Originally Posted by Levesque
The more you talk, the more you look ridiculous.
12.2M pixels is not a problem. I'm playing those games right now without any problems at 7680X1600. Do you?
You don't have a Quad GPUs set-up. You don't have 3X30'' LCDs with 12.2M pixels set-up. You don't have a clue what you are talking about. You're just blowing hot air. So why do you even post? At least Vega has the monitors and the 4 580. But you don't.
Synthetic benchmarks are useless to me. I'm a gamer. I play games. And it's really easy to cheat on 3D Mark 11 and, Heaven and Vantage with the tessellation settings... So who cares?
My last post in here.
Vega was wrong. PooperScooper was also wrong. A Quad-Fire 6xxx serie set-up can easily do 7680X1600, and there is no Crossfire bridge bandwidth limitation at all.
Instead of taking everything they were saying for granted, I got out, bought the monitors, and challenged the claims.
I made my point, and won't derail the thread. So everyone can do what they want with Vega's ''results'' and Popper ''claims'' about a Quad-Fire set-up not able to work at 7680X1600.
Mime is working fine.
There is more than one way to sync frames in a multiple monitor, high resolution environment. Eric Demers went so far to say that Sideport was scrapped because their gpu to gpu communication, efficiency, bus-based communication, and whatever else had improved so dramatically.
I can confirm that fluid, exceptional quadfire scaling is possible at 7680x1600 & higher. I've seen it in effect. Similarly, 11.x megapixel performance is stutter-free with consistent, damn-near linear scaling. Pretty often I'm enabling 2 & 4xAA, though obviously not always.
What's interesting is how often nf200 gpu-gpu communication via that pci-e bridge has been shown to favor AMD performance as well as nVidia. I just swapped out my 6990+6970+6970 for my current tandem of cards, and after selling the first 6970 tested triple crossfire at the same configuration as [H]. Without exception, in 5 minute "in-game" benchmarks, at higher settings (4xMSAA instead of AAA, 16EQAA in BFBC2) I was seeing double digit percentile performance beyond [H] results. Not to call into question their testing protocol or fairness; I feel they are second to none, in that regard. The process of re-testing further solidified their commitment to data. It's just been a very strange range of results, indeed. There's so many potential variables to consider, who knows what's causing the gap separation between one chipset, one CPU family, or even similar rigs running similar clocks.