AMD: R9xx Speculation

[NathansFortune]

One would assume that the new build has accurate 6900 series information in it. The embargo is about to come to an end so information has to be put out there now.

 

[Robert Varga]

Sadly?

For me, it's a big chance for 6950/70 to have really great thermals and TDP.

Thumbs up AMD!

 

[DavidGraham]

GPU-Z is an ARCHIVE!

it can't read nothing from GPU.

I know that , Wizzard just posted the last build which supports HD 6900 cards , ie it contains the right technical specifications , otherwise it wouldn't support them .

Also the program isn't completely composed of lookup tables , it can read GPU and memory clocks in real time .

 

[RobertR1]

I'm very curious to see how well the 6970 performs against the 5870 and GTX580, in actual gamepla scenarios.

Efficiency is great n all as long as the peformance is there and ultimately pushing the performance envelope forwad through competition.

 

[-The_Mask-]

GPU-Z is an ARCHIVE!

it can't read nothing from GPU.

It can read streamprocessors. I remember a HD6850 with 1120SP's and a GTX580 with 512CC.

But it doesn't really matter, because Cayman has 1536 SP's.

 

[Broken Hope]

What I don't understand is, is that 6970 is around GTX 570 speed, and the GTX 570 isn't massively faster than 5870 as far as I know, so where on earth does the 6950 fit? And also why did AMD bother to redo the shaders and a ton of the other work for this GPU if it's not really going to be hugely faster than an overclocked 5870, it just seems like a ton of work for not a massive amount of gain.

 

[Robert Varga]

Maybe 6800/6900 is here just becouse of better efficiency?

Remember 40nm stuff.

 

[Rangers]

What I don't understand is, is that 6970 is around GTX 570 speed, .

How do we know that?

and the GTX 570 isn't massively faster than 5870 as far as I know, so where on earth does the 6950 fit? And also why did AMD bother to redo the shaders and a ton of the other work for this GPU if it's not really going to be hugely faster than an overclocked 5870, it just seems like a ton of work for not a massive amount of gain

You could say the same about 580 vs 480. It's "only" ~15% faster. Not hugely faster than an overclocked 480...and 570 is almost exactly as fast as 480. So the 6950 could be analogous to the 570.

 

[Broken Hope]

How do we know that?

The German guy posting benchmarks and pics of his retail card? Unless that is somehow all faked.

You could say the same about 580 vs 480. It's "only" ~15% faster. Not hugely faster than an overclocked 480...and 570 is almost exactly as fast as 480. So the 6950 could be analogous to the 570.

Except the 6870 is already almost as fast as the 5870 so having a card that is just slightly faster than the 5870 just seems like too many cards occupying the same performance spot.

 

[jimbo75]

What I don't understand is, is that 6970 is around GTX 570 speed, and the GTX 570 isn't massively faster than 5870 as far as I know, so where on earth does the 6950 fit?

Simple. It's in a class of it's own.

 

[Rangers]

The German guy posting benchmarks and pics of his retail card? Unless that is somehow all faked.
.

Well the medium benchmark seems to scale quite linearly with game performance and 4700 for 6970 would put it above 5870, but it was probably with older drivers. I don't think we can draw any conclusions until ATi release their new drivers for the VLIW4 architecture. There may be a significant performance boost to be had.

?

I thought that guys benches had been put into question by posts like the above and others. According to him it's barely faster than 5870 which really makes little sense.

 

[LunchBox]

What I don't understand is, is that 6970 is around GTX 570 speed, and the GTX 570 isn't massively faster than 5870 as far as I know, so where on earth does the 6950 fit? And also why did AMD bother to redo the shaders and a ton of the other work for this GPU if it's not really going to be hugely faster than an overclocked 5870, it just seems like a ton of work for not a massive amount of gain.

Well, the rework done to the shaders can always be seen as a building block of what's to come in the next chip they will churn out. I for one am a little glad since it is looking like I won't have to upgrade this generation as I've already got myself a 5870 when it came out..

 

[Broken Hope]

Well, the rework done to the shaders can always be seen as a building block of what's to come in the next chip they will churn out. I for one am a little glad since it is looking like I won't have to upgrade this generation as I've already got myself a 5870 when it came out..

I got a 5870 at launch also, I was just really looking forward to the 6900 series and if the performance increase is as small as it seems it looks like I'll be skipping a generation also.

 

[PSU-failure]

The German guy posting benchmarks and pics of his retail card? Unless that is somehow all faked.

Except the 6870 is already almost as fast as the 5870 so having a card that is just slightly faster than the 5870 just seems like too many cards occupying the same performance spot.

Let's be serious...

Would you use "Class of its own" to describe something only barely higher than your own lower product? Would you if it were closer to your competitor's 30% faster board?

The performance chart floating around is probably a fake because of this and we should probably replace "HD5970" by "HD6970" and "HD6970" by "HD6950".

As far as I know, the only really useful benchmark asked to Ducatti750 (the German poster), 3DMark's MT fillrate test (so that we could see if the 24 SIMDs are present and working at the intended frequency), doesn't show up.

 

[Unknown Soldier]

So, that makes either the score provided for HD 5870 or the benchmark itself even more un-bound by pure compute. Or do I miss the obvious explanation as to why a HD 5770 should outperform a HD 5870 by more than 50%?

edit:I am just running this ComputeMark 2 on a GTX 580. Seems like the drivers have been holding Fermi back quite a bit. Plus, it doesn't crash on Extreme any more.

It could've also been a driver issue? I think so since the 5870 test was done quite some time back.

 

[CarstenS]

It could've also been a driver issue? I think so since the 5870 test was done quite some time back.

Yes, certainly possible. Right now, only thing I have is my most recent GTX 580 result and the alleged score from the guy who already could buy a HD 6970. I am switching back and forth between three cards this weekend and my trusty HD 5870 is certainly getting it's stab at Computemark as soon as I am done with the rest.

 

[Sinistar]

It'll take some more time apart from what we've done so far. Because other work is piling up VERY fast. It doesn't seem, like I'm not going to touch graphics cards this year at least.

Hmm....

 

[Unknown Soldier]

Kewl

 

[CarstenS]

Hmm....

Work != spare time.

 

[Lightman]

Yes, certainly possible. Right now, only thing I have is my most recent GTX 580 result and the alleged score from the guy who already could buy a HD 6970. I am switching back and forth between three cards this weekend and my trusty HD 5870 is certainly getting it's stab at Computemark as soon as I am done with the rest.

So you're not testing HD69xx for upcoming review on your site?