>180w tdp >up to 311w power peak power consumption in game

Ket me guess slim shady 180w without boost?

>>54914976
>metro last light
people still play this?

>>54914976
>using amd's HARDWARE
Good goy

>>54914976
End yourself

>>54914976
AyyMD shills are out in full force again it seems

>>54915051
I see the shill is in fear of unemployment

>>54915005
Kind of. It's more along the lines of being taxing for absolutely no reason, and its benchmark is decent.

>>54914976
Funny how AMDrones always gasp at straws.
Last generation power consumption didn't matter and now all of a sudden it does because AMD can't push out anything more then muh efficient midrange GPUs?
Pathetic

>>54915018
>tfw still on a HD5870
480X CAN'T HAPPEN SOON ENOUGH

>>54914976
Kek amds current lineup needs over 400W

>>54915090
Damage controlling corporate lies, eat literal shit, your indian streets could use cleaning

>>54914976
>AMDs current gen is slower and less efficient than the Nvidia GeForce GTX 1080
>BBBUUUUUUUUUT POWER CENSAMPSHN
Top lal amdrones

>>54914976
>POOlaris still can't compete

>>54915090
>he forgot stock 980ti reaches 430w, which is higher than Fury X

>>54915145
Lol no less than 100w under load, difference is amd didn't lie

>>54915178
Meant for
>>54915109

>>54915090
So now nvidiots are saying power consumption doesn't matter?

LEL

Haha oh boy

Reminder that Nvidia has always ALWAYS had shit peak power draw.

>311w
JUST tiers of disappointment and failure

File: 1462499908080.gif (1MB, 540x540px) Image search: [Google]

1MB, 540x540px

>>54915160
>480, not even 480X, 490/X
>$200
>beating cards over double the price
>can't compete

Wat

Reminder that the 480 is only 96w despite being advertised as 150w.

http://youtu.be/PJ5wYVu-tM8

>>54915385
150W is the maximum. It would never reach that. Besides that I think the low wattage can be accounted for by the ~51% gpu utilization. Think he mentioned that but forgot to factor it in because it would have been a 'mess'.

>>54915451
>only 50% gpu utilisation

IT'S OVER, NVIDIA IS FINISHED

>>54914976
What happens when your PSU only can deliver less than 250-300W to the card?

>>54915512
Your PSU delivers over 1000w as it breaks

>>54915462
Let's not get ahead of ourselves. I want AMD to wreck shit as much as anyone, but 50% utilization is almost certainly due to poor crossfire scaling, due to early drivers.

The upside is that this is very likely to improve before release, but you won't be seeing 100% usage out of crossfire anytime soon.

>>54915512
It shuts down.

>post the toms articles the past couple days
>nobody here even remotely capable of discussing architectural nuances
>OP just starts shitposting while ignoring all of the real issues it highlights
God this board is awful.

The GTX 1080 with stock clocks momentarily touches 311w in a normal gaming workload. It's average power consumption is in line with its power target of 175w.
The same card still with stock clocks during Furmark will only reach 224w at its upper limit while maintaining the same power target. Pascal has the exact same behavior of Maxwell, and for a 16nm FinFET chip to be behaving identically to a 28nm planar part is not good. It is entirely thanks to the arch's load balancing(throttling with extreme granularity so fast that true clock speed can't even be displayed in software) that Pascal is competitive. Nvidia lost all of the process advantage brought by TSMC's 16nm node.

http://www.tomshardware.com/reviews/nvidia-geforce-gtx-1080-pascal,4572-10.html

Compare to Fury X

http://www.tomshardware.com/reviews/amd-radeon-r9-fury-x,4196-7.html

The peak line draw of Fury X is 453w while the average is 220w in game. This AMD's PowerTune IP at work.
During Furmark however we see exactly what one would expect. Average draw is up to 347w and peak is about the same at 448w. All the SIMD lanes in the CUs are being fully utilized at a fairly steady rate, and those transistors switching are using power. Power consumption goes hand in hand with utilization, the two never become uncoupled. Maxwell, and Maxwell 1.1 aka Pascal are architectures tuned specifically for gaming, and they exhibit entirely different behavior in constant high throughput workloads.

