• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

ATI cuts 6950 allocation

If the 14% faster than 580 is true then it is faster than the 5970, that would be an awesome feat on the same size process, I'll believe it when I see it of course
 
not sure if posted before
a guy on a forum in malaysia ( i think its a reputable malaysian forum ) is selling these on pre order
spec is as

687050.jpg


Via twitter ( another spec )

http://www.hardwareluxx.de/index.ph...md-radeon-hd-6970-alle-technischen-daten.html
 
Last edited:
If true, that is a veryt impressive score and puts it faster than a gtx580. 4.2Ghz i7 get 6000 with a gtx580. That would put the 6970 about 20% faster than a gtx580 in 3dmark11 at least.

I think we have to take into account that graphic card drivers from NVIDIA in relation to 3dmark11 are crap. This is a fact, as they cant get even SLI to work properly.
On the other hand AMD drivers work VERY smoothly on 3dmark, and getting great scaling with 2-3 and even 4 crossfire.
Expect this 15% lead to be reduced when Nfail manage to pull their finger out and get a decent set of drivers out there...
 
AMD is claiming that these 4SP clusters offer "similar performance to" the old 5D clusters, yet take up 10% less die area.
They are theoretically 20% slower, but I believe the problem was that most shaders are not easily optimisable to use all 5 SPs, so AMD cards were essentially wasing die space on a 5th SP that sat idle for much of the time. That is also why there was always a bigger disparity bwetween theorectical TFlops and actual performance with ATI cards than Nvidia.
 
I think we have to take into account that graphic card drivers from NVIDIA in relation to 3dmark11 are crap. This is a fact, as they cant get even SLI to work properly.
On the other hand AMD drivers work VERY smoothly on 3dmark, and getting great scaling with 2-3 and even 4 crossfire.
Expect this 15% lead to be reduced when Nfail manage to pull their finger out and get a decent set of drivers out there...
NVidia drivers SUCK! I'm never buying an NVidia card again! etc. :)
 
I think we have to take into account that graphic card drivers from NVIDIA in relation to 3dmark11 are crap. This is a fact, as they cant get even SLI to work properly.
On the other hand AMD drivers work VERY smoothly on 3dmark, and getting great scaling with 2-3 and even 4 crossfire.
Expect this 15% lead to be reduced when Nfail manage to pull their finger out and get a decent set of drivers out there...

Only the SLI driver bit is broken. My single gtx470 overclocked is beating every single 5870 in the 3dmark11 thread by some margin so I fail to see how you can say Nvidia has poor drivers? 3dmark11 results barring the SLI issue is spot on for both Nvidia and AMD cards in how powerful they are to each other.
 
Only the SLI driver bit is broken. My single gtx470 overclocked is beating every single 5870 in the 3dmark11 thread by some margin so I fail to see how you can say Nvidia has poor drivers? 3dmark11 results barring the SLI issue is spot on for both Nvidia and AMD cards in how powerful they are to each other.

Its beating it because of tessellation, in tests 1-2 your 470 is getting beat by a decent oc 5870. Simples really.
Where 4-series makes ground up and overtakes a 5870 is test 3-4.
 
Its beating it because of tessellation, in tests 1-2 your 470 is getting beat by a decent oc 5870. Simples really.
Where 4-series makes ground up and overtakes a 5870 is test 3-4.

Would you not expect a oc'ed 5870 to beat a 470, i know i would. I do not see what that has to do with the drivers.
 
gibbo, its been asked on another thread
but i will ask here
you have the money in your hand to buy both
what do you buy ?
 
i can see it now
gibbo in his office with door shut, sat on his leather recliner seat, feet on the desk, laughing to himself saying " those kids on the forum, all this guessing, if only they knew,"

superstock1569r9018318.jpg
 
Last edited:
So basically the new architecture has 1536 shaders, but thats the equivalent of 1920 shaders on a 5870 architecture, that correct?

mhh 1536 shaders (4D) = 384 shader units yea maybe they mixed it up something.

you should know better gibbo u have the cards

well if 6970 has only 1536 they might go for low price , but i doubt it we want 1920
 
Last edited:
Are you saying the 6970 has 1536 shaders gibbo?


:)

I am not confirming anything. But I am writing my specs out for the cards and want to try and make sure they are spot on.

We have to make sure punters understand the description. If the card is for example 1536 shaders based on a new architecture, we need to get across to punter that is actually equivalent to a 1920 shader equipped 5870.

Its a bit like when CPU's changed from MHz to PR ratings, so we just need to make sure the none knowledgable customers understand exactly the kind of performance they are buying into.

In order to test as fair as possible I am setting up a rig with 5870 in and will run the last 3D Mark releases on it and then drop in a 6970 and do the same. So then we can get some numbers on how it compares to the card its replacing, the 5870. :)
 
Back
Top Bottom