AMD Radeon Vega Frontier for 1,199 USD

>>60917936
WHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHA

>>60917936
>it's cheaper than gp104 Quadro
Noice.

>375W HOUSEFIRES

AYYMD IS FINISHED & BANKRUPT

AYYMDPOORFAGS CONFIRMED ON SUICIDE WATCH

>>60917936
>2 8-pins

holy shit

$1200 for a workstation Vega GPU?
Thats pretty big. Not too long ago a flagship workstation card could have been $3000.

>>60917995
I guess they want marketshare and want to push ROCm.

File: 1472497826735.jpg (91KB, 660x708px) Image search: [Google]

91KB, 660x708px

>>60917936
>Double 8-pin

>cheaper than quadro
huh

>>60917936
hahahahahahaha

>>60918014
Any AIB gp102 is also 2x8pin.

File: Radeon-Frontier-vs-TITAN-Xp.jpg (80KB, 1352x524px) Image search: [Google]

80KB, 1352x524px

>workstation card vs geforce on pro benchmarks
>not compare with quadros
>375w
WEW LAD

>>60917936
I was expecting something to bridge the gap between 480 and 1080. Who are workstation GPU's actually for and why do they spend that much?
CAD engineers? I know maya, udk all only use the GPU for view ports but the final render is done on the CPU.

http://wccftech.com/amd-radeon-vega-nvidia-pascal-gp100-performance-benchmarks/

>If we take a look at some benchmarks posted by PCPerspective a while ago, the Quadro P5000 (a GTX 1080 equivalent card) manages to score 152.92 in Catia. Vega scores 135.75 in comparison which is near the P4000, that is a cut down GP104 SKU. In Solidworks test, the Quadro P5000 scores 168.11 versus 114.88 on Vega. The score is even lower than Quadro P2000 which is based on a cut down GP106 die.

Quadro P5000 literally destroys 375W card in performance, how embarassing

AMD doesn't even dare to compare to Quadro P6000 or Quadro GP100, which would be even more embarassing the margin of defeat

File: 69f9979993d63791c84fdee415e942c505249c02056f0330732fa7ec80881829.gif (3MB, 476x341px) Image search: [Google]

3MB, 476x341px

>>60917936
This shit is slower than Quadros and draw more power. Vega is Fury X 2.0 confirmed.

>>60918089
>draw more power
Did you measure it from the rail, Huang?

>>60918106
>amd gpu
>2x8 pins
>high tdp
>aio option

>>60918055
>>60918075
get a load of this shill

>>60918139
Since when power connectors mean anything? Fucking AIB 1080tis have 2x8pins.

375w is beyond PCI-E spec. Wut.

>>60918160
>TDP 300w-375w

>>60918183
>>60918176

>>60918181
>trying this hard
your bad at it m8

File: AMD-VEGA-Geometry-Pipeline-1.jpg (101KB, 1066x600px) Image search: [Google]

101KB, 1066x600px

>Complaining about wattage
>AMD is claiming 50% improved IPC and 2x performance per watt

4096 cores @ 1600mhz is going to BTFO of nvidias flagship period

>>60918236
More than that, 375w card simply can't pass PCI-E certification.

File: power_maximum.gif (72KB, 400x1227px) Image search: [Google]

72KB, 400x1227px

AYYMD went full retard as usual and overclock the chip well out of spec because they can't compete in performance, just like the Fury X

>>60918296
Did you forget dustbuster?

File: power_maximum.png (30KB, 500x650px) Image search: [Google]

30KB, 500x650px

>>60918296
>not the updated version with pascal

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

78KB, 500x500px

>>60917972
>>60918181
never forgetti

>HBC
so it isn't HBM2?

>>60917974
That's nothing nigger my card has two 8 pins plus a 6 pin

>>60918361
It is.
>>60918359
Nothing beats 5800 ultra.

>>60918361
16gb of hbm2

Fury X 2.0

>>60918402
Fury X was 275W.

>>60918379
>5800 ultra
https://www.youtube.com/watch?v=qspdnAYiiug

HOLY SHIT!!!!

File: 1481545606774.gif (2MB, 320x240px) Image search: [Google]

2MB, 320x240px

>>60918430
>Fury X
>275W

>>60918430
This is my point, overclocked Fiji with HBMeme2 like we said 4 month ago...

File: GTX590Single.png (103KB, 565x750px) Image search: [Google]

103KB, 565x750px

>>60918379
Single GTX 590 is a 600w card on one PCI-e

>>60918461
That's it's rated TDP. Do you know the difference between TDP and power draw? If course not, fucking /v/edditor.
>>60918460
The original housefire.

