1st RX Vega Custom AIB Review - Vega is confirmed full of shit

>>61968065
http://hexus.net/tech/reviews/graphics/109078-asus-radeon-rx-vega-64-strix-gaming/?page=8
>A quick word on overclocking. We managed to run the Asus Radeon RX Vega 64 at 1,980MHz core and 1,000MHz
>1,980MHz

NVIDIOTS BTFO
FINISHED AND BANKRUPT
HOW WILL THEY EVER RECOVER?

Also,
>Hexus in charge of not being retarded
How can these guys even call themselves reviewers?

>>61968072
Oh jesus lord...what a joke.

>>61968088
Jesus Christ.

>>61968117
>1980 MHz
>Massive power consumption
>Still barely better than a 1080
>No availability

>>61968117
https://www.computerbase.de/2017-08/asus-radeon-rx-vega-64-vorserie-test/

>>61968117
>7% of improvement for 200W
Poor Vega that shouldn't

>>61968117
>...The downside of this voltage-induced performance is that power consumption rose to 532W

>>61968164
At least it isn't voltage locked like Pascal.
>Overclocking is a bad thing t. Nvidiots 2017

>>61968117
vega doesn't scale above 1650Mhz
https://www.youtube.com/watch?v=m2-xZvv5wd8

why only 4k tests though?
also those temps, asus sucks at amd cards as usual

File: 49d6aede-0195-4195-a960-4aefa25e2978.png (18KB, 756x227px) Image search: [Google]

18KB, 756x227px

So my 600W PSU actually will hadle it just fine, will it?

>>61968065
Holy shit, so much for waiting for aftermarket. Fuck I guess I will just have to get the 1080ti like is hold of months ago instead of waiting.

The worse part is that it means that the 4k monitor I buy will be Gsync instead of freesync and will tie me in to nvidia until I upgrade to 8k, vr, or holograms/whatever.

>>61968247
Depends on rest of your parts. (ram[inc clock speed], mobo, cpu, how many hard drives and what type, etc)

>>61968262
Blame AMD for that, their sheer incompetence and blatant lies have destroyed them. I'm going to be surprised if they can even make it to Navi.

>>61968293
2500K an SSD and two hard drives.

>>61968200
I'm agree tho, I think Nvidia locked the voltage because they want make a refresh Pascal, but now Vega is so shit than Nvidia don't need to do that.

https://www.techpowerup.com/231751/evga-geforce-gtx-1080-ti-overclocked-to-over-3-ghz-under-ln2

>EVGA GeForce GTX 1080 Ti Overclocked to over 3 GHz under LN2

File: EtkcsPCe.png (128KB, 464x512px) Image search: [Google]

128KB, 464x512px

>>61968321
So they can re release it twice now?
First with new memory, then with new memory and unlocked.

It runs cooler than reference, but uses 44 more watts. Interdasting.

File: 1489801396082.jpg (96KB, 751x775px) Image search: [Google]

96KB, 751x775px

wait...what?

>it's so bad it's sold out everywhere

>>61968088
Poor Volta.

Jesus Christ these cards suck.

Just gonna hang onto my 390x and get a ryzen CPU these Vega gpus are terrible.

>>61968417
>Nein, Ja

>>61968417
oh, I get it now, they basically simply increased power limited and let boost go wild? that's low even for asus
so it simply doesn't throttle like reference, they didn't do anything

>>61968429
> make just a few cards
> WOW, IT'S OUT OF STOCK OMG

>>61968447
Lüfter abgeschaltet (2D)

that's for passive fan control during idle

File: 14992235892430.jpg (16KB, 424x501px) Image search: [Google]

16KB, 424x501px

>>61968455
>Vorserienexemplar
>pre-serial sample

it's not final. they say they are gonna test all partner cards in late september

Still is faster than reference, guys. Cool down.

