Reminder that AMD has a fully enabled Polaris 10 GPU at 95W TDP which they sell to a even larger market than the consumer one, while the consumer one (you) gets the leaky 150W version others get the 95W version of the very same chip with the same clocks.
That means you're an idiot.
It also means AMD is fucking retarded and can't be bothered showing how efficient their chips actually are to the consumer market so all we can do is scream housefires.
AMD's marketing incompetence is astounding
Other urls found in this thread:
techpowerup.com
techpowerup.com
twitter.com
>look how much of tech illiterate retard I am
There is no difference between these embedded parts or discrete cards. The embedded parts don't maintain their clocks like the discrete cards, they have a lower power target, a setting that you can change yourself on any RX 480.
They have the same TFLOPs and last time I checked clocks are a part of TFLOPs.
Also even if the embeded parts run at 100MHz less that's still nowhere close enough to shave off 55W off the TDP no matter how high voltage the consumer chips are
lower tdp =/= more efficient. perf/watt will be the same no matter how much you reduce clock speed and voltage outside of a margin of error.
>tech illiterate retard
Nice projection Pajeet!
The things we dislike most in others are the characteristics we like least in ourselves.
if AMD has a rx480 at 95w then Nvidia has a Titan XP at 95w
Nvidia has a 1060M at 95W if that's what you're looking at.
>muh teraflops
is this the new gigahertz?
>They have the same TFLOPs and last time I checked clocks are a part of TFLOPs.
The image very clearly says 'up to 5.8 TFLOPS'
So you will only achieve that performance by OCIng said chip back to the stock desktop clocks, which will bring it back to ~150W power draw under load.
AMD has half the power efficiency of Nvidia, 480 and 1070 both consume 150w, 1070 destroys the 480.
No, it just means the clocks are unchanged from the desktop chip because TFLOPs = ALU * clockspeed
And rx480 having the same FLOPs and number of ALUs means it has the same clockspeed.
You need to learn how mobile GPU binning work. It's not like the mobile nvidia GPUs don't use less power than desktop.
What's strange about this? AMD chips have very high upper voltage ranges out of the box, these ones are just binned and come at a more mature process.
They've dropped the m part of the name, just to ensure confusion.
They probably arrived at that figure with some very liberal turbo boost targets that the card will never realistically hit. It'll be interesting to see what the actual performance is like.
We'll need to wait and see what the mobile offerings are like. For all we know, the rx480 could have been clocked above its "sweet spot" in order to compete against an NVidia part that performed better than what AMD were expecting.
>what is binning
Why would they waste there good chips on bargain basement consumer cards where TDP isn't even an issue when they can make massively more money selling them to enterprise customers?
Do you people have brain problems?
TDP IS ALWAYS A ISSUE I RUN MY HOME GRID OFF THE SUN
Enjoy paying $10000 for a quality and high wattage inverter that sounds like a jet engine unless you want plenty of low quality and dirty electricity
>>what is binning
binning doesn't magically get you chips that draw 2/3rds the power for the same performance. you MIGHT get some chips with less leakage that can pull off a 5-10% reduction, but the only major difference between these embedded/mobile GPUs and desktop GPUs are the clock speeds and voltage, and any extra disabled cores/rops.
You are massively underestimating the effects of binning.
You can undervolt a Fury X to use 210W at its stock clocks..
AMD simply has high voltages for consumer chips so they can sell more of them.
^This
Higher power target = higher yields = higher profit per die.
Wonder if the iMacs etc will get this version of the card
>what is fury nano
Hell the nano is around as effecient as maxwell is. One could argue the nano (and possibly the pro duo) are the pinnacle of the fiji lineup.
But this really ruins their image with the normalfag crowd that doesn't know about undervolting.
I knew something felt weird when I saw that Polaris had the same perf/watt as Fiji, a 28nm arch.
>Hell the nano is around as effecient as maxwell is
the nano wasn't any more efficient than other fiji cards. stop confusing perf/watt with overall power consumption.
also, the nano's '''efficiency''' was entirely created by throttling. it went to ~800mhz core clock even under weak 2d loads.
What the hell were you smoking? Perf/watt = efficiency, Nano was 10% slower than a Fury X but used 35% less power, ergo it's way more efficient.
The nano is approx the same perf/w at the fury non-x for similar performance to the fury non-x. Not sure where you're getting the idea that it's anything else.
techpowerup.com
techpowerup.com
I'm not sure why the Nano even existed it was essentially just a $650 fury non-x performance and power wise.
why do nvidiots create illiterate amd slander threads every day?
i sure you morons get paid for all this counter marketing
Slander? AMD are incompetent and put out shit products. If anything there is constant circlejerks about NVIDIA and Intel being evil and unethical by Indian AMD employees here and on plebbit.