File: 1497283935507.gif (2MB, 480x360px) Image search: [Google]

2MB, 480x360px

>>60918181
But hes right, you're a shill

>>60918473
That's dual GPU board.
>>60918469
Where the fuck did 50% more transistors go?

>>60918488
>Where the fuck did 50% more transistors go?
I bet in the HBM die or lies

File: PCspecs.jpg (123KB, 781x602px) Image search: [Google]

123KB, 781x602px

>>60918488
>>60918473
>>60918460
>>60918379
>>60918475

Only the devil knows this kind of suffering
>tfw the CPU is @ 4.9ghz with 1.554v too
You ever seen a system completely max a 1500w PSU?

>>60918504
I know you.
>>60918501
What fucking HBM die you fucking retard? Vega10 is ~500 mm^2 die you nignog. Its still 4096 ALU. Where did 50% more trasistors go?

>>60917936
>wccftech

People just don't know any better.

File: Image-15-768x1005.jpg (210KB, 768x1005px) Image search: [Google]

210KB, 768x1005px

>comparing a workstation graphics card to a consumer graphics card
God, this board used to be filled with people who actually knew about technology. Now it's just children arguing over which video game box can get them the most fps in cow-a-dooty.

Pathetic.

>>60918545
It's videocardz

File: 1495478151944.gif (1MB, 320x425px) Image search: [Google]

1MB, 320x425px

>>60918296
Running a fury full throttle on a synthetic furmark heater is kinda retarded.
When you thrash all your SPs it's going to get hella hotter because is has 4k of the damned things on a bigass die.

Comparing full load on raw geometry produces better real world results.

Plus: Holy fuck you can push 432 watts through a single die what the fuck

>>60918552
I blame AMD. They can't beat the QUADRO, so they bench Vega Frontier with the Titan X (pascal)

>>60918552
Even so the rated tdp is way beyond reasonable. If it don't deliver better performance than p5000 at least is a shit.
Also considering this whole arch, it seen fury 2.0. only time will tell if this is a reasonable going for gaming or pro work.
On paper fury was a beast but in practice it was just a heater.

>>60918183
>TDP 300w-375w
You do realise you are quoting Videocardz guesswork with no actual evidence, right?

>>60917936
Wait, isn't this the workstation card? Isn't this a great thing?
It's looking like AMD is doing a Vega for workstation and a Vega for consumers.

>>60918609
Yes, workstation Vega even comes first, even though launching gaymen ones is easier. Weird.

>>60918591
They should have compared it with p5000 because it will be on same league.

Anyway, Titan Pascal xp is actually almost the p6000, it's the same die, minus drivers.

>>60918617
That is no wonder. Perhaps Vega is too hot and perform below expected so another flop.

>>60918349
>>60918296
All Nvidia cards throttle heavily while running Furmark. 1080ti would consume 350w or more if it actually ran furmark properly

>>60918596
>2x8

>>60918636
You also can't go past 1.2v without modding since Maxlel.

>>60918638
Go shitpost somewhere else. Most cards sold with 2x8pins have TDP in the 250W range

>>60918045
Does it do cryptocoins?

File: AMD-VEGA-Rapid-Packet-Math-NCU-IPC.jpg (147KB, 1066x600px) Image search: [Google]

147KB, 1066x600px

>>60918580
Its on a smaller fab process it should run a lot cooler than a fury x
>>60918552
This is true but performance can be judged from it indirectly
>>60918524
Ayyy
>>60918469
Vega is a all new design it could be compared to Bulldozer to Zen

>>60918638
Any AIB 1080ti has 2x8. So?

>>60918609
>Wait, isn't this the workstation card?

Only when comparing it to gaming cards, and with a huge possibility that benches were skewed to favor it completely.

>Isn't this a great thing?

Not if you need a workstation tier card.

>It's looking like AMD is doing a Vega for workstation and a Vega for consumers.

It looks like AMD's marketing dept is just having another giggle m8.

It's best to just wait until release especially if the information is from wccftech.

>>60918046
Scientific simulations, for one. Engineering stress simulations. Anything that wants a lot of floating points crunched. Vega's party trick is perfect FP scaling with size. FP16 is twice as fast as FP32 is twice as fast as FP64.

>>60918617
If gaming chips are binned Vega, you need to bin enough chips to actually have supply. Low cost workstation card and late lower-end chips could mean Vega has very, very good yields.