File: 1495996614451.jpg (103KB, 1272x437px) Image search: [Google]

103KB, 1272x437px

>>61968502
it's asus, it's final

I don't understand why it's got power draw this higher when it runs exact same clock as reference

>>61968065
It's over. AMD is finished.

>>61968532
Clearly not the same clocks?

>>61968532
they say it right in the tl:dr, it's the engineer sample

File: 290ef8d7-b3eb-4929-9a24-dfd21ac6edfa.png (18KB, 756x227px) Image search: [Google]

18KB, 756x227px

>>61968455
>they didn't do anything
they slapped a better cooler on it.
That's their job.
You can adjust the clocks yourself to fit your system and cooling conditions in your case.

>>61968182
Their test bench consumes 200-275 watts

>>61968065
>More than 100 W higher power consumption than a GTX 1080
>Still slower
>200€ more expensive

Fuck you too, AMD. Despite my Freesync monitor, I'm going to buy a GTX 1080, maybe a 1080TI. And some Nvidia shares.

File: 2017-08-18-13-46-44.png (10KB, 790x170px) Image search: [Google]

10KB, 790x170px

>>61968737

>>61968816
>At the wall

>>61968873
Did you read the thread? >>61968117

>>61968072
>>61968088
>reviewers don't specify what power mode they were using the card in
>vega 64 max goes over 400W

There is no way they could have gotten that consumption without running the card in turbo mode, which is a stupid thing to do because fps only goes up around 3% at the cost of 60-70W extra power consumption, plus a hella lotta throttling

>>61968065
>Massive cooler
>Three fans
>Still worse than reference liquid cooler

Why buy an aftermarket card when you can just get a liquid cooled one?

If they are capable of making GPUs that clock at 1900mhz why doesn't AMD just throw better memory on polaris to feed the increased compute, a few more cores, and run that bitch at 1900?
It's already really good at 1450 or whatever the stock clocks for the 580 are. Surely another 500mhz and better memory would put it ahead of even the vega 56.

Friendly reminder that Vega is pretty efficient at lower clocks. When they drop their performance goal down for APUs and maybe the Polaris replacement next year it should do well.
These cards were an after thought. Radeons never outsell NVDA even when they are strictly superior.

>>61968987
Because historically AMD doesn't overlock that well and power consumption shoots up drastically

>>61968302

It would be tight, if you under volted/under clocked the CPU and GPU then yeah maybe, but I would just go with a 800W Gold+ PSU.

>>61968816
ya like i said they tested at the wall that's total system power consumption that skylake chip at 4.6 takes min 200 watts

File: 1499893689022.jpg (422KB, 1920x1080px) Image search: [Google]

422KB, 1920x1080px

>>61969008
His efficiency would be horrendous as well wouldn't it?
I bought an expensive PSU with 80+ gold and 1000 watts because I figure it's going to last an eternity and save some electricity. My last PSU lasted 10 years and that was just when I retired it, still works. I think they put the best shit in the high power PSUs so if you pull 450-600 watts out of a 1000 watt psu your efficiency will be ideal and the thing should last a life time.
It's also very quiet.

>>61968816
>at the wall

>>61969008
>800W
>>61969053
>1000 watts
What the fuck are you talking about mates?

If they measure 460 at the wall (total system)
and my PSU is 600w 80 cerified (no precious metals) then my efficiency is way above 80 mark somewhere near the 90s

>>61969053

He's still running a 2500k, 800w would be the min.


>>61969100
plus the 100W+ your CPU needs, plus the 50W your mobo needs, and the 15 or so watts your drives need.

>>61969100
PSUs are most efficient at 50% usage. I know you do not own a 1000 watt power supply. I'm saying I bought one even though I don't use any where near 1000 watts.
Your 600 watt should power it but you are going to loose efficiency and the fan in the psu is probably going to be a fair bit louder.

