AMD Cayman info (or rumor)

[waway625]

Here is the performance score...

Under the GTX 570 score....

GTX 570 @ stock + Core i7 950 @ 3.7Ghz

vantage

3DMark Score
25841.46 3DMarks
Graphics Score
...21606.09
CPU Score
62734.18
Jane Nash
64.95 FPS
New Calico
61.57 FPS
AI Test
3235.9 operations/s

3dmark 11

3DMark Score
X1701
Graphics score
1540
Physics Score
7992
Combined Score
1794
GraphicsTest1
7.85 FPS
GraphicsTest2
8.03 FPS
GraphicsTest3
7.45 FPS
GraphicsTest4
4.74 FPS
PhysicsTest
25.37 FPS
CombinedTest
8.34 FPS

[Nintendork]

Without 10.12 it's useless to post more "leaked benchmarks". This will be hell if we continue for 3 days more. As for now, I'm more interested in chip/pcb nude pics.

The 6970 vs 480 powerconsumption comparison of hardwareluxx was done with an undervolted 480.

[SpuTnicK]

Source

[cold2010]

Um ,no.
all resolutions :
5870 90
580 112 => 112/90=1.25x

1024x768:
5870 87
580 104 => 104/87=1.19x

1280x1024
5870 89
580 111 => 111/89=1.24x

1680x1050
5870 89
580 113=> 113/89=1.27x

1920x1200
5870 91
580 115=> 115/91=1.26x

2560x1600
5870 92
580 116 => 116/92=1.26x

As I've written,25% is very doable in games . End users could care less for Synthetic benchmarks.
In the end,we get smaller chip that performs on par with 40% larger chip,while costing the same or less.

http://www.computerbase.de/artikel/g...ating_mit_aaaf

[informal]

Sorry,i trust techpowerup more.

[Dami3n]

From Catalyst 10,12:

AMD Radeon 6950 - HD 6900 Series "= ati2mtag_NICayman, PCI \ VEN_1002 & DEV_ 6718
AMD Radeon 6970 - HD 6900 Series " = ati2mtag_NICayman, PCI\VEN_1002&DEV_6719

Is the german user testing a 6950?

[DarthShader]

Source

Review sites trying to generate hits, compiling info from the internets.

That theoretic peak performance table is hardly insightful either.

Sorry,i trust techpowerup more.

You should look closely at the choice of benchmark's used there. Pick a SLI/crossfire review to see the issue at a more exegerated level.

While their review can be used to compare relative positions of cards, or to look for driver problems, it's unreliable for direct percentage comparisions of cards, because the result differences are flattened. I hope W1zzard will update the choice of benchmarks to a more 2011 level soon, cause I like their overall methodology.

That computerbase.de table isn't a reliable indicator either, since it compiles only 4xAA results, which is a "best-case" for nVidia. At noAA and 8xAA the differences are a bit smaller.

[Dimitriman]

I feel that many people were right to call to question AMDs ability to deliver Antilles with 2x 1920 shaders and it seems like AMD chose to castratre Cayman in order to be able to build and deliver Antilles. Personally I would not have cared if Antilles were 2x 1536 shaders while Cayman was 1920, Cayman had to be a good card so that Amd could escape the position they were in vs the 400s but instead they chose to be at the exact same spot as last generation. If you look at these numbers and they are true, nothing has changed, the 6900 series will be in the exact same position as the 5800 series were all of 2010 without the added 6 month glory time.

[TheBreezyBB]

Let's just hope that the card ( 6970 ) isn't slower than the GTX580 and @ the same price. This alone would make the card Utter failure IMO!!

[eric66]

I't is when you don't know what conditions they were testing under, whether the drivers fully supported the card etc....

You will only find this out on Wednesday.

Andy '

hmm now i start to think that this german guy doesn't have real cayman drivers .....

[TheBreezyBB]

I'm actually waiting to decide on what to get in January.

[zalbard]

Fully supported GPU-z SS from @Ducati750ss:

http://www.abload.de/img/gpu-zneuvxuk.jpg

Ok, so 1536 shaders it is then. The number is a bit disappointing.

http://bbs.expreview.com/thread-38497-1-1.html

I think we've seen a part of this slide before. I think the gains are nice. The question is, is it faster than GTX570 or not? Seems pretty close. I like what I'm seeing!

