A new study has been made public detailing the GPU market share as of this quarter, among other things, and things look really bad for AMD

http://jonpeddie.com/press-releases/details/add-in-board-market-decreased-in-q215

To AMD: That's what happens for leaving the 9x0 series without proper competition for half a year, and then launch mostly rebranded cards.

Damn, S3 still exists?

I made the switch to Nvidia two years ago when I purchased the GTX 770 from upgrading from my AMD  5750s.  Don't think I'll be switching back anytime soon unless AMD releases somethign awesome.  Plus I'm looking forward to share play by Nvidia.

Extra features are always nice to have, but they usually come at a cost.

Personally, I'm down to the basic performance and price metrics and I tend to be objective, althought I must say that Nvidia makes it hard sometimes with things like GameWorks.

I was *this* close to buy a 970, but the low stock stopped me from doing it when I could, and then the 3.5+0.5GB thing came out and I decided against it.

AMD had to concentrate both their development ressources and their money on their new upcoming CPU Design since the last year, which is why GCN only gets incremental upgrades and rebrands right now. With the shrink to 14/16nm and the first tape-outs coming from Zen and Arctic Islands the focus should now gradually shift back to the graphics division.

The R9 Nano and Fury could boost sales up again, but I doubt they will get above 25% again before the release of the Rx 400 Series (Arctic Islands)

I have a hard time believing that neither Nano nor the Fury cards will improve things for AMD.

Leaving Fury X aside (because unless space is a problem the Fury X is a poor choice over the similarly priced but better performing 980Ti), it's true that Fury is a good card as it offers better performance than the Vanilla 980 while costing a bit more and almost the same performance of its Fury X sibbling while costing $100 less, but Nano...

Nano has the same specs (4096 SP, 4GB, etc.) and price of Fury X, but its power consumption is 100 W lower and yet somehow it's supposed to go "up to 1,000 MHz", only 50 MHz slower than Fury X? I really, really doubt it. Even with the best of the best chips, it will have to run a lot slower than that claimed speed, and that will limit its performance a lot, making it pointless.

AMD has already stated that the base clock will be around 850mhz and 1ghz will only be achieved in bursts. Still, even 850mhz is quite an achievement for that TDP. First benches give the card a performance slightly ahead of the Fury (non X) and is apparently very quiet.

Those benches come from AMD, and when they launched Fury X their benches showed that it was on par or even better than Nvidia's 980Ti, and we know how that ended...

And yes, I knew about the speed of the card (it was briefly, but we talked about it on the AMD related thread about Zen and Fury), but even that is weird. With both card sharing the same core specs, does it mean that Fury X needs to use 100W more than Nano to get those extra 150-200 MHz?

Oh, and if Fury showed something is that while on paper it should be between 10 or 15% slower than Fury X due to its lower specs, in reality it is only between 5 and 10% slower, usually closer to the former rather than the later. Why? Because some tech savvy people agree that when AMD scaled up GCN from Hawaii to Fiji, the end result was decompensated and lacks the ability to fully us the extra hardware.

That's why Fury ends up being closer than it should to Fury with both cards running at the same speed, but Nano will be noticeably slower than Fury X and that could hurt it more than we think.

Nvidia seems to have been caught with it's pants down in regard to it's DirectX 12 drivers. Thankfully the situation will improve.

I have a question for our more hardware knowledgeable members, what do you all think of this monster?

I'm far from being a tech savvy, but that thing is, indeed, a monster.

4x8-pin connectors! And with a cooler so big that makes the card a triple slot one, that thing must be the heaviest card ever made. And if they are giving a Razer Ouroboros with it, also one of the more expensives.

Also, apparently DX 12 isn't the only thing where Nvidia is lagging. In the latest podcast from TechReport, it is mentioned that Nvidia has also problems with "preemption context switching", something that apparently is important in VR. They mention that, in the words of some people at Occulus, it is best on AMD, Pretty Good on Intel and Potentially Catastrophic on NVIDIA.

But all of this is pointless because Nvidia will likely fix all those problems with Pascal, and with DX 12 games still away, it won't be much of a problem for current Maxwell users.