File: delusion.jpg (44KB, 640x480px) Image search: [Google]

44KB, 640x480px

>>61969069
>>61969028
>implying it draw less power than reference vega

https://youtu.be/A6PXixRnrkA?t=4m5s

it's really quiet though, damn i'd take 3 slot card any day vs loud card

>>61969008
I had an old building that ran fine.
>FX 8350 housefire
>16GB 1600mhz RAM
>SSD and HDD
>R9 290 VaporX housefire but vaporchamber
>650W PSU

As long as he doesn't go full retard and stick to the "normal" GPU setting, whatever that is, he should be okay.

>>61969123
not on my glorious seasonic titanium 650w PSU

VEGA hits a hard wall above 1650. There is very little gain to be had. Also there is a bug that over 1800 and it returns to stock clocks even though it is reporting higher. Run 3Dmark and it is stock results at anything higher than 1800.

>>61969203
Sick. Yeah I really want one of those, such a nice product. NH D15 of power supplies. The 850 evo of power supplies.

>>61969214
there is also a bug that at ln2 OC discard goes haywire and discards things that it shouldn't

File: AMD-Radeon-Vega-Frontier-Edition-Liquid-Cooled_350W-740x377.png (77KB, 740x377px) Image search: [Google]

77KB, 740x377px

>>61969144
>it draw less power than reference vega
This says vega draws 350 watts tops
>>61969119
>plus the 100W+ your CPU needs, plus the 50W your mobo needs, and the 15 or so watts your drives need.

So going crazy it's like
350+100+50+30+30 this makes like 560 considering i'm stress testing my Vega, my CPU and my fucking drives.
In such situation my PSU would still be within 80+ territory. and it also has to have some extra power above the listed value to make sure it doesn't fail at exactly 600.

Where are you getting your 800 and 1000 figures?

>>61969198
>As long as he doesn't go full retard and stick to the "normal" GPU setting

this is /g/

>>61969203
Chieftec strong.

>>61969252

how is 560 80% of 600? It is 480, 80% of 800 is 640.

I don't know guys. Yeah, I know, it's really not a very good product and it's expensive. But no matter how I look at it I'm saving money buying it and a freesync monitor over going gsync.
I really wanted this shit to be good but I guess it's still better than cross fire rx 580 or the fury X.
Nvidia please adopt freesync.

who cares about 64, show me 56 strix

File: screenshot-www.computerbase.de-2017-08-18-14-22-22.png (35KB, 679x351px) Image search: [Google]

35KB, 679x351px

>>61968737
>>61968873
>>61969028
>>61969069
>>61969252

>the RX Vega draw ~300 watts
>the RX Vega Asus Strix draw ~320 watts
>b-b-but the overcloked Asus Strix will draw less watts

https://www.computerbase.de/2017-08/asus-radeon-rx-vega-64-vorserie-test/3/#diagramm-leistungsaufnahme-der-grafikkarte-the-witcher-3

AMDrones are funny

>>61969278
>Nvidia please adopt freesync.
yeah, not until they get 100% over ROI on the IC
so wait 5 more years, then wait 3 more when they fix drivers for it

>>61969278
If you can grab a Vega 56 at the right price it will be a good deal. 64 is a huge fail and more so with the shitty pricing.

>>61969274
The PSU is certified to have 80% efficiency coefficient while delivering 600 watts.

This means that when delivering 600 watts it consumes 750 watts and heats itself up by 150 watts.

>>61969296
unrelated question, what's the difference between 1080 gaminx x and gaming x+?
x+ costs 10% less here

https://www.computerbase.de/2017-08/radeon-rx-vega-64-56-test/#abschnitt_amd_radeon_rx_vega_64__56_im_test

>At the same clockspeeds as an R9 Fury X, the Vega 64 is, on average, 6% faster.

