AMD Cayman info (or rumor)

[zalbard]

god, this thread went to hell

Agreed...

[generics_user]

not again...

[SKYMTL]

additionally somethign is fishy with that hardware canucks chart, the difference between 4xAA and no AA is significantly higher than in every other review out there were 5xxx and 4xx/5xx AA scaling is much closer, their results are completely different to most user tests and professional reviews and completely overrated....

Oh? I have have yet to see a review where AMD's cards don't take a massive hit when AA is enabled. It's an architectural limitation which AMD will likely resolve with upcoming products.


As for the results, I am 100% confident in saying that on average with the latest WHQL drivers, HD 5870 is at least 30% slower at 1920 x 1200 4xAA than the GTX 480 in newer games. I'd even venture that this carries over into quite a few DX9 games as well but it won't show up on charts simply due to CPU bottlenecks artificially impacting the scores.

To me, DX11 and to a lesser extent DX10 really is all that matters in terms of performance when testing $150 and above GPUs. Even a lowly GTS 450 has no issue pushing out 60 FPS+ in the current crop of DX9 games at 1920 x 1200.

[Oliverda]

[tbone8ty]

http://www.fudzilla.com/games/item/2...r-than-gtx-480

10-20% faster than gtx 480 does not beat gtx 580 and the price reflects it.

[Oliverda]

http://www.fudzilla.com/games/item/2...r-than-gtx-480

10-20% faster than gtx 480 does not beat gtx 580 and the price reflects it.

"AMD's own benchmarketing data claims that Cayman XT is slightly below 20 percent faster than a GTX 480 in 3Dmark Vantage. "

Fuad has eaten the fake slide... again..

[TheBreezyBB]

http://www.fudzilla.com/games/item/2...r-than-gtx-480

10-20% faster than gtx 480 does not beat gtx 580 and the price reflects it.

You better go check 2 pages back to see this and... oh well you should figure it out if you look at the slides

[generics_user]

http://www.fudzilla.com/games/item/2...r-than-gtx-480

10-20% faster than gtx 480 does not beat gtx 580 and the price reflects it.

this article is so wrong in so many ways (GTX 570 over 20% faster than GTX 480?????

we know too little to belive any rumor out there they go form "best card ever" to "almost no improvement", i think that it's somewhere in the middle but performance slower than gtx570 is highly unlikely....

[tbone8ty]

Regardless Antilles will be insane!

[Dezmen]

Stop post duplicated news -.- And FAKE duplicated... xD

[Heinz68]

Actually, this article you refer to will only be proven true or false when trusted benchmark numbers come to light.

Did you bother to read the article?
Just to help you here is the article heading: "Radeon HD 6970 is 10 - 20% faster than GTX 480"

In the article he says "When we compare our own GTX 480 results and add 20 percent on top of that it turns that Cayman XT loses to both GTX 570 and especially GTX 580."

Most reviews had the GTX 570 about as fast as the GTX 480.
So can you explain how the HD 6970 can be slower than GTX 570?

Do you believe the HD 6970 is going to be slower than GTX 570?
Do we still have to wait for for trusted benchmark to prove this article is FUD based on FAKE slide?

It's just another Fuad negative article about AMD plus he has problem with his calculations.

I'm surprised you want to defend such an obvious FUD or is Fuad on the right site of the fence?

[safan80]

Oh? I have have yet to see a review where AMD's cards don't take a massive hit when AA is enabled. It's an architectural limitation which AMD will likely resolve with upcoming products.


As for the results, I am 100% confident in saying that on average with the latest WHQL drivers, HD 5870 is at least 30% slower at 1920 x 1200 4xAA than the GTX 480 in newer games. I'd even venture that this carries over into quite a few DX9 games as well but it won't show up on charts simply due to CPU bottlenecks artificially impacting the scores.

To me, DX11 and to a lesser extent DX10 really is all that matters in terms of performance when testing $150 and above GPUs. Even a lowly GTS 450 has no issue pushing out 60 FPS+ in the current crop of DX9 games at 1920 x 1200.