>>54915558
>tfw HD5770 crossfire
480 can't come soon enough. If it's close to 1070 in performance consider me sold

File: glory.jpg (70KB, 500x650px) Image search: [Google]

70KB, 500x650px

>>54915585

>pls be a grill

>>54915635
plz be in London

>>54915681
Essex ok?

File: 1390832049730.jpg (78KB, 500x500px) Image search: [Google]

78KB, 500x500px

>>54915585
Are you a grill?

File: smoke.jpg (44KB, 398x600px) Image search: [Google]

44KB, 398x600px

>>54915696

>>54914976
Wouldn't power consumption be higher at 1080p or 1440p?

>>54915857
Fair enough

>>54915868
No

>>54915879
Yes. Higher framerate = more energy being consumed

RX 480 literally cannot reach past 225w due to max power output of 1 6 pin and PCI-E x3.0, is specced at 150w, we'll have to wait to see final performance

Everyone knew the 180w spec of 1080 was bullshit. but hitting 300+w? that's pretty bad....

But still, does anyone here actually care? I have an 850w PSU. the rest of my system takes up 150w....

so If I had to choose the best card for the money, I wouldn't give a fuck about power consumption if it meant I would be getting the best perf/$.

>>54915964
Didn't the r9 295x2 pull more power than the power supply cables were rated for?

>>54915070
There's no shills. /g/ loves budget technology like old thinkpads, the Moto g, etc etc. This place is full of cash strapped young people and actual NEETs. Of course there's so many AMD fanboys here.

>>54915950
No

>>54915999
Are you retarded?

>>54915950
>>54915868
Although lower resolutions hit higher frame rates, the GPU is processing less data to make each frame. High resolution workloads are more intensive.
You can't make a definitive statement on such an open ended question regardless. I'm speaking in practical terms.

1080p@144hz might draw more power than 4k@30hz, but that isn't a realistic scenario. In benching one title at two different resolutions the higher res is typically going to use more power.

>>54915964
Again, peak draw happens for millisecond, its not a static load on the line. Only the average matters for accumulative power consumption.
And yet again, PCI-E power spec nonsense that you keep regurgitating is not gospel. Components can and do draw excessive amounts of power over slots and cables. Ratings for load averages are not peak load ratings.

>>54915981
http://www.tomshardware.com/reviews/graphics-card-power-supply-balance,3979-5.html

The 295x2 has extremely strict power management, and it doesn't draw that much under load. Peak hits under 475w, and is generally close to the average under heavy load.

>>54914976
Is this total system power draw or only the card?

Can my PSU with 18 amps on the PCI-E rail run this card?

File: 09-58-19-05-Asus-GTX-960-Strix-75-Watts-Limit_w_600.png (66KB, 600x422px) Image search: [Google]

66KB, 600x422px

Can't wait to see how shit the Asus GTX 1080 will be.

File: rx 480 scaling.png (27KB, 880x183px) Image search: [Google]

27KB, 880x183px

>>54915451
>Besides that I think the low wattage can be accounted for by the ~51% gpu utilization.
Here's some clarification on what that meant. It's not actually 51%, it meant that dual rx 480s perform as well as 183% of a single rs 480. So using the AotS benchmark, x+.83x=62.5 fps , x=34 fps. A single rx 480 gets 34 fps in that benchmark. Make of that what you will.

>>54916084
GPU only isolated with two oscilloscopes and some fancy probes.

>>54916038
Are you?

File: 10-R9-295X2-Power-Consumption-Gaming.png (53KB, 591x934px) Image search: [Google]

53KB, 591x934px

>>54916067

>The 295x2 has extremely strict power management

The magic of the 295x2 is despite its mental power draw it keeps way under the limit for the pcie slot - fuck knows what a lot of mobos would do if that thing actually tested the supposed 75w pcie can provide (catch fire I would guess).

>>54916172
>not even 40w
And then you get Nvidia abominations like >>54916097

>>54915964
The 225W isn't a physical cap, it's a specification cap. Doesn't mean it can't take like 400W with just a single 6 pin. Chances that'll end poorly though.

Also it's more like 150W, because modern graphics cards don't usually take power from the PCI-E slot if they have a PCI-E power connector.

>>54916067
Do these millisecond peak draws make any difference when you're calculating power requirements for your system?

>>54916113
So that's a no? :(

>>54916367
No. When you're calculating power requirements for your system, I assume you are calculating the max possible power consumption.

