• 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 Confirms GCN Cards Don’t Feature Full DirectX 12 Support – Feature Level 11_1 on GCN 1.0, Featur

SiDeards73 said:
Hilarious... a thread about feature levels of an API that hasnt even been released yet, is probably not even final and arguing over who supports what of said non existant as of yet API.

The level of fail in this thread has reached a new high or should i saw all time low.

Seriously folks just lock this thread and move on, discuss it when it arrives, infact discuss it once the first few DX12 games have arrived.
Click to expand...

well for people buying graphics cards in the next few months, e.g. choosing between a 970/980/980ti and a "Fury" and a bunch of rebranded 290's, it could be important to know... as this info has come direct from AMD themselves it should be fairly reliable

after these next couple of cards come out it could be very late 2016 before a proper new batch come out, so based on what MS said, games will very much be out before the new batch of cards comes, so what people are buying now is the same cards we will be playing the first DX12 games on

given that Microsoft said that DX12 was already with some developers, GPU makers are obviously speccing for it, and games are supposedly less than 6 months away from supporting it, I would be extremely surprised is the spec for DX12 was not already locked down
 
So on Twitter AMD Roy is now saying Nvidia can't do Asynchronous Shaders.

Which is confusing me. If this is a feature of DX12 and Nvidia can't do it how can they support 12_1?
Are there features of DX12 that don't come into the listed tiers?
 
Asynchronous shaders aren't really a feature of DX12, they are a feature of GCN, built in to the hardware, it would be like trying to make CUDA a mandatory part of DX12, DX12 supports AS bit they cant make it mandatory as the Asynchronous Compute Engine built in to GCN is proprietary to AMD

nvidia have HyperQ which is similar but different and nvidia haven't done the same type of PR piece showing if/how DX12 can work with that
 
Last edited:
Geck0 said:
What? a 2+yr old architecture (Hawaii) doesn't fully support a yet to be released API?

ZOMG!!!
Click to expand...

Yeah it is not really surprising and not really a big issue because by the time games really support it AMD will have a load of new cards out that do support 12.1 features (if AMd are still around).

What gets my goat, and many others who are just interested in technology, is the constant FUD from the AMD Fanboys shouting how amazing AMD are and how useless Nvidia are, how only GCN is fully DX12 compliant etc, etc.
 
andybird123 said:
Asynchronous shaders aren't really a feature of DX12, they are a feature of GCN, built in to the hardware, it would be like trying to make CUDA a mandatory part of DX12, DX12 supports AS bit they cant make it mandatory as the Asynchronous Compute Engine built in to GCN is proprietary to AMD
Click to expand...



This, It is like saying GSYNC should be part of DX12. It is a proprietary technology.
 
GoogalyMoogaly said:
So on Twitter AMD Roy is now saying Nvidia can't do Asynchronous Shaders.

Which is confusing me. If this is a feature of DX12 and Nvidia can't do it how can they support 12_1?
Are there features of DX12 that don't come into the listed tiers?
Click to expand...

It's probably best to wait for a slightly more credible source than AMD's PR given their recent escapades.

Also, somebody tell Roy about HyperQ.
 
Last edited:
Kaapstad said:
NVidia Maxwell does not support the full implementation of DX12 either.

Best wait for DX12 and Win10 to see what is what in real use rather than debating theoretical points of order.
Click to expand...

/endthread. We have no idea how these feature sets are going to be implemented, what timescales they'll be implemented on etc. By the time 12.1 becomes relevant we might be on a whole new generation of cards. It might be amazingly important, but I won't be making my decision based on this. It'll be down to performance and cost.
 
Why is this even an issue? It's such a non point I can't fathom how people have got into a crap slinging match over it. Some older tech supports a certain level of DX12, some other older tech doesn't, both vendors newer tech supports it fully. By the time DX12 is ubiquitous all these cards will be pretty irrelevant. It's sad how some blind fanboys get so worked up and butt heads over such things.
 
Kirby Wurm said:
Why is this even an issue? It's such a non point I can't fathom how people have got into a crap slinging match over it. Some older tech supports a certain level of DX12, some other older tech doesn't, both vendors newer tech supports it fully. By the time DX12 is ubiquitous all these cards will be pretty irrelevant. It's sad how some blind fanboys get so worked up and butt heads over such things.
Click to expand...

Thats just this forum, if one "side" has something that the other may not then it suddenly becomes a massive issue like its some kind of be all end all feature.
 
Well i have to search for the article, but i remember that the hawaii chips are much stronger (faster) in async compute than maxwell chips.
Edit:
Sorry this is the correct compute data:

AMD GCN 1.2 (285) 1 Graphics + 8 ACEs = 64+ Compute (64+ queues)
AMD GCN 1.1 (290 Series) 1 Graphics + 8 ACEs = 64+ Compute (64+ queues)
AMD GCN 1.1 (260 Series) 1 Graphics + 2 ACEs = 16+ Compute (16+ queues)
AMD GCN 1.0+ (Kabini) 1 Graphics + 4 ACEs = 8+ Compute (8+ queues)
AMD GCN 1.0 (7000/200 Series) 1 Graphics + 2 ACEs = 4 Compute+ (4+ queues)*
NVIDIA Maxwell 2 (900 Series) 1 Graphics + 1 Compute = 32 Compute (32 queues)
NVIDIA Maxwell 1 (750 Series) 1 Graphics = 1 Compute (32 queues)
NVIDIA Kepler GK110 (780/Titan) 1 Graphics = 1 Compute (32 queues)

*S.I. GCN support 2 hardware compute queues per ACE

Kepler and Maxwell 2 has 1 graphics and compute device (32 queue)
Maxwell 2 has a separate compute device (32 queue)
GCN 1.0 and 1.1 (260) has 2 ACE unit each handles 8 compute jobs (16 queue)
GCN 1.1 (other) and GC 1.2 has 8 ACE units each handling 8 compute jobs of 64 total queue.


Async compute is the biggest bang in DX 12 IMHO...finally handling task parallel can bring a great performance jump
 
Last edited:
Gotta love the fan boys of both sides. This is just a repeat of the same guff as the last time. We support tier 3 and you don't, nah nah nah nah nah. Aye but we support 12.1 and you don't, nah nah nah nah nah. Grow up the lot of you, honestly. Both sides will absolutely fully support DX12, end off. :D
 
Gregster said:
AMD not being able to do ROVs or Conservative Rasterization and these are 2 important parts of DX12.
Click to expand...

Please do tell us why you believe these two features to be such an important part of DX12. I'd love to hear your factual and technical thoughts on the matter.

Feel free to include the wealth of future game titles that are confirmed to be implementing these two features.
 
Dave2150 said:
Please do tell us why you believe these two features to be such an important part of DX12. I'd love to hear your factual and technical thoughts on the matter.

Feel free to include the wealth of future game titles that are confirmed to be implementing these two features.
Click to expand...

+1 please do Greg

From my understanding these two features are there to help devs work easier and not something we will see in a game I.E Graphics or performance!

Information here!
http://www.anandtech.com/show/8544/microsoft-details-direct3d-113-12-new-features
 
You must log in or register to reply here.
Back
Top Bottom