Originally Posted by guitarmageddon88
1. So you say that a 2gb 6950 xfire setup that does a consistent, lets say, 100fps, that drops to say, 50 fps in the hard parts with lots of processing, is the same as a 1gb 6950 setup that suffers from a lack of vram and actually jolts down to 50fps due to the lack of vram which affects more than just raw fps? Both readings will show the same min/max/avg fps, but the reasons one is dropping are different from the other. Thats why I say, show me the readouts, either via gpu-z, or via afterburner, charted over time in relation to GPU usage, then we can see what its really about.
Also, if you use the command "render.perfoverlayvisible 1" in game, while I was on the 570's the gpu line would spike all over gods creation when settings were maxed. of course, I could turn it to medium settings, ssao on, and no msaa, and it wouldn't spike, but that's not my goal. Now I get a nice consistent flat line with everything maxed to the utmost.
2. I would prefer to run msaa over fxaa any day of the week in its current form because either via NVCP or via injectors, it looks just plain bad compared to msaa. And if Im going to spend a decent chunk of change on videocards, I may as well go all the way, spend the extra 100 bucks as I did over the 570's I did have, and get the card that lets me max every single setting there is. Otherwise, I wouldve kept my 5870 which stuttered just as bad as SLI 570's under maxed bf3
Well, shoot for $100 more a card vs. 570's, that's one thing. That's a different scenario that what we're talking about here though
And I'm not being critical of your choice there, nor anyone's choices for that matter. Hell, if you got bucks to throw around and thus have the luxury of insisting on the highest possible settings at all times no matter what ... go for it, that's awesome! And I'm jealous.
But what I'm debating is the idea that 1.5GB 580's are inadequate for 1080p, since you did not originally specify 570's or 580's in the post I replied to.
In 'backup' for what I've said, it appears that the only difference that having extra vram MAYBE makes on BF3 MP is the level of AA and AO you can run.
Have a good look over the review here
If you study the whole page and MP section carefully, you'll discover that 1.5gb SLI 580's actually hold their own VERY well even against 7970's in xfire and 680's in SLI in terms of max playable settings.
In single player at 5760x1200, you have to lower MSAA from 2x to 0x, and lower AO by one notch, with 580SLI vs 680SLI or 7970XF.
And here's what you're looking at in terms of max playable settings in MP:
680 SLI (2 gb) - 5760x1200, 0xAA, FXAA, Motion Blur On, HBAO, otherwise maxed
7970 XF (3gb) - 5760x1200, 0xAA, FXAA, Motion Blur Off, SSAO (lowered from HBAO), otherwise maxed
580 SLI (1.5gb) - 5760x1200, 0xAA, FXAA, Motion Blur Off, AO Off, otherwise maxed
So as you can see, even at 5760x1200, the 2gb 680's clearly outperform the 3gb 7970's in terms of max playables, and the 1.5gb 580's aren't far behind the 3gb 7970's at all. I'll leave it to others to extrapolate on what these result implies about the supposed inadequacy of 1.5gb of vram in BF3 MP ... given that these tests are done at +333% more pixels than 1080p.
AFA you analysis of the 1gb vs 2gb card goes ... I'd bet money that when you see a chart that says this
It's pretty much guaranteed that the FPS over time charts (like on [H]) will look identical for both cards.
What you're speculating, that there's different 'causes' of the min fps level, but yet this results in the exact SAME minimum fps reading AND average FPS reading for the two cards ... well, lets just say ... Occam's Razor has a bone to pick with your logic Edited by brettjv - 4/16/12 at 11:06am