• 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

not long to find out now. I think before we sell the fish as bad before it arrives, why don't we wait?

No point listening to dubious sources, one way or the other :)
 
Has anyone been following this Ducati guy who has been carrying out some of the tests? Did he bother doing any comparisons between available drivers or did he just go headlong into doing every benchmark he possibly could without asking himself why they are hugely surprising?

For all we know he could have forgot to plug in the 6 pin thinking only to put in the 8 pin because its more pins so better.
 
Has anyone been following this Ducati guy who has been carrying out some of the tests? Did he bother doing any comparisons between available drivers or did he just go headlong into doing every benchmark he possibly could without asking himself why they are hugely surprising?

For all we know he could have forgot to plug in the 6 pin thinking only to put in the 8 pin because its more pins so better.

or maybe hes running only on 1x PCIEX??

And there is still chance that this baby will be 50% quicker than 580GTX.

Let the games BEGIN!
 
It's an example of something silly being done, not a prediction on whats wrong.

I'm asking do we have any context on where these benchmarks are coming from or do we just blindly trust results that are hugely surprising and should be taken as done on the professional level which we expect from sites like guru3d, TH, Hardocp, etc.
 
I'm asking do we have any context on where these benchmarks are coming from or do we just blindly trust results that are hugely surprising and should be taken as done on the professional level which we expect from sites like guru3d, TH, Hardocp, etc.

Clearly they are not done "professionally" like the reviews we will see on Wednesday from the big name sites (Anandtech, HardOCP, guru3D).

That said, they're coming from someone who has bought a retail version of the card (verified by box pictures, and serial numbers matching between hardware and software). He has shown how GPU-Z recognises the card, he's posted a number of benchmarks that show a fairly consistent level of performance, and he's compared the results to a GTX480 (his previous card). Performance seems similar to the GTX480 overall, but varies from test to test (as would be expected).

It's entirely possible that he is doing something "strange" with the benchmarking or driver setup, but it seems extremely unlikely that the entire thing is fabricated. He is using the "out of the box" drivers that many review sites will be using. It's possible that the newer 10.12 driver set may add some performance. A couple of other results from different sources have also cropped up, and fall pretty much in line with this guy's results.

Once you accept the limitations of a non-professional benchmark, I think that it's fair to consider the results to be reasonably representative of the performance of the card.
 
Is it just me, or does it just seem like utter madness to be releasing a card that is barely better than a 5870 ? I mean seriously... There is making something efficient, and then just screwing yourself over, just doesn't add up given the performance of the 6870 cards.
 
Clearly they are not done "professionally" like the reviews we will see on Wednesday from the big name sites (Anandtech, HardOCP, guru3D).

That said, they're coming from someone who has bought a retail version of the card (verified by box pictures, and serial numbers matching between hardware and software). He has shown how GPU-Z recognises the card, he's posted a number of benchmarks that show a fairly consistent level of performance, and he's compared the results to a GTX480 (his previous card). Performance seems similar to the GTX480 overall, but varies from test to test (as would be expected).

It's entirely possible that he is doing something "strange" with the benchmarking or driver setup, but it seems extremely unlikely that the entire thing is fabricated. He is using the "out of the box" drivers that many review sites will be using. It's possible that the newer 10.12 driver set may add some performance. A couple of other results from different sources have also cropped up, and fall pretty much in line with this guy's results.

Once you accept the limitations of a non-professional benchmark, I think that it's fair to consider the results to be reasonably representative of the performance of the card.

Ok well i've only seen stuff people have posted here, I assume its a forum he is posting on. Do you have a link?

Interesting you mention his 480, I remember a screenshot of his desktop showing shortcuts for Nvidia 3D vision so i'd be curious if he mentioned how he uninstalled the Nvidia drivers before testing the 6970 or if he bothered to do it at all. Could be Nvidia software (or remnants) recognizing the AMD hardware and forcing incompatibility.

Basically i've had enough of people posting this stuff and absolutely no context included. At first it was just funny but with people taking it seriously without any justification I would like to add some intelligent analysis by reading up how he went about this.
 
Is it just me, or does it just seem like utter madness to be releasing a card that is barely better than a 5870 ? I mean seriously... There is making something efficient, and then just screwing yourself over, just doesn't add up given the performance of the 6870 cards.

It does seem somewhat odd.
 
there's no chance amd will release 300 and 400 pound cards (inc vat) and be much slower than their respective peirs (570 and 580). The price will be according to performance, so unless both caymans have some crazy new "ability/function" I fail to see it not living upto the hype. It will be close or better than the nvidia's.
 
6970 to be a maximum of 10% slower (10% I reckon) than the 580 imho. Antilles will continue AMD's pattern of securing the top spot by a dual gpu card. This time though nvidia will have dual gpu card out in the not too terribly distant future. So I don't see AMD's flagship card reigning for that long, unless that is the crossfire scaling, as seen so far with the 6800's, carries on with the improved scaling.
 
Last edited:
While obviously none of us can know for certain until the cards are released on wednesday and we have legit reviews and pricing info, I'm starting to feel that I made the right call choosing to skip this new generation of cards. We should be seeing 28nm cards this time next year, and I'm sure my 5850 can handle any games until then.
 
I'm still gonna wait see the pro reviews - with the short supply of 580's the 6970 being poo would be v bad news and at that price amd would do better not to release it if them figures are true - not much better than clocked 6870. I find it very hard to believe they shoudl get such a large dose of stupid all of the sudden.
Seems a bit to good for nvidia for them figures to be true but will wait and see.

I am buying one or the other of the 580/6970 and I can wait.
 
Ok well i've only seen stuff people have posted here, I assume its a forum he is posting on. Do you have a link?

Here is a link to the thread he's posting the results on. It's a German site though, so unless you speak German you might want to use a translator of some kind.

It's possible that drivers have been improperly removed. But, results from other sources have also been trickling in, and seem to fall in line with those reported in the above thread. Proves nothing of course, but it does ad a little weight to the results we've seen so far.
 
Crysis 5870 v 6970

This guy had the 5870 in his system before the 6970 so no Nvidia drivers to screw things up.

6970 12% faster in the crysis benchmark over a 5870.




 
Last edited:
Back
Top Bottom