R700 Inter-GPU Connection Discussion

This completely disregards the aspect of visual fidelity (i.e. smoothness in this case).

 

Ahem:

Not to start some perception argument, but I can tell the difference between the two, especially in games where the camera spends any amount of time turning. In RTS's it may not be very noticeable to me, however.

Wasn't it Project Gotham Racing that had the scene render at 60 fps while the reflections only updated at 30fps? That used to annoy me to no end.

 

My comment was specifically WRT your previous comment. If I took that out of context I apologize.

 

I certainly agree but i'm mostly refering to the high end setups which are capable of playing Crysis at very high settings at modest resolutions vs other games at insane resolutions.

Although I think the same argument applies (albeit not as strongly) to playing Crysis at High settings with say a resolution of 1920x1200 vs playing other games at 2560x1600 or 8xAA etc...

 

I agree with Chris. If you're going to quote someone, use the Quote feature. That way it's easier to follow who you're replying to. Use it, please, and keep the thread on topic.

 

This is my target hw set as well. Perhaps others are referring to more mainstream/performance hardware?

Depending upon u-arch you may be right, although I suspect this comparison is rather ATi-friendly and NV-unfriendly. Crysis just seems to run poorly with HQ settings + AA/res no matter what hardware is being utilized.

 

410520, I think you're forgetting something important here .. dx9 vs dx10

Crysis is a DX10 game and most sites bench it using DX10 .. hence +- 30fps where as COD4 is strictly DX9 .. hence getting well over 100fps(over 200fps for R700).

Once Crysis hits DX9 .. you see +70fps <-- Not gonna google Crysis DX9 results, you can do it.

http://www.driverheaven.net/reviews.php?reviewid=588&pageid=3
http://www.driverheaven.net/reviews.php?reviewid=588&pageid=6
http://www.legitreviews.com/article/745/6/

US

 

The means by which the graphical results are achieved is irrelevant from an IQ perspective. If (hypothetical) COD4 runs @ high res + high AA and looks as good or better than Crysis with a much lower resolution and "just" high settings rather than high quality (to achieve the same level of playability in both titles), who's the "winner"?

 

IMO, graphically Crysis looks way better than COD4 irrespective of how fast it runs.

I haven't touched Crysis since it came out and I got and finished it, but I still remember how beautiful it looked. Sure COD4 looks beautiful too, but it's a different beauty. I play Multiplayer COD4 online. That said, I love CSS and play that online too and think it's really beautiful, better than COD4.

US

 

But are you running them with the differential settings as suggested? i.e. COD4 with much higher settings than Crysis.

 

Ok put both games at settings that give 60 fps average and decide which one looks better. This is getting far off topic.

 

But thats not a fair comparison for people who are happy with 30fps in Crysis (which i'll bet in non biased user testing would be an overwhelming majority).

You also have to consider the hardware its running on.

Settings which an R700 can handle both games at 30fps.....

 

That's exactly what's been said. Same level of performance gives different level of quality. To pull off 60 fps in Crysis you need to drastically reduce settings (res and AA) compared to COD4 (or Bioshock, if you prefer).

 

Average of 30 fps means MIN fps in the teens or lower, which is not acceptable for a shooter where slow down occur during intense moments, IMO

 

Nope, I play at 1680x1050 4xaa 16xAF

US

 

What about Crysis?

 

I think I played it at 1280x1024 4xaa,16xaa, been a while since I played it though. and yes, it was slightly laggy but it still looked very pretty.

US

 

Ok. Yeah, not nearly enough of a difference to fit the framework of our discussion. No wonder Crysis looks better to you

And I'm not saying you're wrong either, Crysis is arguably the best-looking game on the market when you crank every setting and run AA + high res. Problem is, there's not a hardware setup in existence capable of pulling this off with playable framerates, and this is the crux of our discussion. IQ is useless if the game is unplayable.

 

Yeah I agreee on that one. But the same comparison could be applied to a "relatively steady 30fps or above". That would allow for the odd dip into the 20's but nothing too significant.

In one of the very high quality settings benchmarks posted earlier, a 9800GTX+ averages 34fps but never drops below 23fps. If that means its at or very close to 30fps most of the time then I would consider that playable.

 

The problem with setting the bar so low is that I'd think that by the time you got COD4 down that low, you've long since passed the point of diminishing returns.

I thought it was that Crysis is inefficient and slower per "IQ metric" than other games, hence why the next Crysis engine is being "fixed" to run better at the same visuals.