Adult 6970 discussion Thread

That has to be fake, no way they would compare a brand new next generation VideoCard to something like 2 generations ago.

That would be like GM saying, hey look at our brand new 2013 C7 Corvette, it beats the fuck out of the 1985 Ford Mustang :rolleyes:

well the 85 did have a 302 and the new Mustang also has a 302 (with a handful of mods)


Apples to apples :p
 
If the 6970 is only 10% faster then the 480, does that mean the 6950 will be slower then the 5870?
 
If the 6970 is only 10% faster then the 480, does that mean the 6950 will be slower then the 5870?

Not really possible to say just yet. But if the leaked numbers are real, then the HD 6950 should be around HD 5870's performance, probably a bit faster. Obviously it will always depend on the game tested. The architectural improvements may pay off in some games.
 
Not really possible to say just yet. But if the leaked numbers are real, then the HD 6950 should be around HD 5870's performance, probably a bit faster. Obviously it will always depend on the game tested. The architectural improvements may pay off in some games.

if that holds true then wouldnt the 6950 only be marginally faster than the 6870? that seems to be a stupid place to put a new card.
 
if that holds true then wouldnt the 6950 only be marginally faster than the 6870? that seems to be a stupid place to put a new card.

Yes that's true. Which is why the performance numbers we are seeing are baffling. Why is the 6900 series performance so underwhelming compared to the 6800 series?

Remember the 6870 is supposed to be displacing the 5770 but for marketing reasons and the big leap forward in performance, AMD decided to upgrade its model number. The 6870 comes very close to beating a 5870 and actually does beat it in CrossfireX mode because of superior scaling.

The 5870 was roughly double the 5770 in stats and performance. The 6970, according to the leaks, is offering nowhere near that kind of performance boost even with the new shader architecture or perhaps because of it.
 
It isn't like they are 50/50. It is like 90% for obviously slower than 580, and maybe 10% it might have a chance to catch it if you believe in unicorns.

All the leaks that look like official ATI slide deck, support it being a fair bit slower than 580.

All the leaks show it being 256 bus, Nearly same Z-Stencil/Rops as Cypress. None of this indicates a big boost.

About the only dispute among leaks is the SP counts. This alone wouldn't be enough to indicate a 580 beating monster given the Cypress level components of the rest of the chip.

So no the leaks don't support both.

Except most of those leaks, say ATI confidential and not AMD confidential? :) and most of those have already been debunked as copied 6870 slides with the performance graphs altered.
 
Last edited:
Yes that's true. Which is why the performance numbers we are seeing are baffling. Why is the 6900 series performance so underwhelming compared to the 6800 series?

Remember the 6870 is supposed to be displacing the 5770 but for marketing reasons and the big leap forward in performance, AMD decided to upgrade its model number. The 6870 comes very close to beating a 5870 and actually does beat it in CrossfireX mode because of superior scaling.

The 5870 was roughly double the 5770 in stats and performance. The 6970, according to the leaks, is offering nowhere near that kind of performance boost even with the new shader architecture or perhaps because of it.

That's why I've been thinking that they must have had to make some kind of compromise. More tessellation performance at the cost of some of the regular performance, possibly.
 
The 5870 was roughly double the 5770 in stats and performance. The 6970, according to the leaks, is offering nowhere near that kind of performance boost even with the new shader architecture or perhaps because of it.


5870 is double the 5770. Double the transistors, double the memory bus width, double the shaders... It is essentially two 5770 chips in one piece of silicon.

It was always quite clear that 6970 would not be double the 6870. So the top lower version 6870 would be much closer to cut down upper version 6950 that in 5850 was to 5770.

If you remember there was a lot of criticism that the gap between 5770 and 5850 was actually too big. Now it will be too small. They just can't win. :D

But otherwise it does look somewhat like 4 Unit shader cluster is not all it is cracked up to be. Because if the leaks are correct total shader count went down but transistor usage went up dramatically 2.1-2.6. It really looks like they could have easily built 1920 old style shaders with that budget and they would have performed a fair bit better, considering what the 6870 is doing with only 1120 of them.
 
Ah yes! Gibbo himself! The hardly ever wrong guy that claimed the HD 2900 XT was faster than the 8800 GTX, while consuming less power and producing less heat! :D We'll see who's right very soon, but Gibbo is not really trustworthy.