Which is literally impossible to achieve without artificial load. You'll hit maybe 50-70% of maximum power consumption while gaming if you're lucky.

>>54916367
Yes. If you're using a 500w psu and your system spikes to 550w several times a minute, even though the average power consumption might be only 400w your psu will still wear out much quicker and might actually cause damage to other parts.

If you have a gtx 1080 and i5 6600 you should have a 600w power supply at least.

File: 1440113118491.png (1MB, 1062x815px) Image search: [Google]

1MB, 1062x815px

Will this power supply and motherboard support a GTX 1080 or RX 480?

The wattage is there, but the amps on each rail seem to indicate a GTX 1080 is out unless I connect it to two different rails at once.

>>54916413
But I'm not only going to be gaming. I may do some h265 encoding from time to time.

>>54916548
That uses less power than gaming, you know.

The only way to get what is typically regarded as 100% power consumption is to run Furmark and IntelBurnTest at the same time (assuming Intel processor).

>>54916038
He's not wrong. If you want the actually best performing, longest lasting, high quality hardware, go for nvidia and Intel. If you are poor as fuck and need performance per dollar, go AMD.

I worked a summer and bought a nice $3000 machine. There's no way I could get better performance with amd hardware than an i7-4790k and a 980ti.

>>54916584
I7 5960x and fury x crossfire.

File: 1449272809486.gif (2MB, 500x500px) Image search: [Google]

2MB, 500x500px

>>54916584

>>54914976
Yes

>>54915051
>>54915070
>>54915090
>>54915109
>>>/v/
/g/ isnt the place for summerfags like you

And I'm just sitting here waiting for the 460x.

>>54916530
That's not how power supplies work. There is no other "rail" that you can connect a PCIE component to. They need the 12V rail.

Dual 12V rails are inferior to a single rail with higher amperage. The maximum wattage that one of your 12v rails can supply is just over 200W. That PSU isn't great.

>>54916530
>delta
>dual rail

HOUSEFIRES

>>54916584
> $3,000
>1 GPU

Easiest bait to spot

>>54914976
>oscilloscope
electrical engineer here, i am almost CERTAIN they fucked up the measurements.

They could be picking up noise from the GPU, not real power.

>>54918638
NOPE

>>54918638
Sure anonymous guy on /g/, I bet you are right that the guys that do this for a living "fucked up the measurements."

>>54918638
>I know best because of my learnings, everybody's dumb but me!

End yourself.

>>54918657
>>54918667
Want me to explain why?
By the way,
1. I like AMD, and I use their GPUs. I am an AMD person.
2. I work for NVIDIA.

>>54918686
No. Nobody cares about your stupid opinions. Go away.

File: 1443245432606.gif (2MB, 325x235px) Image search: [Google]

2MB, 325x235px

>>54918686
>1. I like AMD, and I use their GPUs. I am an AMD person.
>2. I work for NVIDIA.

>>54918759
You would too if you saw how they made those GPUs.

>>54918686
Why I think it might be off:
- GPUs and other extra high speed digital electronics spew out a ton of noise and you can easily pick them up with an oscilloscope if you don't take preventative measures. I really don't think these fuckers know what they are doing.
- "Peak power" doesn't matter, because almost all that shit is decoupled away.
- Speaking of decoupling, even IF it is not noise, how the hell is it possible that you have so much high frequency noise from the power bus? Why is it not decoupled onboard?

Why I think it might be correct:
- Their current probes are made by Rohde and Schwarz, and they know what the fuck they are doing. I (hope) that their products are idiot-proof.

>>54918759
>>54918768
I quite like the company actually. Lot's of really cool engineers. I just don't like some of the business people.

>>54918792
Additionally, if I wanted to do it, I'd use a data collection bench meter, not a scope.

File: 1455067772133.jpg (15KB, 167x223px) Image search: [Google]

15KB, 167x223px

>>54918792
>>54918812
I don't doubt your reasons for being skeptical but it's not like Tom's Hardware hasn't been around for 20 years. Do you think you're the only EE that's ever questioned their methodology? I'm sure if we both googled hard enough we could find a detailed explanation from a previous article of what they do and why it works etc.

>>54918838
Is there an article associated with that picture? I kinda wanna know, for my own good.
Tom's hardware has been around for that long, but I'm not sure if this is his first time doing this particular measurement.