Performance with the 10.10e drivers worst than the 10.10d drivers for. The 10.10 whql and 10.11 whql drivers are worst for me than the 10.10d. Honestly I could care less about whql drivers they suck from both nvidia and ATI.

[E30M3]

i think that the 6970 will be slower than the 580 by 5-10%
and even slower if the driver quality output setting, is the same as nVidia. (No cheats)

[generics_user]

Oh? I have have yet to see a review where AMD's cards don't take a massive hit when AA is enabled. It's an architectural limitation which AMD will likely resolve with upcoming products.


As for the results, I am 100% confident in saying that on average with the latest WHQL drivers, HD 5870 is at least 30% slower at 1920 x 1200 4xAA than the GTX 480 in newer games. I'd even venture that this carries over into quite a few DX9 games as well but it won't show up on charts simply due to CPU bottlenecks artificially impacting the scores.

To me, DX11 and to a lesser extent DX10 really is all that matters in terms of performance when testing $150 and above GPUs. Even a lowly GTS 450 has no issue pushing out 60 FPS+ in the current crop of DX9 games at 1920 x 1200.

sorry to burst your bubble but you are the only review site showing such extreme differences in AA scaling between amd and nvidia

sure amd loses some 5-10% more fps in AA but in your summary it seems like it takes a hit 2-3 times bigger than nvidia which is the complete opposite to every single other review out there

[safan80]

sorry to burst your bubble but you are the only review site showing such extreme differences in AA scaling between amd and nvidia

sure amd loses some 5-10% more fps in AA but in your summary it seems like it takes a hit 2-3 times bigger than nvidia which is the complete opposite to every single other review out there

This is nothing new, when the did their 580 GTX review they had the 5970 in it which beat the 580GTX and they chose the 580GTX over the 5970 without even talking about the fact that people who already owned a 5970 would not want or need to change to the 580GTX. Their reviews are not the best anyways, they don't offer anything different (competitive advantage wise) than other review sites. What is worst is they don't like to test crossfire or SLI much, but when they do they prefer to do SLI. They are pretty much a Nvidia mouth piece kind of site.

[xdan]

Just to help you here is the article heading: "Radeon HD 6970 is 10 - 20% faster than GTX 480"
In the article he says "When we compare our own GTX 480 results and add 20 percent on top of that it turns that Cayman XT loses to both GTX 570 and especially GTX 580."
Most reviews had the GTX 570 about as fast as the GTX 480.
So can you explain how the HD 6970 can be slower than GTX 570?
Do you believe the HD 6970 is going to be slower than GTX 570?

GTX 480 is 13-15% slower than GTX 570 in Vantage.

Still my guess is that 6970 will be at midle between GTX 570 and GTX 580.
Probably 6950<GTX570<6970< gtx 580.

[SKYMTL]

Did you bother to read the article?
Just to help you here is the article heading: "Radeon HD 6970 is 10 - 20% faster than GTX 480"

Most reviews had the GTX 570 about as fast as the GTX 480.
So can you explain how the HD 6970 can be slower than GTX 570?

I never said which PART of the article was true or false since Fudo contradicted himself more than once as you can see by the highlighted portions above.

sorry to burst your bubble but you are the only review site showing such extreme differences in AA scaling between amd and nvidia

sure amd loses some 5-10% more fps in AA but in your summary it seems like it takes a hit 2-3 times bigger than nvidia which is the complete opposite to every single other review out there

What in the world are you talking about? 2-3 times? I see a 15% loss on average between AMD and NVIDIA when AA is enabled. The true difference runs the gamut from 0 all the way up to ~20%.

[Oliverda]

GTX 480 is 13-15% slower than GTX 570 in Vantage.

Still my guess is that 6970 will be at midle between GTX 570 and GTX 580.
Probably 6950<GTX570<6970< gtx 580.

A rumor from few days ago:

"6950 X score is good against the 480, P score is good against the 5970... weird.. "

[Sam_oslo]

GTX 480 is 13-15% slower than GTX 570 in Vantage.
http://lab501.ro/wp-content/uploads/2010/12/vantage_p_gpu.jpg[/IMG]
Still my guess is that 6970 will be at midle between GTX 570 and GTX 580.
Probably 6950<GTX570<6970< gtx 580.

