Originally Posted by Xuvial
I'm not understanding the disappointment with Pascal. What am I missing?
1080 completely destroys
previous generation cards in ways I haven't seen in a long time, and it's not even the Ti version. At 1440p and even 4k it absolutely CHOMPS up those frames.Warning: Spoiler! (Click to show)
If people's disappointment is with the pricing, then I totally understand - but the pricing situation is purely because AMD hasn't stepped up (yet), and for no other reason.
From a pure graphics horsepower perspective this architecture is crazy
powerful, and 1080 Ti is going to be even crazier.
It's only about 25-30% faster. Considering this is 1.5 die shrinks worth of (16/20nm hybrid process), that is actually quite underwhelming performance wise. You always have to consider the context. I was more impressed with Maxwell because we were looking at 35% more performance per mm^2 on the same node. I have been expecting more - perhaps a >35% leap over the prior generation. Yes this is a midranged die, but in terms of performance over the last generation, this does come short.
The other issues such as the dishonest marketing around the "cool as a cucumber" when this GPU throttles on the stock blower also get to me. Finally, the charging of a premium for the Founder's edition alienates my support for Nvidia as well.
As it stands, it cannot run 4k on maximum settings on the most demanding of games at an acceptable frame rate. We may be approaching that with the 1080Ti and the Vega 10, but I suspect that true playability at 4k may have to wait for Navi and Volta.
I'll note as well that these GPUs will not age well. They have no hardware scheduler and therefore no async ability. What that means is that when Volta comes out, they will follow the same trajectory compared to AMD cards that Kepler followed after Maxwell came out, which is a decline in competitiveness, while the AMD GCN GPUs will age better.
Sigh ... it seems like both AMD and Nvidia have disappointments to me. The Fury X was AMD's disappointment - 45% more shader, but only perhaps 25% more performance.
Making things more complex is how DX12 handles multi-GPU setups. I bet many developers may simply not implement support.
Originally Posted by Kinaesthetic
I'd spend the extra $300 for the 1080 over the two 480s in XFire because XFire and SLI have been the most miserable experiences of my PC gaming life. They both are a complete headache to use. I don't have enough time in my life to troubleshoot why dual GPUs aren't working in the game I'm playing. I have more important things to do, and when I want to relax, I just want to be able to fire something up and have some fun.
Dual RX 480s over a single card solution is in my personal opinion just stupid. And I laugh when AMD and Nvidia market their cards in a way that dual GPUs are god's gift to mankind. Because it just sucks to use. Ughh I could rant forever on this topic.
It is better to wait and see what Vega and the large Pascal bring to the table. It's probable that the large Pascal, if the 3840 SP rumors are true, will be perhaps 40% faster. Vega is an unknown as it may be a new architecture (or a major revision on top of Polaris and GCN). Then get the custom PCBs, because it's likely the stock blower will throttle, unless of course they release a CLC.
Agree though that the dual GPU setups are often asking for problems. There are just too many driver issues. You will spend more time as an IT support guy than a guy enjoying games in many titles. No, by the way, that isn't hyperbole.
Originally Posted by sherlock
Should doesn't mean they would, as always Nvidia is financially justified to charge whatever they can get away with because they beat AMD to the punch with their next Gen card AGAIN. Upset about the price? You got nothing but AMD's incompetence to blame. Nvidia is not responsible to you with regard to price, but to its shareholders
If Vega is what AMD fanboys think it is, those 1080 price ( 780 price dropped once 290/290Xs showed up) will come down but for now Nvidia is entitled to profit on their performance lead.
With 80% Nvidia marketshare, it is a monopoly like Intel in x86 CPUs. They can get away with it because they are a monopoly. Same with Intel. Do you really think that the 5930K would be so expensive if AMD had a competitive offering?
At this point, I'd support anti-trust actions against both companies. Consumers and innovation tend to win when there are multiple companies vying for supremacy, not when we have a monopoly or a cartel.
Originally Posted by xlink
Depends on how they define performance...
Maybe they're referring to frame consistency and latency a bit more.
TBH, I probably care more about the sd(frame rendering time) more than I do the mean(frame rendering time.
This is the part that strong affects playability. Standard deviation or frame time variance (which is STD^2). I assume that by mean, you are referring to frame rate.
I am hoping that in the future, a stronger emphasis on split frame rendering can help.Edited by CrazyElf - 6/11/16 at 5:37pm