AHAHAHAHAHAHAHA
*Breathes in*
AAAAAAAAAAAAAAHAHAHAHAHAHA

>6 (SIX) %

>>61969327
>at the same clockspeeds
>mouthbreather
checks out

>>61969344
>AMD apologist
>Huge retard
checks out

>>61969327
At least they have an IPC boost.
Nvidia did cut the IPC from Maxwell, lol.

>>61969353
You don't know what same clockspeeds mean, mouthbreather?

>>61968065
Fuck, my reference card arrives tomorrow.

Hory sheet you have to read this plebbit post. Literal rage quitting after buying Vega.

Abject failure by AMD/RTG and buyers alike.

http://www.reddit.com/r/Amd/comments/6ugqql/i_bought_vega_64_mrsp_and_regret_it/

>>61969430
Pls give benchmarks in thread

>>61969327
Even Crossfire reference RX480 beats this thing for about equal power draw.
I can tell because I has one.
Also, I bought Vega to play Doom at 4K, because Vulkan doesn't crossfire.

File: 1422475902008.jpg (44KB, 640x640px) Image search: [Google]

44KB, 640x640px

>>61969430
>buying reference
>not waiting for aib
>not waiting for drivers

AMD said Vega would have

>2X performance per clock
>4X performance per Watt

None of it is true.

When Nvidia promise something they at least deliver.

>>61969481
It looked like AIB were at least a month away.
Guess I was wrong.

>>61969418
It means their "new" architecture is barely an improvement over Fiji.
All this time Raja was telling lies.

why AIBs do not make 3 slot cards? really, i'd pay for it, it would be cool under 600RPM

JUST WAIT

DAIJOUBOU!

>>61969515
fiji can't get even close to p6000
their driver team obviously focused on pro drivers

>>61969480
>Crossfire anything
No

>>61969498
They are.
It's an engineering sample, they arent coming to shelves tomorrow.

I'm still not buying reference though, they just suck.

>>61968065
VEGA seems incredibly unbalanced. I wonder if a cut down version might be better? Has anyone tried undervolting/clocking VEGA? Maybe AMD would be better off replacing polaris with VEGA?

>>61969545
asus says early September
but i'd stay away from asus

>>61969558
> I wonder if a cut down version might be better?
It would be was it not for the hard power lock.
It cant go above 300 watt so it's cant OC too much.
>Has anyone tried undervolting/clocking VEGA?
They did it does a better job keeping the boosclock.

>>61969543
Honestly, my last experience with it was Radeon HD 4850s and that microstutered as fuck.
RX 480 worked pretty well in titles that supported it. Namely F1 2016, Rotr, GTA V, Middle Earth whatever, and Elite Dangerous, that I have played.
Actually, now that I think about it, most major titles for me did support crossfire, and it was a mostly bug-free/stutterless experience.
At least, I'll get to play Doom 4. I heard it's a pretty good game.

File: 1502378645894.jpg (89KB, 1400x700px) Image search: [Google]

89KB, 1400x700px

>>61968065
I just want a 580 for $200. THE PRICE IT IS SUPPOSED TO BE

>>61969594
Same with Polaris. My RX 480 used to down clock when running at default voltage. I underclock and allow the power limit to +50% and all is good. Actually I may try overclocking my memory too and see if it remains stable. I know I can get to 1420/2200 with +100mv voltage

>>61969661
Sorry I meant to type 'undervolt'

>>61969558
Vega perf/watt improved a lot with undervolted. The stock BIOS is absurdly overvolted to increase yields. Gamers Nexus got their Vega 56 to be stable at ~1.025v with a +50% power offset and +180mhz memory OC where Vega 56 will tie a Vega 64 Air in every year for 240watts total card consumption.

File: mLpXllb.jpg (189KB, 1920x1080px) Image search: [Google]

189KB, 1920x1080px

>>61968117

Pufff...

2100 MHz core / 5005 MHz memory / 1.05v

>>61969695

>5500 MHz memory

fixed

>>61969659
>tfw you wanted to buy a rx 570 when it was really cheap and you didn't
>tfw you could have sold it and get a ryzen 1600 + b350 board + 16gb of ddr4 ram
fucking hell

>>61969315

.. that's not how that works.