Also, for reference, the Computemark score of 454 sounds bad compared to the 580

http://h-5.abload.de/img/6970computemarkxmc9.jpg

But look at what the 5870 scores:

http://www.geeks3d.com/20100606/gpu-...80-vs-hd-5870/

238!

The 6970 scored 90.7% higher than the 5870 in the same compute bench!

I suppose it's just optimised for Nvidia.

Still loses in Heaven? Odd...
Other than that, looks somewhat decent.

According to @Ducati750ss, these are today's last benches from him:

Bad Company 2:

GTX 480 = min 42 , max 124 , avg 77,0

HD6970 = min 25 , max 130 , avg 70.4

Very bad min fps.
Hopefully 6970 will be ahead of GTX480 min FPS wise, this is one area where 5870 was seriously lacking.

Source

They should have boosted ROPs more. Other than that, looks more balanced than GTX570.

From Catalyst 10,12:

AMD Radeon 6950 - HD 6900 Series "= ati2mtag_NICayman, PCI \ VEN_1002 & DEV_ 6718
AMD Radeon 6970 - HD 6900 Series " = ati2mtag_NICayman, PCI\VEN_1002&DEV_6719

Is the german user testing a 6950?

No, he is not. Let's wait for the reviews...

[Nintendork]

Ducati750ss failed hard. All the nvidia gadgets installed and who knows what.

[blindbox]

From Catalyst 10,12:

AMD Radeon 6950 - HD 6900 Series "= ati2mtag_NICayman, PCI \ VEN_1002 & DEV_ 6718
AMD Radeon 6970 - HD 6900 Series " = ati2mtag_NICayman, PCI\VEN_1002&DEV_6719

Is the german user testing a 6950?

I raeged if that's true.

http://www.xtremesystems.org/forums/...eply&p=4662941

EDIT: Removed oversized image. Post below describes the true situation I hope.

[cold2010]

From Catalyst 10,12:

AMD Radeon 6950 - HD 6900 Series "= ati2mtag_NICayman, PCI \ VEN_1002 & DEV_ 6718
AMD Radeon 6970 - HD 6900 Series " = ati2mtag_NICayman, PCI\VEN_1002&DEV_6719

Is the german user testing a 6950?

DEV_6719 = HD 6950

DEV_6718 = HD 6970



Source

http://www.arabhardware.net/forum/sh...postcount=1633

[andos]

lolz.

Look in description.

http://www.czechcomputer.cz/product.jsp?artno=84725

http://www.czechcomputer.cz/product.jsp?artno=84726

[TheBreezyBB]

Forget that Seems that this wasn't the case.

[badboy18187]

Like few people have said, until trusted sites actually post their reviews, I won`t believe in any of these leaked "real" benchmark numbers.BTW, I`m sorry if I missed it, but did AMD disclosed the fact that HD6970 has only 1536 SP`s?

[Nintendork]

The mod of that forum says Ducati doesn't even have a 6970...

[-Sweeper_]

if cayman XT (250W TDP?) cant beat a fully spec'd GF100 I'll be really disappointed

lets just wait a few more days..

[zalbard]

fully supported gpu-z ss from @ducati750ss:

6718! So, is it 6850 that he is testing with a 6870 sticker?! O.o
Then again the VRAM clocks appear to be too high for 6850 part...

[cold2010]

6950/6970 GPUz 0.5.0

[Hicksetto]

From now on @harry97 started to compare HD5870 vs. HD6970 with the same card!

http://www.abload.de/img/69704mxh.jpg



http://3dmark.com/3dmv/2791306

http://3dmark.com/3dmv/2791228

Keep in mind @harry97's resolution is 1680x1050!

[Hicksetto]

HD6970 vs. HD5870

RE5:
http://www.abload.de/img/5870wcb1.jpg

http://www.abload.de/img/6970qeqs.jpg

Unigine Heaven 2.1:
http://www.abload.de/img/6970_uni6dnx.jpg

Table:

[TheBreezyBB]

So the Texture Fillrate on 6950 is the same as the one in Ducatis750ss photos! interesting!!