AA Rendering and other things

[Lux]

Not necessarily. If the rendering pipeline isn't limited by fillrate then adding/removing extra pixels to be rendered won't make much difference to performance. This is also why lowering your screen resolution won't necessarily speed up the rendering of a badly-performing game.

 

Did some testing and I stand corrected. I took some screenshots clearly showing the AA effect and what I found is, you can not change a video setting and have it applied without completely restarting the game. Restarting the level will not apply the new settings. The game must be completely reloaded for the new setting to take effect.

 

I tried turning off AA>0 in game and forcing it in the driver and there is a difference although, at least on a 780gtx, its not discernible and the fact that no matter what the game caps your FPS @60 its hard to tell what performance impact is occuring.

 

I did notice though with 0xAA selected in the driver and Vertical sync OFF in game along with forced off in the driver, on the loading screen FPS went up to the 1000FPS cap. With 32xCSAA +8x Supersample, the FPS on the loading screen only went up to 255. So the difference is there and it is substantial but with such an old title if its being rendered >80fps no matter what settings are applied (screenshot FPS fluctuates between 78-80) then its hard to actually see the performance difference.

 

Nvidia has always had superior OpenGL performance but for newish ATI cards to be getting 20FPS in this game sure does seem like something is broken considering the load on the video card is quite small. I hope they figure out what it is.

 

I recall running Crysis at 2xAA and FPS being around 50ish FPS at the time and trying to run 8x brought the FPS down in to the 8-15FPS range. Every game I've ever played always exhibited this behavior so I suppose when I first played through Doom 3, I never even tried higher levels because of what I knew about AA in every other game. The old "elephant and the string" adage. Thanks for the lesson!

 

It would be interesting as per motosep's comment, to turn off hardware rendering and run a test only using software rendering on each of the affected systems just to get an idea of what each of the systems CPU's is contributing to the mix. See if there is any correlation. It definitely seems like the video card is not correctly handling what the CPU is handing off to it.

 

Sorry to be off topic, Biker, we can't delete posts that I'm aware of but feel free.

[stumpy]

doesn't 'virtual sync' lock the frame rate to 30fps or 60fps if that is on or was that removed from the source, are the tests being run with virtual sync off.

[lowenz]

It's *Vertical* sync

 

IDTech4 frame cap @60 FPS isn't the same thing.

Edited October 25, 2014 by lowenz