>>61969686
Well that's the first thing I'm planning to do.
I'll make an update, as it seems no-one has got the card, or doesn't share results.
You have to be aware, though, that undervolting could possibly not work on your gpu.
AMD has a broad selection, and you could end up with a turd that actually needs 1.2V to run those clocks.

>>61969719
This is true, but I'd expect the vast majority to work at 1.1v tops. Gamers Nexus is the best resource for Vega undervolted info at the moment as most reviewers out there have no idea WTF they are doing with undervolting or overclocking on launch drivers (like core overclock being mostly broken on launch drivers). The way to go for now is to set power limit +50%, max out memory OC, then max out undervolt but leave core clocks stock.

File: 1422845562338.webm (376KB, 390x293px) Image search: [Google]

376KB, 390x293px

>>61969315

>>61969719
Vega64's base clock is 1247Mhz.
If that's not flashing red in your mind, you're thinking it wrong.

File: 1473719567457.png (29KB, 487x372px) Image search: [Google]

29KB, 487x372px

>>61969123
Most decent power supplies today have only slight drops in efficiency around their peak, usually around 3-8% or less. They're only very low at low utilization, probably less than 10% load will cause efficiency to drop quite a bit.

See image for an EVGA B3 which is their budget series in a high temp load test.

File: Jap.jpg (24KB, 324x479px) Image search: [Google]

24KB, 324x479px

>>61968614
>>61968247

That god awful TDP, hot enough to cook pasta..

Jesus fucking christ...

You done it AMD, ain't buying your shit this time.

>>61968247
since that is total system draw, I don't see why not?

>>61968445
I was going to get a new card, I might just strap an aio on my 390 and go ham with the clocks instead.

File: 1502385495418.jpg (99KB, 640x840px) Image search: [Google]

99KB, 640x840px

>>61969791
>>61969715
That is.
https://en.wikipedia.org/wiki/80_Plus
>The efficiency of a computer power supply is its output power divided by its input power; the remaining input power is converted into heat. For instance, a 600-watt power supply with 60% efficiency running at full load would draw 1000 W from the mains and would therefore waste 400 W as heat. On the other hand, a 600-watt power supply with 80% efficiency running at full load would draw 750 W from the mains and would therefore waste only 150 W as heat.

It really looks like AMD has just given up on making Graphics cards.
Now, they make compute devices that just so happen to be able to run games.
And even that is gonna be BTFO by Volta's tensor shit.
Should they sell the Radeon brand to someone actually interested in making GPUs?

>>61970051
that would be a good idea

>>61969100
80+ is is <87% your efficiency is around 85% tops. Which is good enough
>>61969119
plus the 100W+ your CPU needs, plus the 50W your mobo needs, and the 15 or so watts your drives need.

erm, "at the wall" means everything.
>>61969123
th3 old 50% argument is stupid, your pc isn't operating balls to the wall 24/7. Efficiency will be about the same shitposting, maybe a little better on a lower wattage psu

>>61969482
we did get lied to and shafted on that, I remember the slides RTG.

>>61969719
>AMD has a broad selection, and you could end up with a turd that actually needs 1.2V to run those clocks.
This is true, I have a 390x that will do 1080mhz max at 1.2v [email protected] and not much more no matter how high you crank it.

>>61968146
Oc don't do shit when the hbmeme is choking bandwidth

>But muh wider bus
It's several order of magnitude shitty than gddr

>>61968417
Confirmed hbmeme can't oc

>>61970500
it can OC +180mhz easily

HBM is fine. It's the GPU that can't OC for shit and is a power monster. Also either terrible drivers or terrible architecture or both.