>>54918884
They do it with every card.

>>54916367
Not necessarily. A quality PSU will be able to safely deliver far more wattage than its rated for. Some Seasonic 400w units are capable of outputting a sustained 630~ so for a unit like that it would never be a problem. If you're relying on something far less robust then you could get a bluescreen as components crash from not getting enough power. Thats pretty unlikely though.

>>54916433
A differential that small would never cause any problems.

>>54919077
Yes it would.

>>54915178
Says who lol
Not him but I'm running a 980ti sli with CPU and gpu overclocks and I've had zero issues with my corsair ax760 under full load

>>54919396
Says Linus, of all people
http://www.tomshardware.com/reviews/amd-radeon-r9-fury-x,4196-7.html

>AMD does a great job managing power, and we're happy to see the company competing on a more even footing with Nvidia in this area as well. The Radeon R9 Fury X even manages to beat Nvidia’s reference GeForce GTX 980 Ti due to the Fury's better cooling and the resulting lower leakage currents.

TDP is the rating it has for COOLING.
As in what cooler is required to cool it, so it'll be the peak average rather than the peak overall.
Wattage usage spikes are perfectly normal.

>>54918686
>I like AMD, trust me guys, I'm an AMD person just like you all!

>>54919432
>reference
Into the trash it goes
Have a sc acx2.0
A g1 gaymin windforce

Thanks for the info though anon

>>54919443
Yep, and given how dense the CU is, and how dense 14nm FinFET is, I don't see the RX 480 drawing 150w at stock clocks. Not even close. We know that the little Polaris 11 draws just under 50w. I'd place the 480 at about 100w.

>>54919558
>moving goalposts
The post

>>54919627
I don't know this meme
All I'm saying is my anecdotal evidence suggests that 760 watts albeit great efficiency is enough to run an entire computer with 980ti sli, let alone two gpus alone at max spike load totalling almost 900w
The gigabyte card even uses two 8 pins as opposed to evga 8+6

>>54919657
That's not how tdp works, anon.

>>54919675
That would make sense considering I haven't looked into it
My system did shut down once when increasing power usage past 100% on afterburner though so I put it back to 100 and it's been fine ever since

How often do you replace PSUs?

Is it ok moving them from different builds

Have Seasonic X660 very happy with it, 4790k oc/980ti oc

File: 1464355436659.jpg (106KB, 797x694px) Image search: [Google]

106KB, 797x694px

>>54915070
>nvidiot trying to counter benchmarks by shoting "shill"

>>54918458
>>54918365
Do either of you know anything about what you're talking about?

Sometimes I don't know why I bother with /g/.

>>54919855
Replace them when they stop working

>>54915598
It won't be that close, but the 1070 is going to be a LOT more money than the 480.

770GTX 2GB
A lot of these newer games can't keep stable 60fps. DOOM is the most hurtful one since it doesn't seem to matter if it's on super low or super high the frame rate is 60fps about 80% of the time.

Which one of these would be better?
1070
or the new AMD one. I seriously don't use AMD because I've always been intimidated by the drivers.

>>54919999
They've really stepped their drivers up since early 2015/late 2014.

I think it's best to wait and see what the 480 offers. I know your pain, I'm on a 680 myself.

>>54920015
Well, which one would allow me to use my old card as a slave or something?

I honestly don't know what I would do with it since there is no real place to resell it here and I'd hate to just shove it in the closet.

>>54920015
FUCK OFF.

Nvidia is the only option.

>>54919999
>I've always been intimidated by the drivers
So was I, but after having driver issues left and right with a pair of GTX 460's, and hearing from friends with similar driver issues that never got fixed in newer revisions, I gave it a chance and bought a 7950.
Haven't experienced a single issue on Windows. If you're more into Linux, then stick with the open-source driver. Haven't given me any trouble either.

The "no drivers", is just a meme, nothing more

>50w for entry level Polaris
>100w draw for RX 480
>200w draw for Fury replacement
>Summit Ridge CPUs
>Raven Ridge APUs
>enterprise Opterons scaling up to 32 cores with 8 channel memory
>a handful of new semi-custom design wins including enterprise parts


AMD will have a solid year of revenue in 2017.

>>54921420
So keen. Thinking of replacing my i5 750 and r7 260x with an APU. Will have to see how they perform.