I hope this doesn't make a new inferno, but when I said something similar a couple of weeks ago, it created a lot of feelings, to put it in a mild way.

The timing for GTX570 suggests that it was meant to fight 6970, but it's positionable that AMD has improved some (using that handy time of component shortage ). But we have to wait and see, and it will be interesting to see some real tests.

[DarthShader]

Oh? I have have yet to see a review where AMD's cards don't take a massive hit when AA is enabled. It's an architectural limitation which AMD will likely resolve with upcoming products.

As for the results, I am 100% confident in saying that on average with the latest WHQL drivers, HD 5870 is at least 30% slower at 1920 x 1200 4xAA than the GTX 480 in newer games.

Why pick out 4xAA? What about 8xAA?

http://www.hardwarecanucks.com/forum...review-18.html

10% slower at 25x16, 8xAA at worst. (I realize the sample is small, but I blame the reviewer for not doing more )

Compared to the 30% from 4xAa, is this an "architectural limitation which nVidia will likely resolve with upcoming products." ?

[Heinz68]

I never said which PART of the article was true or false since Fudo contradicted himself more than once as you can see by the highlighted portions above.

One more time, here is exactly what you said.

Actually, this article you refer to will only be proven true or false when trusted benchmark numbers come to light.

The fact is, the article is not based on "AMD's own benchmarketing data" as the FUD claims, therefore the article is false period.

Looks to me you suggested the article might not be false or at least we should wait before coming to such conclusion.

At the same time it also looked like you were defending the FUD

[Zoran]

...and even slower if the driver quality output setting, is the same as nVidia. (No cheats)

Please, go look for some reviews where they tested both default AND HQ settings and you'll see that in MOST of the cases there is NO USABLE performance difference. I personally have two 6870s in CF and I shall yet see a game where that setting makes a difference DURING GAMEPLAY, not while standing still and analyzing pixel per pixel (not too many of those either!). I absolutely am on AMD's side with this setting as it matters more to me whether I am able to achieve playable framerates than a beautiful slideshow.

With the spare horsepower I have everything on the max IQ side with VSYNC on and if a game is not playable I lower the IQ as would anyone else who knows how to.

The Casual user is keen on being able to play the game and not too keen on knowing what they have to lower(adjust) to make it playable I believe.

Regarding the speed of Cayman XT - I am absolutely sure that AMD would not bother releasing a bigger (read: more expensive to manufacture) chip if it was only competitive to the GTX570.

Taking the "absolute" average performance differences (calculated from a whole bunch of sources - be they biased to one side or the other) there is about 10% difference between 6870 and the 5870; about an additional difference of 13% between 5870 and GTX570; and an additional 18% between the 570 and the 580 at 19x12 4xAA 16xAF so, the difference between the 6870 and the 580 is roughly 47%.

If the roumor of 480SP (1920/4) is correct and we compare it to the 6870's 224SP (1120/5) (because it has almost all architectural improvements [compared to the 5870!] already built in and possibly a very similar clock as well), and we consider the "VLIW4 has a similar performance as VLIW5" to be 15% lower performance on average then we get 480/1.15=417; 417/224=1.86 ~86% higher theoretical power. We do not know for certain, but basing on previous inter-architectural ratios everything else (except the memory interface!) will also be "doubled" compared to 6870. so I expect a similar situation as we've seen with the 5770/5870 (roughly 60% performance difference).

Thus, I expect Cayman XT to be 60%*86% ~ 51% faster than the 6870 that would mean ~5% faster on average than the GTX580 and we could call that "trading blows" as well as the different architectures behave differently in different games.

I may still be wrong though

[Solus Corvus]

real answers will comme very soon and some of us will be athorized to release the famous :

"i told you so"

we're still not sure about what it is "they told us so"

my guess is :

*rolls a 100 dice"

50% faster then 5870

Hey! You can't guess that. That's MY wild ass guess. This is like the price is right. You have to guess 49% or 51% or something.

[Zoran]

51% is mine

[Aten-Ra]

I am amazed that people still don’t know how to read the Reviews or they read them the way they want.