So did Gibbo claim that 2900xt was better than the 8800GTX while testing both cards?

I don't know what his claims were in the past, but he works for OCUK and they have the cards to sell and has been testing them on his own...
 
What version drivers are they using?

I'm reading either some people or Gibbo himself calling them "old drivers" because they don't officially support 6900. With the new shader architecture and some other improvements, driver support of those features really are necessary to ballpark the real performance.
 
I just saw where AMD has the 10.12 out now on there web site so maybe one of these people ( that say that have these cards) can get them.
 
I just saw where AMD has the 10.12 out now on there web site so maybe one of these people ( that say that have these cards) can get them.
The interesting part is the 10.12s don't list the 6900 series as supported. Maybe we are waiting for a 10.12a to drop with support?
 
The interesting part is the 10.12s don't list the 6900 series as supported. Maybe we are waiting for a 10.12a to drop with support?
I'm pretty sure that 10.12 will support the 6900 series. This came up on a thread on Guru3D from the release notes which were up on AMD's site earlier but have unfortunately been taken down for maintenance but 6900 was mentioned in the first few paragraphs.

Highlights of the AMD Catalyst™ 10.12 Windows release includes:

New Features:
Support for DivX on the AMD Radeon 6800 Series
With DivX® software installed on an AMD Radeon™ HD 6800 Series and later GPU and the DivX Accelerated™ solution, consumers can enjoy an improved DivX video playback experience by lowering CPU load, extending computer battery life on notebooks, reducing noise on home theater PCs and improving overall computer operation.
Grab the DivX Accelerated download from here (will be available Dec 15) http://go.divx.com/accelerated
Support for OpenGL 4.1
New features introduced in OpenGL 4.1
Full compatibility with OpenGL ES 2.0 APIs for easier porting between mobile and desktop platforms
The ability to query and load a binary for shader program objects to save re-compilation time
The capability to bind programs individually to programmable stages for programming flexibility
64-bit floating-point component vertex shader inputs for higher geometric precision
Multiple viewports for a rendering surface for increased rendering flexibility
New ARB extensions introduced with OpenGL 4.1
Linking OpenGL sync objects to OpenCL event objects for enhanced OpenCL interoperability
The ability to set stencil values in a fragment shader for enhanced rendering flexibility
Features to improve robustness, for example when running WebGL applications
Callback mechanisms to receive enhanced errors and warning messages
AMD Stream 2.3 SDK release
Performance improvements for AMD’s OpenCL toolset
Support UVD video hardware component through OpenCL driver (Windows 7 only)
Support for the Stream Profiler on Linux (command line version)
Support added for AMD Radeon HD 6800 Series and AMD Radeon HD 6900 Series
Support for FFT and BLAS-3 libraries
Stream Profiler enhancements (including timeline visualization)
Resolves a number of bug fixes from the Stream SDK 2.2 release

http://forums.guru3d.com/showthread.php?p=3808892
 
I'm pretty sure that 10.12 will support the 6900 series. This came up on a thread on Guru3D from the release notes which were up on AMD's site earlier but have unfortunately been taken down for maintenance but 6900 was mentioned in the first few paragraphs.



http://forums.guru3d.com/showthread.php?p=3808892
It only lists the 6900 series under the SDK. Look at the product compatibility on the actual release notes. Only 6800: http://support.amd.com/us/kbarticles/Pages/AMDCatalyst1012ReleaseNotes.aspx
 
I tried to read this thread..i really did. But i just couldnt.

Grr...

Hey lets just wait for he next 9800Pro...it should be back around and released again pretty soon. :eek:

Now there is an upgrade cycle...from one 9800 to the next.....the "8 year plan"
 
I tried to read this thread..i really did. But i just couldnt.

Grr...

Hey lets just wait for he next 9800Pro...it should be back around and released again pretty soon. :eek:

Now there is an upgrade cycle...from one 9800 to the next.....the "8 year plan"

egh..... I'll just buy every damn generation like I do already :cool:
 
Looks like 6900 drivers were omitted from 10.12? Only support for up to 6800s listed, though that could just be on paper.
 
So are we continuing on the idea that the performance leaks are meaningless as they are using 6800 series drivers and this is all meant to be a "trick" to make the people leaking info look stupid when they come out wednesday mush faster than whats being seen?
 
