by Q2 it just wont matter

That's just what I was thinking!

WAY too late AMD.No one is gonna bother buying 2 really hot chips @28nm when there will be 16nm and 14nm chips kicking ass

by Q2 it just wont matter

Pretty much depends on price as 14/16nm pricing is still very ignored.

WAY too late AMD.No one is gonna bother buying 2 really hot chips @28nm when there will be 16nm and 14nm chips kicking ass

If you had temperature my Fury X reaches, you would still be alive, just with high fever. If you had temperature of your card, You would be in late phase of decomposition. 2x Fiji with same thermal settings as in Nano are easily cooled. But truth remains that AMD had dual Fiji card ready and should have sell it at same time as Fury X. Because back then they could have ask higher price as Fiji proven to be best CF scaling (modern) chip. Maybe best ever. At least in games where CF works. I presume we will see single chip cards performing 1.5 times faster than GTX980Ti/Fury X next year, but they will come at 2 ~ 2.5 times higher price than GTX980Ti/Fury X did launch with. Touch of reality will be pretty hard for those people willing to spent only around $250 for GPU and hoping in 14/16nm salvation.

Q. On the E3 Livecast, Lisa committed to shipping Fiji Gemini by Xmas. What happened? Is Fiji Gemini delayed? A. The product schedule for Fiji Gemini had initially been aligned with consumer HMD availability, which had been scheduled for Q415 back in June. Due to some delays in overall VR ecosystem readiness, HMDs are now expected to be available to consumers by early Q216. To ensure the optimal VR experience, we're adjusting the Fiji Gemini launch schedule to better align with the market.

What?? Are they doing this on purpose or are they lying their asses off? This piece of information is retarded on so many levels I can't even begin to describe how flawed the approach is. Delaying the Gemini so much makes absolutely no sense. Q3 2016 is the time for Arctic Islands. And they want to release a dual-GPU last-gen card a few months before that? That makes no sense. It's like this: MSRP will probably be around 1200$ in Q2 at launch. A few months later in Q3 it will drop massively to make room for Arctic Islands. I'm reckoning 700-800$. OR they could release it in January and get the whole Q1 on top of Q2 of 1200$ sales and then drop the price in Q3 like in the other scenario I presented. The first scenario makes absolutely no sense and leads me to believe that a. they are sabotaging themselves (highly doubtful). b. the card is actually not ready and the VR statement is just a smokescreen. I'm starting to question AMD's actions. This reminds me of Maxwell2 launch when the 970 was selling like hotcakes but they delayed the 290x price drop for more than a month. The 290x should have had its price dropped the moment 970 got out of the factory. I've said this before.

I presume we will see single chip cards performing 1.5 times faster than GTX980Ti/Fury X next year, but they will come at 2 ~ 2.5 times higher price than GTX980Ti/Fury X did launch with. Touch of reality will be pretty hard for those people willing to spent only around $250 for GPU and hoping in 14/16nm salvation.

I don't think the prices will be much higher. 200-300$ is the sweet spot where most of the profit comes from, I highly doubt they'd risk losing market by increasing the price. I mean look at Intel, they've already been sitting on 14nm FinFET for quite a while now. Their prices have increased, yes, but if anything that's due to lack of competition. On the other hand the GPU market is still volatile, Nvidia and AMD are trading blows. Nvidia won this round considering its sheer amount of sales, but you don't know what's around the corner and AMD has been known to fight back with awesome prices. Nvidia won't be able to increase prices too much as long as AMD is in the game. And I bet the same goes for AMD increasing prices. Nvidia may be greedy faks, but they did charge respectable money for the 970. The 290x was over 500$ at the time of 970 release. They could have simply placed the 970 at 290x price but they didn't, they knew profit comes from the 200-300$ segment. And they even overpriced the 980 to make the 970 more appealing. Quite brilliant actually, same kind of move as with Titan X and 980Ti. Anyway, I'm getting sidetracked. I strongly believe the high-end prices won't move that much. GPUs are crazy cheap to manufacture in large volumes, they can drop prices significantly before they would hurt their overall profits, since lower price = more customers buying. At least theoretically.

I highly dought they need this card to make money I think the delay is till nv gets out whatever top end they have coming. that said good luck with the "new" games that don't scale well in cf/sli that wont look good on paper at all.

Actually makes sense to me since AMD will need to ensure they have enough GPUs to support the launch of HMD / VR gear at that time. They don't want to hand this over to the competitor and then have vendor specific optimizations due to lack of marketshare. Many people will be upgrading post Christmas anyways (post holiday sales, after tax rebates, etc...). When the whole ecosystem is ready, it is better for consumers 4K TV, VR and GPUs to run it. Getting new GPU on an old rig sometimes is not enough bang for the buck and does not invite game or movie/tv producers to support it. I'm one of those waiting for the GPU + 4K to be really ready to really enjoy the overall experience.

It will still outpace anything 16nm and 14nm coming around the same time.

AMD have dropped the ball on this one this Dual GPU needs to be out now who in there right mind would purchase 1st gen HBM when newer GPU's will be out. It may be faster than single GPU but it's competition will be newer crossfire/Sli GPU's and they will be faster and it will be slower and may consume more power and be hotter.

I think the Anandtech article covered the reasons behind the delay pretty well. It's something I've been saying here recently - SLI/Crossfire are basically dead until DX12 gets rolling. Most modern engines are using techniques that just aren't compatible with AFR. The only non-DX12 systems that are going to benefit from multiple GPU's is VR, which is why the X2 is being delayed.

If you had temperature my Fury X reaches, you would still be alive, just with high fever. If you had temperature of your card, You would be in late phase of decomposition.

hahaha 🙂)) nailed it.

Dual card is so pointless, waste of time...

Wooooo! AMD continues to "get it done!"

It's something I've been saying here recently - SLI/Crossfire are basically dead until DX12 gets rolling.

depend your use... you are 50% right lol 🙂

Dual card is so pointless, waste of time...

depend your empty space in your box about any dual GPU it is only for the show, manufacturer earn money with mainstream (right now it is the gtx950 that is in nearly ever PC for nvidia and the 370 380 for AMD)... so the target is how people can buzz about it, put the brand in the light and not to sell plenty of them (despite it can be cool, but sadly there isn't too much people that can affort them :banana:).

Not a clue about this so have to ask.. Do these dual GPU cards (AMD or Nvidia) act as one single card or are they basically crossfire/SLI on one board and will show up on a system as such?

VR need two separate data streams, because each eye has its own display. In the best case scenario, each of the two cards, creates a separate image for one eye. Placing two chips on one card allows a better pairing of collaboration between these cards. Gemini from the beginning was designed as dedicated to the VR.

Not a clue about this so have to ask.. Do these dual GPU cards (AMD or Nvidia) act as one single card or are they basically crossfire/SLI on one board and will show up on a system as such?

They are basically SLI/CFX. Being on the same board usually results in better scaling, but not by much.

Too late to the party. Old tech.

With the driver optimizations coming to the Oculus runtime (asynchronous timewarp) and with the implementation of Gameworks VR and LiquidVR, a 980 Ti is already almost enough to max anything that'll come out for VR in the next year. This'll probably be for those looking to supersample (very small demographic) or to max every last setting. Star Citizen (arguably the game that'll be the toughest to run in VR) isn't coming out for a while and VR support will take even longer to come out.

I agree that a dual Gpu card with 1st generation HBM is rather pointless since it is delayed to 2nd QTR 2016. It would have been needed to be released some time this month or next month to be relevant. However there will be buyers for this there always are even though the buyers are few and far between.