 path. This indicates that the conditional rendering feature of SM3.0 is actually faster than using multiple shaders (which gives NVIDIA 6 series cards a performance advantage when multiple shaders would have been required).
Facebook
Twitter
RSS
36 Comments
View All Comments
Illissius - Friday, July 2, 2004 - link
For these benches, were nVidia's trilinear and/or anisotropic optimizations on or off? (This would help in comparing results with other sites, for example.) I don't recall seeing them mentioned, but they're getting to be as important as the driver revision these days.DerekWilson - Friday, July 2, 2004 - link
DAPUNISHER:I'm not sure about the 64bit version of any game, as game developers are much more likely to hold everything until MS releases WinXP64 than hardware vendors. My guess is that we can expect not to see any visual improvements or differences with the 64bit move. There's much less reason to alter the graphics of the game when gaining more registers and memory address space than when you add the ability to do conditional rendering, floating point frame buffers, instancing, and all that...
Zak,
We could try to guess performance based on these numbers:
http://anandtech.com/video/showdoc.html?i=2044&...
But we didn't use those because they're based on the 1.1 version of farcry under dx9b and Catalyst 4.4 ...
Our focus was the impact of SM3.0, not on overall relative performance, but in the future we will include older generation cards even when looking at next gen features. You are right, it does provide a way to relate to the numbers, and those cards should be in there for completeness' sake as well. Thanks for the suggestion.
Zak - Friday, July 2, 2004 - link
I wish you guys would include one or two benchmarks on some older video cards to give a point of reference for those, such as myself, who still run R9800 and older generation cards. Without seeing how the game performs on R9800 or eqivalent card it's hard to relate to these benchmarks.Zak
DAPUNISHER - Friday, July 2, 2004 - link
Very impressed with the GT's performance in this version. When can we expect your preview of FarCry 64bit version with the SM3 path Derek? and will 64bit bring some new eye candy or more performance? Inquiring minds want to know :-)Warder45 - Friday, July 2, 2004 - link
Interesting. Some odd stuff like ATI's X800 line actually decreasing in performace with the 1.2 patch. I wonder if thats a driver issue that now needs to be fixed, but if I was an ATI owner I'd stick with the 1.1 version of the game. I'd really like to see someone benchmark with omega's drivers for ATI, and see if there's any difference in performace there.#9, That article at tom's is from the NV40 review months ago. This new verison, 1.2 fixes most of the IQ problems nvidia was having.
http://www.xbitlabs.com/articles/video/display/far...
So far the only IQ problem I've seen mentioned with the new version.
araczynski - Friday, July 2, 2004 - link
very nice, that 68UE sure is kicking some tail, before AND after the SM3.perhaps this will lead to developers optomizing (to some small degree at elast) their code for the 2 camps? (or at least for the camp that pays them the most...)
in any case, here's to hoping the 68U/UE are priced acceptably by xmas, or at least next tax time :)
Shad0hawK - Friday, July 2, 2004 - link
"Then, during the ATi refresh,we will all be greeted the the Geforce 6900, 6900 Ultra, 6900 Turbo and 6900 Ultra Hyper Fighting Edition."actually that will probobly be after ATI anounces the super golden/silver platinum extra extra XT edition with not only one but TWO "free" certificates for games not out yet
nserra - Friday, July 2, 2004 - link
#1, #2, #3, #4:I think the huge hit is because nvidia is not doing AA to the all scene as ati does.
The new drivers from nvidia have this ability. How do you think nvidia have come to top so soon, after some driver release ... Trilinear optimizations, Shader optimizations and now AA optimizations...
I also don't understand why only toms site notes differences between ati and nvidia image quality...
http://graphics.tomshardware.com/graphic/20040414/...
ZobarStyl - Friday, July 2, 2004 - link
Bearxor, though I agree the overclocked editions are silly, don't act like ATi doesn't do the exact same thing...Ultra Extreme = XT Platinum Edition
Ultra = XT
GT = Pro
vanilla 6800 has no direct competitor, but it held it's own occasionally against the Pro in the review.
Both of the double-named cards are just the top end overclocked, so I tend to ignore them in the reviews, but then the GT was beating all of the ATi cards in some of those demos too...
RyanVM - Friday, July 2, 2004 - link
#6, Ditto :p. Dell 2001FP for life :D