Looks like 6900 drivers were omitted from 10.12? Only support for up to 6800s listed, though that could just be on paper.

Hopefully it's just on paper. I honestly believe your post has depressed me. I was hoping 10.12 would bring support and a huge boost for Cayman but it appears to me now that the driver being used is the driver in the retail box that the german guys were using. I'm beginning to cope with the idea that it will not beat a GTX 580. The only thing left for AMD to screw up would be the pricing. If they release a part that trades blows and mostly loses with the 570 for the rumored $450 then it's a clunker launch if the performance is what we've been seeing for the last few days. To pull this off they can launch the card around GTX 570 price of $349.99 or around that area.
 
To pull this off they can launch the card around GTX 570 price of $349.99 or around that area.
If the 6970 trades blows with the 570 they have bigger problems than price.
For example, a 6950 that's trading blows with their existing 6870. Whoops.
 
Hmm...that should put the 5870 down to a solid $200. I still cant decide if i should get a second 5870 and deal with x fire headaches as 570 performance would DEFINATELY not be worth $350 investment over my current 5870 i am not paying much above $350 for any video card period. (unless it shows at least a 50% or more performance improvment over current). As it stands 2 5870s are faster than a 580 as long as you keep the resolutions down (which i do) and play well supported games (which I do).
 
I just checked the .inf files for the 10.12 drivers. There is no mention of the 6900 series. I suspect the drivers for the card will show up with the reviews and be different from the ones in the retail box.
 
I wouldn't despair quite yet, AMD spent a whole year on the 5870-6970 transition. It would defy explanation if they can't manage at least a 40% performance boost with a 33% watt increase, so a 5% performance/watt increase in total to be on par with a 580.
 
If the 6970 trades blows with the 570 they have bigger problems than price.
For example, a 6950 that's trading blows with their existing 6870. Whoops.

Good point I didn't even think of that, perhaps the 6850 would perform just in between but slightly above a 6870, this is the thing that keeps me thinking there is no way these things can be true.

I suspect the drivers for the card will show up with the reviews and be different from the ones in the retail box.

I hope your right
 
I hope your right
The interesting part is often before launch Kyle or Brent will drop some comment about performance or new cards before the NDA breaks. Neither have even posted if they have the cards. Along with as much disinformation and conflicting reports on the cards, I think AMD is playing their cards very close to their chest this launch. Most likely not wanting to give time for Nvidia to adjust pricing right before launch like they did with Barts.
 
The interesting part is often before launch Kyle or Brent will drop some comment about performance or new cards before the NDA breaks. Neither have even posted if they have the cards. Along with as much disinformation and conflicting reports on the cards, I think AMD is playing their cards very close to their chest this launch. Most likely not wanting to give time for Nvidia to adjust pricing right before launch like they did with Barts.

Funny you mention that I thought about that before. From my recent memory they'll usually come into a thread when everything is skewed completely and drop a hint via a quick comment that gives everyone a hint that they're barking up the wrong tree. They never really violate NDA and I'm sure that's not what you meant either. That is the part that bothers me about this too, if the card did indeed perform amazingly they would have chimed in by now and give that hint that we may be barking up the wrong tree. That's the other part that's bothering me. It leads me to believe that they are not barking up the wrong tree, but we're pretty dead on with the info.
 
Funny you mention that I thought about that before. From my recent memory they'll usually come into a thread when everything is skewed completely and drop a hint via a quick comment that gives everyone a hint that they're barking up the wrong tree. They never really violate NDA and I'm sure that's not what you meant either. That is the part that bothers me about this too, if the card did indeed perform amazingly they would have chimed in by now and give that hint that we may be barking up the wrong tree. That's the other part that's bothering me. It leads me to believe that they are not barking up the wrong tree, but we're pretty dead on with the info.

The benchmarks from people with the card have been conflicting though. In some it barely seems an increase over the 5870, while in others it's soundly beating the GTX570 by so much that you'd think it'd even beat the GTX580 or at least equal it.
 
That doesn't make sense since 6950's are cut-down 6970s. You can't cut down 6950's cause then you're just wasting cut-down chips.

He's hinting that the 6970's are yielding just fine, hence there aren't enough 6950's to go around
 
Back
Top