• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

6970 3dmark11 score

if the shader count is indeed 1536 then I wont be surprised if it is a little slower then gtx 580. Also I was expecting it to be 1920.
 
And there motivation behind that would be what again? Lose potentially hundreds of thousands in revenue from those customers waiting on the results? AMD delayed there card for a reason, to scramble and do what they can to get a few more frames out of it. You will see next week.

So you're saying hundreds, perhaps thousands, of people are going to run out and buy a GTX 580 or 570 because they can't wait a week for results? That might actually be true.

Or because they really are running into issues stockpiling enough for launch

Obviously, AMD was devastated by the 15-20% increase of the GTX 580, and decided not to reveal information out of fear...........;)
 
And there motivation behind that would be what again? Lose potentially hundreds of thousands in revenue from those customers waiting on the results? AMD delayed there card for a reason, to scramble and do what they can to get a few more frames out of it. You will see next week.

If the 6970 ends up faster those who were impatient enough will scramble to sell off their 580 anyway.
 
The BS about amd got surprised by nvidia is nothing but bunch of crap. I am more ready to believe that they might have driver issues. 580 was just what it was suppose to be, I mean it didn't blow any expectations away. My GTX 480 is with in 10% of gtx 580 at the same clocks. GTX 580 was the right thing for nvidia to do in the short term. I also think that 570 is the best buy for the money right now.
 
After waiting and the delay, even if these cards suck I'm still picking one up to make myself feel better, and as a personal Christmas gift.
 
Those slides are so fake it's not even funny. Besides the 6970 is a 1920 shader part. The 6950 is the 1500+ shader part. So far just a whole lotta FUD.

After waiting and the delay, even if these cards suck I'm still picking one up to make myself feel better, and as a personal Christmas gift.

If they DO SUCK, then there's no reason not to go with Nvidia this round. But I wouldn't believe the FUD until we get some real testing. [H] and sites like Guru3d and Tom's I'd trust for benchmarks but not these bogus slides.

I henceforth claim the 6970 shall be 20% slower than the 570.
Quell it if you dare, sir... If. you. dare.

I hope you're joking. That would make it identical to the 6870. ^_^
The 570/480 is roughly 20% faster on average than the 6870 in [H] testing.
 
Last edited:
My guesses:

AMD would be stupid to have included 2GB as the default memory size on a slower GPU. If rumors are true, then the 6950 also has 2GB of memory - which would be idiotic if it were slower than, say, a 5870.

Thus, if the 6950 does indeed have 2GB, then I expect it to be faster than the 5870, and comparable to the GTX 570 (perhaps slightly slower... where the hell is the GTX 560? An overclocked 560 would be its competition).

I'm expecting the 6970 to be at least comparable to the GTX 580 (I'm still expecting the 580 to be faster and have more OC'ing headroom).

I hope I'm right... I wish they would follow nvidia's lead with something like that Vapor-chamber heatsink... that's awesome!
 
Last edited:
Those slides are so fake it's not even funny. Besides the 6970 is a 1920 shader part. The 6950 is the 1500+ shader part. So far just a whole lotta FUD.

The latest talk over at Beyond3D is that the 6970 will actually be a 1536 part, not the 1920 earlier assumed. Guess we'll know in a week.
 
The latest talk over at Beyond3D is that the 6970 will actually be a 1536 part, not the 1920 earlier assumed. Guess we'll know in a week.

Yes, if it's only 15% bigger than Cypress, then 1536SPs makes more sense.

Well, next week should be fun!
;)
 
Yes, if it's only 15% bigger than Cypress, then 1536SPs makes more sense.

Well, next week should be fun!
;)

The size #'s dont exactly line up to 15% either though... because SIMD's arent the full chip, and a 4-VLIW supposedly takes 10% less room than 5-VLIW, meaning that (24/20)/1.1 * 334 = 364mm^2 is if the entire chip were SIMDs, which it isnt. Obviously we have more TMUs too, and changes to the rest of the core, but I don't think that'd be enough to make up 8% of the chip. Then again, 50% more SIMDs would be too small to fit in just 15% more so there's more to the card than we know :eek:
 
The size #'s dont exactly line up to 15% either though... because SIMD's arent the full chip, and a 4-VLIW supposedly takes 10% less room than 5-VLIW, meaning that (24/20)/1.1 * 334 = 364mm^2 is if the entire chip were SIMDs, which it isnt. Obviously we have more TMUs too, and changes to the rest of the core, but I don't think that'd be enough to make up 8% of the chip. Then again, 50% more SIMDs would be too small to fit in just 15% more so there's more to the card than we know :eek:

Then again....
http://forum.beyond3d.com/showpost.php?p=1501803&postcount=5980
So, 24 SIMDs = 1536 shaders

But I agree, for 1920SPs it would probably need more than +15% die area....

Oh, well, time will tell!
:)
 
Last edited:
Yes, I saw that
- who is that Chispy guy?

Supposedly some reliable source on Guru3d.....I can't take his word, since I know nothing about him; and I'd find ludicrous if the 6970 is only 1-6% better than the 480. AMD wouldn't do anything that stupid.

Anyway, here's some more craziness.

WTF is a slow board?
 
Supposedly some reliable source on Guru3d.....I can't take his word, since I know nothing about him; and I'd find ludicrous if the 6970 is only 1-6% better than the 480. AMD wouldn't do anything that stupid.

Anyway, here's some more craziness.

WTF is a slow board?

Wrong color pcb, and ATI logo not AMD Logo?

