-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
AMD Polaris architecture – GCN 4.0
- Thread starter Boomstick777
- Start date
Caporegime
- Joined
- 18 Oct 2002
- Posts
- 30,151
Ahhhhh FuryX2, beyond defunct at this stage 
Ahhhhh FuryX2, beyond defunct at this stage
honestly it has been defunct since last december. As soon as they delayed it 1st time, it was too late to market. If it was released september, I would have gotten one for myself. But now so close to polaris announcements, I just cannot justify buying last gen product.
Soldato
Any dual card is defunct when crossfire and SLI doesn't work correctly with *all* games. The only games in my gaming library i got working with crossfire (when i had 4x7970s for Bitcoin mining) was Half life 1 and 2, but who needs 400-600fps?!
Not an issue in VR. FuryX 2 is basically a niche VR market product.
Soldato
- Joined
- 30 Nov 2011
- Posts
- 11,356
What? How is it not an issue? VR games don't just magically work in crossfire, they still need support - are there any VR games that actually support multi-GPU right now?
Both AMD and Nvidia have released additional API's for enabling per-eye SLI/Crossfire, but no game is actually using them yet, so yes game support is still a massive issue for VR
And people are overblowing supposed xfire/sli support. I have multiGPU set up and am happy with it, since all the games which have problems be run at 4k on single GPU support xfire, those which don't support xfire run fine on single GPU.
Caporegime
Caporegime
We'll find out in 9 days ^^^^
Soldato
AMD will be talking a lot about Low abstraction API coding at GDC. They are also having a big talk on GPUopen and mGPU coding. So hopefully they will add mGPU code and examples to GPUopen and we should see some decent mGPU support coming as default.
I bet you they don't say much/anything about Polaris at GDC
It's going to be all about GPUOpen and that kind of thing.They may have some "demos" set up, like the power efficiency demo, but will not release any concrete information, as we're still too far from release. Still won't know price or performance by the end of it.
You wait and see
The problem is AMD twitter guys know how to talk the talk, but when it comes to us, tech geeks, they won't say anything new we already did not know 
I do hope they will surprise us with polaris info, but I really doubt. I expect endless push for recent Ashes bench results as a result of their fine work
I do hope they will surprise us with polaris info, but I really doubt. I expect endless push for recent Ashes bench results as a result of their fine work The problem is AMD twitter guys know how to talk the talk, but when it comes to us, tech geeks, they won't say anything new we already did not know
+1
Remember fury x? Plenty of fluff and no performance news until we got the benchmarks from reviewers.
Plenty of shouting about async performance in ashes will likely be the order of the day.
Caporegime
http://wccftech.com/sk-hynix-hbm2-mass-production-q3-2016/
so they are saying 4GB hbm2 stacks in Q3 '16 and 8GB hbm2 in Q4 '16
I read sometime ago about Samsung being a little ahead of Hynix in hbm2 production, even so, this would imply that hbm2 GPU's will not be on the market until Q3?(guess) and even then there WILL be stock issues.
i'm getting curious now on what Polaris 10 and 11 will be and what type of memory they will be using.
so they are saying 4GB hbm2 stacks in Q3 '16 and 8GB hbm2 in Q4 '16
I read sometime ago about Samsung being a little ahead of Hynix in hbm2 production, even so, this would imply that hbm2 GPU's will not be on the market until Q3?(guess) and even then there WILL be stock issues.
i'm getting curious now on what Polaris 10 and 11 will be and what type of memory they will be using.
Associate
Yeah a "little": http://www.digitaltrends.com/computing/samsung-mass-producing-hbm2/
They started mass production in January.
It doesn't make any sense. Hynix is developing the standard, and they license production to Samsung. So Hynix is waiting several quarters to start production while Samsung is pumping volumes out already.
Associate
Samsung $$$$ > Hynix $$$$ i guess.
Samsung is a bloody giant compared to Hynix, and they don't license it, HBM is an open JEDEC standard.
I think as soon as the standard was accepted the truckloads of money started rolling.