SLI AMD 480 performance.worth than 1080?

That has never ever worked that way with any kind of hardware. He didn't even imply that it would be 2x the 1080, he simply said that the get single 1080 performance with bad CFX scaling, which is a different thing.

If the card is indeed close to a 390x, then it should have been faster than that, so that's the reason he said that I guess.

 

Showing (and Raja speaking about it) only the CPU bond value (worst case) as the "GPU usage" was misleading.

This can only lead to the asumption that "GPU usage" means....AVERAGE GPU USAGE as it's the case for 1080.

WIthout labeling properly "GPU usage" as CPU bond and the other two values:less bond , no CPU bond AND without the only really important KEY value: average GPU usage we didn't had correct data.

Is AMD fault to don't expose clearly the results from their own bench in their paper launch.

When all the values and labels are provided it finally make sense.

AMD GPU is ok for the price (in fact very good) PR is a fail...again.

 

Interpreting the Reddit link:

https://forum.beyond3d.com/posts/1918704/

 

Because the OP made a mistake.

Makes sense to me. Plus from what I have said in order for this to be remotely true if game or app supports CF X and the driver has a CF X profile for said game. I do agree sorry AMD nice try better luck next time.

 

Why would normal batch be 51%?

I just tested crazy 1080p 8xmsaa, but i forgot to do my usual 1490/8000 oc for ashes comparisons.

This is at 1480/7800


So basically two 480s are slightly faster than a single 980Ti in this game ? This stinks

 

the test was at 1440p, you did 1080p

 

are you sure ? im certain i saw 1080p 8xmsaa? thats the only reason i tested 1080p

nvm, i checked, it is 1080p

1080 1080p results start at around 60

let me tell you what amd did

they tested the 1080 with the temp throttling or something

 

its in the reddit thread.

1440p VSR on a 1080p monitor.

are we talking abou the same thing?

Robert Hallock clearly said the xfire 480 test was 1440p VSR on a 1080p monitor.

[–]AMD_RobertTechnical Marketing 123 points 14 hours ago
1080p monitor running at 1440p with VSR.

 

Well that makes no sense buddy, cause the 1080 gets 60fps at 1080p crazy settings

So either

A) AMD_RobertTechnical is lying
B) AMD compared 480 CF 1440p results to 1080 in 1080p for some reason

 

nm, he was talking about the doom demo.

My bad...

reading comprehension issue on my part....

 

Well, well, well.

So it turns out the crossfire scaling is 83%

Meaning the 62.7fps the 480 CF setup got at 1080p Crazy was 183% the score of a single 480

This means a single 480 does 34 fps

A single 390x does ~ 38 fps, AT 1440p

Seems like "fury level performance" will be the result of geometry improvements, because in terms of compute this is performing a lot worse than a 390x

 

deleted

 

It's performing a whole hell of a lot worse than a 390x than 5.8tflops vs 5.9tflops would indicate...

34 fps(RX 480) at 1080p vs 38fps(R9 390X) at 1440p

At 1080p a 390x does around 45 fps...

1.7% lower compute throughput, 25% lower performance ?

seriously ? am i going crazy, or has everyone else gone crazy ?

 

honestly, it seems pretty pointless to try to extrapolate performance of a single card from xfire.

we will just have to wait for some more leaks or the 29th.

 

Wow.

Really ?

Just because it's not favorable it's pointless?

AMD themselves claimed 1.83x performance of a single card.

 

https://www.reddit.com/r/Amd/comments/4m692q/concerning_the_aots_image_quality_controversy/

Robert says they used 368.19, which doesn't support the 1080 lol.

He probably just looked up the latest WHQL driver and just assumed they used that.

 

You are right, that was misleading and bad.

That also depends on the driver, it seems to be the most radical departure from the original design to date. We'll see. I don't say you're wrong, but even if you're not the card is still good value and I want to see how it performs with different workloads. How much is a single 390x doing in a similar setting at 1080p, on the same system? Do we actually have a number on that? Also, did they mean 83% CFX scaling, or 83% average GPU usage? These are two different things.

The same guy on Twitter that was claiming that a 7970 didn't have a "scaler" for VSR while the 280x did, that it magically got after a PR ****storm. He's a PR person, they are not to be taken seriously.

 

We should wait the release of 480 and when guru3d detail benchmark this.Then we know the whole story.

 

Driver not being mature is most sensible answer, but why would amd sole and exclusive performance demo be in a game with bad drivers.

Robert hallock was super clear that its 1.83x perf scaling, not gpu utilization. Nobody ever mentioned gpu utilization except the stupid Deceptive 51% in the slides

I found a '390 series' benchmark on aots website, version 1.2.x of the game

5930k

40fps

 

AMD statement on benchmark