FUD Sea Islands AMD supports DirectX 11.1

MentalIlness

Platinum Member
Nov 22, 2009
2,383
11
76
http://www.fudzilla.com/home/item/27897-sea-islands-amd-supports-directx-111

Kabini APU to have it


We finally managed to learn a few details about AMD’s upcoming Sea Islands graphics cores. AMD told some of its partners that Radeon HD 8000 is a good candidate for an official name and now we got some news that the new cores will support yet to be announced DirectX 11.1 API.

Richland A series Trinity successor in 28nm will end up with current generation Northern Islands Radeon HD 7000 cores, while the Kabini gets Sea Island Radeon HD 8000 cores. Many expect to see Sea Islands in discrete and mobile graphics form this year, but this is not something we cant confirm at this time.

Despite the fact that some APUs might end up with the same number or graphics cores (shaders) AMD is quick to point out that next generation cores will be different and that e.g. 384 Sea Islands cores won’t be the same as 384 cores of Northern Islands. In a perfect world the new generation should end up significantly faster.

Nvidia has Maxwell scheduled for 2013 and according to the roadmap from GTC 2010 it is expected in 2013 with double precision of some 15 Gflops per Watt, some three times more than with Kepler. We hope both players in the graphics duopoly will put up a good fight as every industry needs competition to drive prices down and keep innovation up.
 

OCGuy

Lifer
Jul 12, 2000
27,224
37
91
DX 11.1? Hopefully more useful than DX10.1. Hell, hopefully more useful than DX11 up to this point.

Marketing gimmick.
 

ViRGE

Elite Member, Moderator Emeritus
Oct 9, 1999
31,516
167
106
Afaik the current gen already supports DX11.1
Bingo. I would sure hope Sea Islands supports D3D 11.1, seeing as how Southern Islands already supports it.:p
 
Last edited:

-Slacker-

Golden Member
Feb 24, 2010
1,563
0
76
I guess FUD has grown tired of being the butt of all the jokes, and now they're trying to play it safe with a "rumor" that everybody on the planet already knew. It goes without saying that sea islands will support dx 11.1, because the 7000 series already does.
 

BallaTheFeared

Diamond Member
Nov 15, 2010
8,115
0
71
Nvidia has Maxwell scheduled for 2013 and according to the roadmap from GTC 2010 it is expected in 2013 with double precision of some 15 Gflops per Watt, some three times more than with Kepler. We hope both players in the graphics duopoly will put up a good fight as every industry needs competition to drive prices down and keep innovation up.


Bwahahahahahaha.

Yeah right. Maxwell in 2013, good luck with that.
 

f1sherman

Platinum Member
Apr 5, 2011
2,243
1
0
cudafll01.png


http://www.nvidia.com/docs/IO/113297/ISC-Briefing-Sumit-June11-Final.pdf
 

HurleyBird

Platinum Member
Apr 22, 2003
2,814
1,550
136
owned

As a sidenote, seeing as 20nm isn't likely to come online until 2014-ish, one has to wonder what the Nvidia lineup is going to look like. AMD is refreshing the entire product line, while all we know about Nvidia is that they are going to launch big-K. Are the current 6xx parts then going to be re-branded? Or do they have additional new chips in store?
 
Last edited:

Arkadrel

Diamond Member
Oct 19, 2010
3,681
2
0
@ f1sherman

Nvidia limit the double precision floating point performance in the 680.
So them going from 6 ->15 or whatever doesnt seem so impressive when currently it looks like this:


I ran my Sandra 2012SP2 benchmark that I carry around on a USB stick to check GPGPU compute performance in floating point, especially double precision.
nv680compute.JPG



from: http://www.theinquirer.net/inquirer...hed-amds-mid-range-radeon-hd-7870-gpu-compute

Wow! The claim of beating the HD7970 goes right into the thin air, it seems. Nvidia's new GPU is beaten by the Radeon HD 7970 by an order of magnitude here in double precision floating-point, as well as nearly twice in ordinary single precision floating point. One is speechless here. Even the Radeon HD 7870 with its restricted double precision floating-point still outperforms the GTX680 by a noticeable margin in this department, as you can see here.


Looks like what? 40% faster in single, and 500% faster in double? The 680 loses out alot to the 7970.

So ofcourse the next gen called "maxwell" will have better DP performance pr watt... what does that mean?
probably that nvidia wont limit the card artifically with next release.
 
Last edited:

f1sherman

Platinum Member
Apr 5, 2011
2,243
1
0
@Arkadrel

K20 is full-blown DP card from that slide.