>>61968223
Why would anyone test at anything other than 4k? People buying new cards now would want them to perform at 4k which is rapidly becoming the new standard. 4k monitors can already be had for less than the cost of this gpu.

>>61968417
Not even a real oc.

Holy shit balls this is a new low.

Perhaps they just wanted to be the first out the gate with a 3rd party cooler?

>>61969319
More vrm

>>61969278
HDMI 2.1 has it's own VRR spec and since HDMI has a far larger market everyone will have to support it. Both Gsync and Freesync have served their purpose but they will likely matter far less in the coming years.

>>61970575
Perhaps because you are a shill and will say "hey ti is just 15 fps better".
4k is no where near mainstream, hell majority of idiots are still on 1080.

>>61969281

AMD really shit the bed with the Vega64, the 56 is the only thing that is even remotely interesting.
However, how does no one remember that ASUS is a shit brand for GPUs? Go Sapphire or go Nvidia.

WHY AMD?
WHY YOU CAN'T DO A RYZEN IN GPU?
YOU KILLED INTEL AND YOU`RE STILL BEATING HIS ROTTEN KIKE CORPSE. DO THE SAME TO NVIDIA FOR THE FUCKS SAKE

>>61971090
>A RYZEN IN GPU?
It's an EPYC in gpu
It's called NAVI

>>61971108
>It's an EPYC in gpu
4 trashy small gpus, is still a trashy gpu

If AMD can't deliver a good architecture MCM won't matter.

>>61971137
Brian, please.

>>61968907
Did you notice the better performing Nvidia cards in that chart beating Vega on FPS/watt even when Vega is not in turbo mode?

It draws more power even in powersave mode.

>>61971137
Go away Jason NotSoHung

>>61971468

youtu.be/4X5aIyBa0gs

Shit between that and the R5 1600X consuming 40 watts more than a 7600k under load, she really are taking people for fools...

Dunno about you guys but I'm going to buy myself a nice power efficient CoffeeLake with a standard GTX1070 and save money on my electricity bills.

>>61968065

hey rememeber yesterday when i showed Vega 64 was shit based on the gpu userbenchmark site

also every site everywhere is quietly avoiding the i7-7820X which is only 580 dollars and destroys the 10000 dollar threadripper 1950x (and every other ryzen) and also doesnt require a 330 dollar mobo which is the cheapest for TR vs only 200 for x299

combuterpase tl;dr with google translate:
>On Friday, ComputerBase was able to test a pre-production copy of the Asus Radeon RX Vega 64 Strix in the editorial department and compare it with the reference design from AMD. The result is clear: The first partner card with Vega is faster, much quieter and hardly needs more power.

>>61968417
>HBM from Samsung

at least some good news

>ANUS card is dogshit
water, wet.

>>61972007
More like even AIB partners can't make Vega worth buying.

Which is a shame since I like Sapphire cards.

File: bosh.jpg (26KB, 450x429px) Image search: [Google]

26KB, 450x429px

>>61969712
>mfw got a 470+480 for $150 each @ launch
>also got a pair of 6700ks for $200
>put a $10 led kit on each
>sold them to /v/ types for a massive mark up

File: Capture.png (226KB, 1303x659px) Image search: [Google]

226KB, 1303x659px

>>61968907

File: 1499437634755.jpg (8KB, 250x209px) Image search: [Google]

8KB, 250x209px

>>61968117
Don't forget to add
>stuck with limited overclocking because Vega BIOS is locked and unmodifiable
AMD is trash

>>61972288
>>sold them to /v/ types for a massive mark up

you're a retard. you sold them to cryptocurrency miners who probably ripped your garbage led kit off and laughed at how stupid you were

>>61972484
>miners
>earning anything
How long is this meme going to continue?

>>61971944
Literally the only company that makes HBMeme2
What the fuck did you expect

>>61972602
>Literally the only company that makes HBMeme2
who is Micron & SK Hynix for 200, Alex.

