I say an extra 6-7month wait for Vega or Big Pascal to launch would be the wiser decision.
Seems to be the wiser choice indeed, but even then there has to be games that are really GPU limited and kick the **** out of the current cards, but it might still take some time until they're out, and the current cards still are hanging in there even at 4K and playing all my current games with the highest in game quality settings enabled by default......My standard procedure.
One of the likely candidates is star citizen, and the current alpha does indeed have some kickass graphics and present a much higher load for the GPU's, even if it's no where close to be considered a game yet.......Fact is that who knows when the game will get done, even though there are some 300 people working on the project.
One of the ways the Radeon Pro Duo is being marketed, is by adding up all the hardware between both GPU's to make it sound more impressive, but I can play that game too:
11 200 shaders in total.
700 texture units.
256 Rops.
8GB of ram per GPU ( 32 GB total ).
1.3 TB/sec of memory bandwidth.
Fact is that by computer terms it's old hardware and not the latest thing, not the fastest on a per GPU basis in the least but when all that hardware is working together, it's still a lot of graphics power as long as there's a crossfire profile for a given game.....A weakness that the Radeon Pro duo also shares, just like any other multi GPU setup.....
I basically get the feeling that if I changed the cards, and even though the latest thing is faster ( this also applies to big Vega and Big pascal ), i'm still going to get pissed with myself with spending a lot of money, which I can afford to that's not the issue, and prematurely ending their stay in my system when the software to justify said upgrade just isn't there ( yet ).
We're all playing console ports for the most part.....
Looks good only if you want a single card in your rig to minimize heat or some such.
Looks as if it will be curb stomped with dual 980 Tis.
Heat isn't an issue with the amount of water cooling i'm using......My current cards and even with 4 of them, have a hard time going above 50*C in the middle of summer at full load between all 4......Having 3 triple thick EK radiators in their own dedicated loop just for the GPU's does that.....
Same goes for power.....I have a pair of single rail corsair 1200 I PSU's in parallel, with the overall system load evenly split between both PSU, so even in the worst cases at full load, neither one has ever exceeded 700 watts.....Well below their respective maximum possible output, and being single rail units, there's a combined 200 amps at the 12v line for the GPU's to feast upon ( 2400 watts for that line alone between both PSU's).
So even running a pair of Radeon pro duo's and using the worst case scenario where they can draw up to 525 watts for each one ( triple 8 pin + the 75 watts from the PCI slot ), it's still a light load for the PSU's......Bring it on.... The setup is built like a brick shithouse on the cooling and power fronts....
True, Nvidia could release a dual GPU card using a pair of GM 200's too, but given that the next step has already been shown in a tesla version ( Pascal GP 100 ) with it's 15.3 billion transistors, and that AMD has tipped it's hand with the higher end Vega being the one that is meant to compete with it, the release of such a card at this point in time is unlikely.
AMD released this one simply because it's been shown since the middle of last year, but I did hear that the original plan was to use Nano's in the design with their lower clocks and much lower power use, but was redesigned to use the Fury X versions for the version we're getting......Hence the triple 8 PIN power setup.....That's 450 watts on their own that can be delivered to the card.