Here is another slide of interest
- could be the one where Fudo got his HD6950 in a 'class of it's own' quote from

http://forums.guru3d.com/showpost.php?p=3804614&postcount=33

- which is a bit strange at the GTX470 sits there, which is soon to be replaced with the GTX560
 
Supposedly some reliable source on Guru3d.....I can't take his word, since I know nothing about him; and I'd find ludicrous if the 6970 is only 1-6% better than the 480. AMD wouldn't do anything that stupid.

Anyway, here's some more craziness.

WTF is a slow board?

He also said this earlier...

What can be seen cannot be unseen , guys Cayman HD6970-XT is ~30% faster than HD5870 and ~20% slower than 580GTX , those are real numbers and saw with my own eyes the benchmarks and clocks and the whole thing , and sorry i cannot tell you more , just believe what im saying its nothing but the truth and i have no need to lie ,im dissapointed in what i just saw a few minutes ago. Well 580GTX here i come :/

Isn't the 580GTX about 30% faster than the 5870? How can Cayman be 30% faster than a 5870 but 20% slower than a 580GTX?

And if hes going by benchmarks, I wouldn't be surprised if Cayman's numbers in synthetic benchmarks aren't that great, considering the change in architecture...
 
Came across this on the guru3d forums.

Makes no sense - going by OCUK's pricing the 6970 costs £60 more than GTX570. And that performance threshold -- 30% over Cypress would put it roughly on par with a 570.

Unless AMD is abandoning it's usual price/performance approach I don't buy it.
 
Makes no sense - going by OCUK's pricing the 6970 costs £60 more than GTX570. And that performance threshold -- 30% over Cypress would put it roughly on par with a 570.

Unless AMD is abandoning it's usual price/performance approach I don't buy it.

I don't either, and I wouldn't surprised if he was spewing more fud.
 
He also said this earlier...



Isn't the 580GTX about 30% faster than the 5870? How can Cayman be 30% faster than a 5870 but 20% slower than a 580GTX?

And if hes going by benchmarks, I wouldn't be surprised if Cayman's numbers in synthetic benchmarks aren't that great, considering the change in architecture...

yes that doesn't make sense. 30-35% faster than 5870 2gb would be a GTX 580 so if it's 30 % faster than 5870 2gb than it'll trade blows with a GTX 580 not come in 20% slower. Unless they are measuring their benchmarks at 2560x1600 with 8xAA 16AF HQ using a 5870 1GB vs a GTX 580 then that'll start making it look real but I call fud from this guy.
 
He also said this earlier...



Isn't the 580GTX about 30% faster than the 5870? How can Cayman be 30% faster than a 5870 but 20% slower than a 580GTX?

And if hes going by benchmarks, I wouldn't be surprised if Cayman's numbers in synthetic benchmarks aren't that great, considering the change in architecture...

its obviously he is entirely BSing...

Fanboys are obvious..... :p
 
All I need to know is the final MRSP... that'll tell me how it performs in relation to the 580 which it is supposed to be competing against, unless AMD is intending for the 6990 to play that role.

Surely the 6950 and 6970 will arrive at least 20% faster than the 5850 and 5870, respectively.

Guess we'll know for sure in a few days! Damn all of this waiting, it's almost cruel now that Nvidia has already shown its cards in the 570/580.
 
Must say I'm becoming skeptical. I was expecting that the 6970 would be slightly faster than the 580, but now it looks like more and more "sources" are claiming it'll be, at best, slightly slower. And that's backed up by the likely rumor of 1536SPs instead of 1920SPs.

What happened to that whole "69xx is 30-50% faster than 68xx" thing that one of the board makers stated?
 
Cayman XT 6970: 1536 SPs
Cayman Pro / 6950: 1408SPs
The 2 cards are keeping a very good price - performance :)
 
Last edited:
http://www.fudzilla.com/games/item/21128-radeon-hd-6970-is-10-20-faster-than-gtx-480

Surprise, surprise, guess who was wrong!

Update:
It appears that our usually well informed source failed us, and forwarded a photoshopped roadmap. So, we plainly got served. We are in the process of gathering more reliable info on Cayman performance and we will write an update as soon as we get something worthwhile. It will clearly end up quite a bit faster and we apologize for the inconvenience.
 
This is ridiculous.
The only people enjoying all this crap are the guys at ATI, who are no doubt having a hearty laugh at all the fake info.
 
Looking forward to the flood of 5970's hitting the market for cheap after these hit haha
 
I don't get why they won't give out any info what so ever till release. I got a 570 in my Newegg cart and I'm about to pull the trigger.
 
I don't get why they won't give out any info what so ever till release. I got a 570 in my Newegg cart and I'm about to pull the trigger.
I don't know why, either.
Maybe so they can make some quick changes on the fly, and no one will know about it. If they promise X amount of shaders, and have to decrease it for some reason, they won't have a mob of gamers angry at them.
 
So according to that latest Faud article it'll now end up 10-20% faster than the GTX480 and thus on average losing slightly to the GTX580. Meaning it's pretty much the 4890-GTX285 and 5870-GTX480 all over again?
 
4890-GTX285
Except the 4890 was meant to compete with the 275, not the 285.
5870 vs 480 also isn't a good comparison, either. If the 6970 is between 10% and 20% faster than the 480 (assume 15% average) then it's a lot closer to the 580 than the 5870 was to the 480.

I... don't think there's any valid comparison for single GPU's.
Not in recent years, anyway. ATI has never been that close to Nvidia.
 
Back
Top