Some RX 480 4GB stuff leaking out.1080Mhz Clock

Dont mind him, hes ignoring most of us here. Not a huge loss.

 

More hype
posted image

 

I hate to bore you guys with numbers and logic and reasoning, but IF the rumor posted is valid, and a 20% OC brings a 480 on par with a 1070...

It necessarily means that at stock the 480 is 83% (5/6) the performance of a 1070.

120% 480 = 100% 1070 => 100% 480 = (100/120)% 1070 = 83%.

We have two reference values; 3DM, and AotS.

1070; 55fps AotS, 8300 3DM Extreme

480; 34fps AotS, 5700 3DM Extreme

34/55 = 62%

5700/8300 = 68.6%

Either the rumor posted is untrue, or AMD underrepresented the performance of the 480 by 25%, which do you think is more likely ?

Where's the 3DM score ?

 

considering the lack of any hype whatsoever from AMD, its not a stretch to believe they are underplaying their hand to keep nvidia from reacting.

On the other hand, i find it hard to believe that nvidia hasnt got their grubby mitts on a sample.

 

You don't think it's a stretch that AMD intentionally lowered the performance numbers of the 480 in their own announcement by 25% ?

 

I just posted the pic here,It could be a GTX1080 for all I know.

 

they haven't really announced any performance numbers.

its all been derived.

 

That looks like a screenshot of a video, did you not take the screenshot?

 

So this isn't Raja Koduri? Or is he not an AMD employee ? Or was this presentation staged, just like the moon landing ?

This. This is a stretch.

It would need a 58% overclock compared to stock (and linear performance scaling wrt clock) to hit 1070 numbers.

58%.

Someone. Is. Lying.

Either it's AMD, making their card seems worse than it is in reality, or it's whoever started this rumor.

 

C&P from guy streaming video with Overwatch.I did not watch the video but screen shot was supposed to be when he alt tab out of game.Shown firestrike.

 

like i said, derived.

trying to get single card performance from that is silly.

Which according to Hallock, is why the did it: So the website have something to reveal when the NDA lifts.

 

Was going to ask what your source was regarding the pricing but then you answered so I meant to say ignore my post.

 

Wasn't it actually that Hallock misrepresented scaling, and then when everyone complained that AMD numbers make no sense, they came up with exact and detailed numbers for fps/scaling. Aots, wasnt it?

1.83x scaling...?

 

yes, but that doesnt tell us much about usage on each card.

 

1080 gets 58.7fps in AMD labs in Aots

1070 is what 15-25% slower, depending on res.

how come 1070 gets 55fps in Aots?

 

No. It was 1.83x multi-gpu scaling. That means the score of dual GPUs is 183% the score of a single GPU. That means we can unequivocally say a single 480 achieves 34 fps.

 

AMD's score is incorrect for the 1080 actually, or rather it's lower than every other score on the AotS benchmark browser; 1080s start at 62+

I got 59fps at same settings used in the 480 demo.

A 1070 most certainly does manage around 55

 

@The Mac

you think this AMD's slide is another example of AMD downplaying its own performance

that they willfully chose an example of bad 480 CF scaling to compare with 1080?

 

i edited my post.

im not so sure its unequivocal without Knowlege of individual card loading.

either way you are stil deriving single card performance.

maybe its right, may not.

dunno.

im certainly not buying into the hype, just posting what i come across.

 

It cannot be incorrect, the only way they can say mgpu scaling is 1.83x is if they compare the 62.5 fps to the single GPU performance and calculate the ratio.

Unequivocal.