NVDA_TeslaK10_K20_689.jpg


@Phynaz what gave you the idea Kepler would be 3x faster than Fermi?

Even SP does not translate linearly to rendering performance, as seen with 7970 vs 680.
 

ShintaiDK

Lifer
Apr 22, 2012
20,378
146
106
So ofcourse the next gen called "maxwell" will have better DP performance pr watt... what does that mean?
probably that nvidia wont limit the card artifically with next release.

Om sure the next GK114 or whatever will be DP limited as well. Remember GK104 aint the highend/Tesla chip. GK110 is.

Just look where the HD7870 is placed on that chart.
 
Last edited:

Spjut

Senior member
Apr 9, 2011
933
163
106
I still haven't found any article explaining what DX11.1 is better for in gaming. Draw calls is apparently becoming a big issue on the PC, hopefully there's some improvements for that.

Guessing(hoping!) DX12 will come with the HD9000/GTX800 series.
 

ShintaiDK

Lifer
Apr 22, 2012
20,378
146
106
I still haven't found any article explaining what DX11.1 is better for in gaming. Draw calls is apparently becoming a big issue on the PC, hopefully there's some improvements for that.

Guessing(hoping!) DX12 will come with the HD9000/GTX800 series.

http://msdn.microsoft.com/en-us/library/windows/desktop/hh404562(v=vs.85).aspx

Direct3D in session 0. Doh.

I dont get the sudden inflation in DX versions. A fully utilized DX10 game would crawl on a GTX680/HD7970GE.

Seems right now we get new DX versions for the sake of new versions. Not because it serves a purpose. And thats actually damaging.
 

Phynaz

Lifer
Mar 13, 2006
10,140
819
126
@Phynaz what gave you the idea Kepler would be 3x faster than Fermi?

Even SP does not translate linearly to rendering performance, as seen with 7970 vs 680.

Because in the chart Fermi is at 2, and Kepler is at 6.
 

GotNoRice

Senior member
Aug 14, 2000
329
5
81
DX 11.1? Hopefully more useful than DX10.1. Hell, hopefully more useful than DX11 up to this point.

Marketing gimmick.

Just wondering, what are you basing this conclusion off of?

While it's true that the number of games written specifically for DX10.1 was quite small, that doesn't mean DX10.1 wasn't ultimately utilized.

When you run a DirectX11 game on DirectX10.1 hardware, it makes use of the DirectX10.1 feature level. That is a different feature level than the DirectX10.0 feature level. That means you're taking advantage of DirectX10.1 features almost every time you run a DirectX11 game. I'd hardly call that a gimmick.

Let's also remember that the primary reason DirectX10.1 didn't see wide support was that Nvidia never supported it until the very end and even then just on their low-end cards. This time, DirectX11.1 has good support from both companies.

DirectX11.1 also makes use of feature levels, meaning it will run directly on DirectX11.0 hardware (along with DX10.1 and DX10 hardware) as compatibility is as simple as shedding the small subset of features that are not supported. Adoption should come quickly. And then, when DirectX12 ultimately arrives, you'll be playing your games using the DirectX11.1 feature level instead of the 11.0 feature level.

DirectX11.1 features will be taken advantage of one way or another.
 
Last edited:

Anarchist420

Diamond Member
Feb 13, 2010
8,645
0
76
www.facebook.com
DX11.1 is no marketing gimmick because of bindless textures and DX10.1 made many things mandatory that should've been mandatory with DX9. I wish nvidia wouldn't limit their double precision fp performance though.
 

BallaTheFeared

Diamond Member
Nov 15, 2010
8,115
0
71
Nvidia makes money with these cards, they purposefully limit their capability in these areas.

They sell it to us for less than $500 then turn around and sell the non disabled version for $2000-$6000.
 

ShintaiDK

Lifer
Apr 22, 2012
20,378
146
106
I honestly couldnt care less about DP performance on a gaming card. Now GK104 aint nVidias DP card. Gk110 is. Just like HD78xx got horrible DP performance around GK104 levels.

And why would I want to be penalized in terms of power consumption, size and performance to get DP? If GTX680 vs HD7970GE is the case for DP. Then screw DP.
 

ViRGE

Elite Member, Moderator Emeritus
Oct 9, 1999
31,516
167
106
DX11.1 is no marketing gimmick because of bindless textures and DX10.1 made many things mandatory that should've been mandatory with DX9. I wish nvidia wouldn't limit their double precision fp performance though.
Bindless textures isn't a D3D 11.1 feature.