AMD: Speculation, Rumors, and Discussion (Archive)

[dskneo]

It is now 83% scaling.

The 51% was being referred to the single batch test (cpu-bound). The entire test average was 83%. This puts a single RX480 with 34FPS

not quite the champion I painted it to be some posts above

 

[CarstenS]

Interestingly, when selecting the two most closely matched runs from this morning (gog-version 1.12.19917, 1080p, high, 980 Ti OC w/BIOS-fixed clocks on i7-6700K) and looking at the presented "weighted framerates" for different batch groups, I'm seeing:
run1/run2
Normal 87,69/87.80 Fps
Medium 80,75/78,30 Fps
Heavy 65,70/64,96 Fps

Now, with the more recent version from Steam (1.12.19928), Fps are down.
run1/run2
Normal 83,82/83,71 Fps
Medium 73,86/75,33 Fps
Heavy 62,33/61,05 Fps

I'm using an older driver on that system though. Might be all well with a more optimized version that melts down the snow more quickly.

--

So does this only happen on GTX 1080 or on all Nvidia cards?
I do know that GTX 1080 is having some problems with some games. Many people still can't get Warframe to run on a GTX 1080. Digital Extremes have been working on it trying to get those users to be able to play the game.

I wouldn't know - my results in here are not even with a 1080, but an overclocked 980 Ti.

It is now 83% scaling.
The 51% was being referred to the single batch test (cpu-bound). The entire test average was 83%. This puts a single RX480 with 34FPS
not quite the champion I painted it to be some posts above

Where do you get the 83% scaling from?

 

[dskneo]

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

second post bellow explains it

 

[SimBy]

It is now 83% scaling.

The 51% was being referred to the single batch test (cpu-bound). The entire test average was 83%. This puts a single RX480 with 34FPS

not quite the champion I painted it to be some posts above

You can't pick and choose like that. Its 51% scaling for the FPS comparison they used.

 

[Silent_Buddha]

I wouldn't know - my results in here are not even with a 1080, but an overclocked 980 Ti.

Hopefully just the 1080, I'd hate to think that Nvidia have found another creative way, in another game, to "enhance" their performance. And hopefully it really is just a bug. Brings back memories of Nvidia using shader replacement in order to boost performance at the cost of graphics IQ.

Regards,
SB

 

[Otto Dafe]

You can't pick and choose like that. Its 51% scaling for the FPS comparison they used.

No it's 83% for the FPS comparison they used. Check the post immediately above yours.

 

[SimBy]

No it's 83% for the FPS comparison they used. Check the post immediately above yours.

I know what Robert wrote. They used single (normal) batch FPS which had 51% scaling. 1.83 is average for all batches..

 

[Love_In_Rio]

I know what Robert wrote. They used single (normal) batch FPS which had 51% scaling. 1.83 is average for all batches..

An so the fps results tell that a 1080 is 1,75 times faster than a 480 overall.Anyway IIRC a 1080 in this bench performs similar to a pair of 980 GTX in SLI.So a 480 should perform roughly similar to a 980 in AoS,a game that seems to favor AMD architecture.

 

[Otto Dafe]

I know what Robert wrote. They used single (normal) batch FPS which had 51% scaling. 1.83 is average for all batches..

FPS is average for all batches.

 

[Love_In_Rio]

FPS is average for all batches.

And so is the 1,83 scaling. So the power point was misleading.

 

[dskneo]

which makes the single RX480 framerate 34, if I am not mistaken

 

[SimBy]

FPS is average for all batches.

Where did Robert confirm FPS shown is overall and not single batch?

 

[Love_In_Rio]

Where did Robert confirm FPS shown is overall and not single batch?

Its not fully clear, but would be quite strange to publish a by nature cpu bound bench.

 

[SimBy]

Its not fully clear, but would be quite strange to publish a by nature cpu bound bench.

I'm sure AMD intentionally made it very unclear. But if they actually shown average all batches FPS and used single batch scaling the whole thing just got a whole lot stranger.

That would put RX480 around 290 level. I think 290 gets around 35 FPS on crazy 1080p.

 

[Otto Dafe]

Where did Robert confirm FPS shown is overall and not single batch?

I interpreted the reddit post we're all referring to as that confirmation, and I didn't understand what you were saying. Now I think you're probably right, and it's some pretty lame cherry-picking by AMD.

Edit: Did they even reveal the framerate for the run that wasn't, in their own words, "cpu-bound" anywhere? I guess they didn't think that would be interesting to us?

 

[dskneo]

I think we all know what this means. The card is not a revolution, but an ordinary progression of price/perf.

We just have to wait for the 29th for confirmation.

 

[xEx]

Talking about the power consumption could it be higher because of the use of normal gddr and cheap(low efficient) electronic components?

 

[Entropy]

Talking about the power consumption could it be higher because of the use of normal gddr and cheap(low efficient) electronic components?

Higher than what?
It uses the same GDDR5 as the GTX1070.
We just don't know the power draw under various conditions until actually tested. Simply have to wait for numbers until we have decent reviews.

 

[SimBy]

Did a quick AotS run with 290@1050MHz, obviously not the same version AMD used. Exactly 35 fps average all batches on crazy 1080p.

 

[silent_guy]

With AMD focusing its launch completely on VR, were there any new VR related features announced?