AMD is finished.
>>54850970
> implying amd live
So are AMD fanboys brainless or what?
2xGPU with 57% utilization match a single GTX1080.
57% utilization then the game is pretty much running off 1 card, 1 card should be able to match the GTX1080 then.
So what was the point of using two cards?
And lets not go into the image quality being lower on the AMD side.
>>54851099
>And lets not go into the image quality being lower on the AMD side.
it's literally the other way around
>nvidiots are this stupid
>>54850970
480 destroys all of your offerings in price:performance, all that's left is for the 490X to destroy your single card performance. Face it, this was Fermi 2.0 through and through; temperature wise and performance wise. You guys lost.
>>54851099
>57% utilization then the game is pretty much running off 1 card
Are you literally brain damaged, or is this your first day out of /v/? It is nowhere near that simple.
>And lets not go into the image quality being lower on the AMD side.
The bad image was on Nvidia side, moron. Computex was in Taipei, Taiwan aka China. They read right to left. That's why the comparison was done in reverse to how a left to right reading country would. Did you even watch the stream?
>>54851099
dont forget the graphical output lag with 2 cards
>>54851894
you mean the lag that was only present because the API's had scheduling issues since they saw 2 separate GPU's while AMD has been doing work into EMA in linked mode to ensure the API's only see one unit of pooled resources so the scheduling doesnt cause added latency?
why do you think amd promoted Mantle, it helped give rise to dx12 and vulkan with Explicit multiadapter in linked mode.
Dont be surprised when you start seeing 2+ gpus on a single card again i the next few years
>>54850970
>AMDs 200$ upper midrange card is on par with the 980ti and Fury.
>Nvidia fanboys on suicide watch, trying to justify their product and spawning shill threads like this one.
Topkek
>>54853844
It's fairly amusing watching them flail. They'd still buy a 960 over an RX 480 though, just to support master.
>>54851099
>2xGPU with 57% utilization match a single GTX1080.
My curiosity is why didn't they just run a comparison on ONE card if it is truly only using 51% (or 57%)?
It really fucking is Zen is worse than skylake and I actually think the radeon are ok
>>54853884
>My program uses 10% of my cpu on 10 cores
>I should be able to get the same performance by running it 100% on one core right?
Kill yourself
>>54853917
If anything they should get _better_ performance out of a single card as there is less overhead, stutter, etc. This isn't multi-cow processing comparison like you retardedly present it as.
>>54853937
>multi-cow processing
I need this
>>54853875
To each their own. But Yeah, it is funny as hell, seeing all these Nvidia fanboys getting mad and trying to damage control.
>>54853950
It's an old image macro explaining multiprocessing but I can't find it.
>>54853765
>Dont be surprised when you start seeing 2+ gpus on a single card again i the next few years
Because the AMD housefire meme has to be strong and alive for as long as AMD exists (not much longer if the console deals aren't' enough to prop up their business).
>we can't beat Nvidia with a single chip Raja what do we do!
>put more chips
Indian engineering.
>>54851099
Dual GPU solutions work as smoothly as a dual toilet solution in India.
http://www.hardocp.com/article/2016/05/27/from_ati_to_amd_back_journey_in_futility
>In the simplest terms AMD has created a product that runs hotter and slower than its competition's new architecture by a potentially significant margin.
AYYMD HOUSEFIRES
AYYMD HOUSEFIRES EVERYWHERE
>>54853884
I suspect that memory bandwidth is the bottleneck on te AMD card
>>54854006
>tfw the crying AMD wojak is hot at fuck because he's wearing a hat for the snow when his GPU heats up his house more than enough
>>54854006
Kill yourself my man
>>54854028
Isn't part of DX12 Async supposed to alleviate that?
>>54853937
Do you even have a remote idea of how a graphic card works ?
If it has multiple cores, then the load is distributed across all the cores (for the sake of arguments, let's count that in number of pixel to process, even if it is immensly more complex than that)
You have X pixels to compute to generate a frame. You have Y cores to compute those pixels. A core is taking Z seconds to compute one pixel.
With one card : Time = (X / Y) * Z
With two cards : Time = (X / (Y * 2)) * Z
And graphics are actually something you can multithread easily. I have wrote a graphics engine, and you don't need to have that much shared memory between the threads (basically just the scene definition). So because threads are not relying on other thread outputs to compute it will get done faster.
Now go back to >>>/v/
>>54854050
>BUTTMAD AMDPOORFAGS DETECTED
>>54850970
>HALF THE PERFORMANCE FOR THE SAME TDP
>BUT IT'S CHEAPER!!!!!!
my sides
>yfw Intel buys AMD after this flops hard just so they can eliminate their competition
NVIDIA WON
V
I
D
I
A
W
O
N
>>54854075
>but it's cheaper
You can literally buy 3 of them for the price of one gtx 1080
>>54853950
>>>multi-cow processing
>>54853937
>>54854110
Yeah, and get 3X the microstutter
http://www.pcper.com/files/review/2016-04-29/GTAV_3840x2160_PLOT.png
>AYYMD
TOPPEST LELKEK
>>54854231
>4K silky smooth microstutter is a feature
>>54854099
they already bought mediatek so nope, probably QUALCOMM gonna buy whole amd
>>54854302
Qualcomm should amd really fucking sucks
>>54851099
its $200 less for 2 gpus with same performance. but I guess if you dont like your shekels, you buy nvidia
>mfw nvidia actually definitively won in my lifetime
>>54854488
>that typical nvidia premature ejaculation
>>54850970
nvidia jews are just mad that they got btfo by a poo in loo