>>61971765
What?
Literally every benchmark the 1950x mood the floor with the 7820x except single core
http://www.cpu-monkey.com/en/compare_cpu-intel_core_i7_7820x-749-vs-amd_ryzen_threadripper_1950x-756

Go home Brian

>>61972632
Companies that don't make HBM2
Samsung is currently the only ones

>>61972636
DELID THIS,SU. DELID THIS RIGHT NOW!

>>61972654
They are making HBM2 soon though

Vegas saving grace may be low power embedded products like raven ridge.

May be.

If you can cut the voltage down low enough and keep clocks at 1000mhz while taking less than 10% performance deficit it could be the new embedded king.

But who are we kidding? It's Amd, and Raja.

>>61972636

>1 singular test
>almost 400 tests

also multi-core is a bullshit test for servers

Multi core mixed speed is a server orientated CPU benchmark. It measures the ability of a processor to perform 32 integer and floating point operations at the same time. This test is more appropriate for measuring server rather than desktop performance because typical desktop workloads rarely exceed two cores. See the multi core integer speed and multi core floating point speed for more details.


Quad-core testing is more viable because of gaming applications

>>61968065
>>61968142
>>61968445
It's an engineering sample running stock clocks with a higher power target so it doesn't throttle as much and its competitive with AIB factory OC'd 1080s. What more do you want from an engineering sample on alpha BIOS and half finished drivers?

>>61972441
Buildzoid and GN already found a way around that by using registry mods to increase power limit way past what the BIOS normally allows. Should be interesting to see what they can get out of Vega 56 now that they can put up to 380 watts through it.

>>61969123
Thing is that with the only extra cost being up front and the fact that a 600w will pull more wats than a 800w to power a 500w draw rig and have more fan noise why wouldn't you spend $20 more to get the 800? You would probrably even bread even once you factor in electricity costs after 2-5 years.

>>61972758
This is why I want to know how vega performs under volted and clocked, but I haven't seen anything like that yet.

>>61972910
Yeah, I just watched that video, but it's still sad you need over 300 watts for an amd gpu to be useful

>>61968117
Jesus Christ can't these fucking idiots do ANY RESEARCH before they try things and hit publish?

Core overclocking does not work properly on Vega launch drivers, and setting clocks too high will cause a silent performance regression. The only way to get a core overclock to work on launch drivers is to test performance with games after each attempt to check for regressions.

>>61972960
It depends. Vega 56, undervolted with an increased power target and HBM2 overclock will deliver consistently better than OC'd 1070 performance at ~240 watts.

At 300+ watts, Vega 56 should be trading blows with an AIB OC'd 1080 for it to be worthwhile, otherwise there is no point. Unfortunately, right now, its hard to test this kind of stuff because core overclocking doesn't work properly in the launch drivers so who knows how much of an actual working core overclock a power modded Vega 56 could take?

In theory, in the absence of the RTG driver team getting primitive shaders working, Vega 56 should be exactly as fast as Vega 64 at the same clocks and power target because both cards are front end bottlenecked by triangle throughput.

>>61972956
Gamers nexus done a fair bit of under volting

File: 1457920886570.jpg (72KB, 752x816px) Image search: [Google]

72KB, 752x816px

>>61973025
Now I'm starting to wonder what would happen if we drove a 1080ti to like 800 watts
Would we make a DeLorean?

File: attachment.jpg (50KB, 625x626px) Image search: [Google]

50KB, 625x626px

>>61971701
>6c/12t CPU consumes more power than a 4c/4t CPU.
Are you retarded?

File: 8Rch6JF.png (55KB, 1509x905px) Image search: [Google]

55KB, 1509x905px

>>61973087
I want to see the full voltage curve for vega, not just what they were able to min max.

>>61973165
Go buy a vega and post the results pls.

