If the dual gpu has the same specs and clock speed of 2 separate cards they will perform the same. It ain't rocket science.
http://www.legitreviews.com/amd-radeon-r9-fury-x-4-way-crossfire-setup-benchmarked_167338 LegitReviews has some bench's by AMDmatt, with 2, 3 and 4 Fury X's in Crossfire. The GPU scaling is very very good even to 4 cards!!! even if it is just on 3Dmark. Can't wait to see some more Crossfire reviews.
Apparently a person behind the tests can't do math for scaling. Thankfully, readers in the comments fixed it for him/them. What does this reminds me off...hmmm...
I think Crossfire is the real bright spot for the FuryX compared to the 980 Ti, since they seem to throttle a lot less than the poor space-compressed and throttling 980 Ti's due to the water cooling. On the other hand, their computer performance is not so spectacular as was made out previously in the thread: http://www.anandtech.com/show/9390/the-amd-radeon-r9-fury-x-review/24
People I know do not have a problem with throttling on 980ti sli. I'm going to join that club very soon. Do not have throttling on my current cards that run a bit warmer.
Yeah you will see a good increase with 1 980ti, I wouldn't with 1 card though. I been buying sli for over 10 years now had to change.
Here's where I'm going to have to respectively disagree with you. What you're providing is mere anecdotal evidence. However, several review sites I've seen doing SLI reviews with the 980 Ti has seen the cards throttle. Could it be they are doing it wrong? Perhaps. However, these are the same sites that we are using to argue benchmarks with each other, and provide a level and controlled conditions. Therefore, I'm going to have to stick with the cards throttling as a rule (with your "people I know" being the exceptions) until review sites point out otherwise.
Thing is most review sites (this one included) use a test bench. Hilbert has not done a 980Ti SLI review. Open air bench "cases" show a huge weakness of the reference cooler.
Perhaps, perhaps not. For that matter, not all the review sites used an open test rig. Bottom line is, until there are reputable review sites that show otherwise, the 980 Ti throttling will just have to stand as a point of fact in our discussions, the same way we say the Fury X uses more power and delivers less performance than the 980 Ti from those same reviews (and we take it as gospel, otherwise it's a "he said, she said" world on this forum and any discussion within the scientific method is pointless, and it becomes an opinion circle-jerk.
There are more reason for card to throttle. even if you can keep it at 45°C, it still can throttle once power limit in vBIOS is reached. My HD7970 was throttling a lot in experiments with TDP as I lowered it to 175W. That means reviewers could have card(s) which needed more juice for given clock than allowed. And different kind of workloads put different load on GPU even if each shows same 100% utilization. (One of reasons why review sited show sometimes quite different power consumption.) But TDP throttling has simple solution, people just have to allow higher TDP limit and who cares if it eats 25/50W more.
Get a load of this bench: https://www.youtube.com/watch?v=6jAeLS-gmbo Also, Look at the other videos that the uploader posted, Amazingly at same exact settings, The Fury X eats alot less vram then the Titan X, Some of his videos are almost the same vram usage, This is very interesting. Strange, the vram usage count on titan X is alot more then fury X, Could be how the hbm works. Hmmm, I can't wait for more updated drivers, I am looking foward for more juice coming out of this card and more voltage tweaks and overclockin', Can't wait for results in the future. I am looking foward to the dual r9 fury x coming out very soon, what? this 15th?
Battle of the kings, L@@K: https://www.youtube.com/watch?v=fFpy2L3B8lk Reason why the Titan X and the 980 Ti Oc stands out more is because of alot more overclocking potential headroom, whereas the r9 fury x can't overclock as much, If it had the same overclocking headroom, the r9 would win, on the overclocking part.
No throttling here and I'm on air - in fact, I can boost them further than I do and they still don't throttle. I've yet to see them break 75c tbh.
I think if the R9 overclocking potential opens up as much as the 970/980 series, I strongly believe, the r9 fury X would easily win. Hopefully Rivatuner fame: !Unwinder!, can unlock great hidden potential within this card. When driver fully matures + Unwinder magic = R9 Fury X FTW. Actually looking more into the dual R9 fury X version on a single pcb. Scaling on xfire on the fury X seems Absolutely fantastic. And damn! 41 Celcius underload on some games, WOW. Waiting to see the normal fury card with the traditional air fans, love to see the difference in temps.
R9 Fury has close to zero OC potential but ok, maybe it will "open up". btw. that YT test is bs. Everyone knows 4th card in Sli barley scales 40%ish...