[spoiler]it's still horrible[/spoiler]

>>61973208
>spoiler tag

fuck off faggot

File: 1501890617131.png (88KB, 489x423px) Image search: [Google]

88KB, 489x423px

56 hurry up

>>61973208
hi /v/

File: 1497959662034.jpg (29KB, 480x562px) Image search: [Google]

29KB, 480x562px

>>61968088
>415W
>415W
>415W
>415W

File: RTG.jpg (132KB, 1280x720px) Image search: [Google]

132KB, 1280x720px

>>61968065

>>61969711
>1375MHZ memory
FIXED

>>61972441
>locked and unmodifiable
It's probably just using the same type of BIOS the Fury cards have given the similarities, and while you can't do anything crazy with firmwares that require signing, you CAN do basic things like mess with the voltage and power limits, same deal with unlocking if everything under the 64 wasn't lasered. Wait for this guy to make a statement

http://cxzoid dot blogspoot dot co.uk/2016/04/possibly-last-batch-of-custom-fiji-bios.html

With that said, I think Vega is more of an UNDERvolting candidate, since the Fury is extremely cool at stock speeds with better paste, and if it's a decent ASIC, you can generally OC and undervolt

File: mlzhff[1].png (39KB, 390x490px) Image search: [Google]

39KB, 390x490px

>>61972602

>Samsung
AMD reference cards are Micron though

Source, my Vega card

File: 1316839860781.jpg (22KB, 195x195px) Image search: [Google]

22KB, 195x195px

>>61968088

>>61968088
DELETE THIS RIGHT FUCKING NOW, I SWEAR TO GOD I WILL FIND YOUR IP, GO TO YOUR HOUSE AND I WILL MAKE YOU REGRET EVER CLICKING THAT POST BUTTON.

File: 1501510708307.jpg (690KB, 1263x1920px) Image search: [Google]

690KB, 1263x1920px

>>61970048
>even infodumping here
good jobu Wikieora

>>61970814
Bease 1440 is like 720. Pointless. The jump is from 1080 to 4k.

>>61972441
Only on reference cards because of self destruction heat problems. Once 3rd party comes out it would be unlocked.

File: kek.jpg (2MB, 2014x2695px) Image search: [Google]

2MB, 2014x2695px

>>61968088

File: perfwatt_3840_2160[1].png (48KB, 500x1050px) Image search: [Google]

48KB, 500x1050px

>>61971336
>Did you notice the better performing Nvidia cards in that chart beating Vega on FPS/watt even when Vega is not in turbo mode?
They do but every mode but the power saving mode is pants on head retarded. They should've just accepted that Vega can't compete with the 1080.
In the power saving mode it's like 8% worse performance and 10C lower load temps while only using 200w instead of 400w.

>>61977763
what the fuck are with these power state profiles? Does other cards have these autistic number of power states?

>>61977763
huh, that's actually pretty cool, 64 seems worth it for msrp money now

>>61977955
I think there is a switch on the card
200w seems damn fine for 4k ALUs, I see why miners want it

>>61968117
>A quick word on overclocking. We managed to run the Asus Radeon RX Vega 64 at 1,980MHz core and 1,000MHz memory, just beating out the clocks achieved by reference, and by doing so we added an additional 6.4 per cent performance in a trio of games. The downside of this voltage-induced performance is that power consumption rose to 532W at the wall.
>>532W
AAAAAAAAAAHHHHHHHHHHHHHHHHHHHHHH

File: 1406748318413.jpg (27KB, 400x400px) Image search: [Google]

27KB, 400x400px

>tfw going to get bullied for buying a vega even though i acknowledge the superiority of nvidia but i need an amd for driver reasons

>>61968200
Yeah, Pascal just has a voltage lock on it, Vega is power limited which is even worse.

>>61978342
>500w monster
>power limited

I hope OC meme dies in next 5 years with advent of 7nm.

>>61978520
this.

>>61968532
is it really that hard to figure out
the power difference is due to the custom cooling solution on the asus card