>>60918718
It's also in new gaymacs pro.

>>60918734
That shit behind a display? God within a week the display will melt unless it's lead shielded.

>>60918748
Why?

>>60918748
Ryzen runs cooler than Intel chips, maybe AMD figured out power delivery.

>>60918770
You just add tiling and larger L2 for more efficient GPU. See Maxwell.

>>60917936
What?! 2x8 pins? Fuck that shit, I am not buying a new PSU for that

>>60918718
It does have very good yields similarly to Zen. The 14nm FinFET is fully fleshed out at this point. We will see a cut down version of vega on GDDR5 for RX 600 series of cards but that is more than a year out.
>>60918703
Work station cards also have to be very accurate. This is one of the reasons gaming cards are not ideal for them. Its also the same reason why in general work station cards perform much worse in gaming. Double precision performance is king in scientific calculations.

>>60918821
Vega10 is 1/16 FP64.

>>60918821
P6000 still owns the top gaming performance believe or not.

>>60918881
I think Titan Xp is better. Both are full gp102.

>>60918898
P6000
>3840 CUDA cores, 12 TFLOP

Titan X Pascal
>3584 CUDA cores, ?? TFLOP

>>60918934
I said Titan Xp, which is 3840 ALU one.

File: hitman.png (15KB, 622x581px) Image search: [Google]

15KB, 622x581px

>>60918934

Titan X (pascal) =/= Titan Xp = P6000

File: aaa.jpg (232KB, 1162x778px) Image search: [Google]

232KB, 1162x778px

>>60917972

>>60918960
Fucking nVidia naming schemes

So if this workstation card is $1200, does that mean the regular Vega 64 will be more like $599?

I just sold both my 480's to coinfags on eBay and made a decent profit. Limping along with an old 960 for now. Should I get a 1080 or wait for Vega? I don't have any brand loyalty, just want to make the best choice. If I have to use this 960 for more than a month I'll probably neck myself though.

>>60918176
Really doesn't matter, 150w each from the two 8 pins + 75w from the slot = 375w. Not like the motherboard knows the difference.

Just makes buying PSUs more difficult.

>>60919006
Agree

>>60919018
>If I have to use this 960 for more than a month I'll probably neck myself though.

RX Vega is at least 6 weeks away.

>>60919034
It does matter when you need PCI-E certification.

>>60918176
It doesn't matter at all, two 8 pins means it's still only drawing 75w from the board. Just makes it harder to find a PSU.

>>60918045
>--
Pathetic. even they know how embarrassing it is.

>>60919060
>Need
That's a funny word

>>60917936
The new iMac will probably be the cheapest way to get a Vega machine

>>60918782
AMD already has tiling, nvidia's is just better.

>>60919572
Current GCN tiles are too big.

>>60919594
I think they're supposed to be like a 1/3 of the scene or something, or divide the scene in 3s (so 9 tiles) or something. I vaguely recall 1/3 being tossed around.

>>60919657
It's 1/4 of the scene.

DOA

File: 1455235944048.jpg (8KB, 247x204px) Image search: [Google]

8KB, 247x204px

>>60919060
>need PCI-E certification

>>60918558
>videocurryz
Still don't know better

>>60919054
And another 6 months before you can actually get your hands on one

File: 79aa399ef02b0b0a7389ced2705271616509cfb509842077138fba45322c7394.jpg (64KB, 1280x720px) Image search: [Google]

64KB, 1280x720px

>>60917936
>$1271.60 Ameribucks for a fucking consumer video card

That's it. I give up. PC's are fucking retarded. I'm just gonna use ebay thinkpads for the rest of my life.

>>60920587
This isn't being sold as a consumer card senpai
In fact it is seriously undercutting Quadro's, though it isn't delivering the same performance

>>60920587
>Consumer video card
where do you get that idea?
the fact that it's being advertized is not fully representative of the direct market. they are marketing Epyc, and it's server grade.

>>60918349
hawaii still near the top
finewine™
>>60918473
isnt this total consumption though, having owned a hd6850 as my first gpu, no way it could pull 250w
>>60918504
the ssd temp and gpu temp seem to be switched :)
>>60920587
thats a workstation card for corporate, hedt use

>>60918296

I get so hard everytime I see how high the 295x2 power draw and temperatures are. It's like seeing the typical trashy slut with runny makeup.

>>60920587

>consumer

too much anime has rotted your mind

>>60917972
>still won't run as hot as a 7700K

>>60918055
It's not like they marketing and pricing it as an in-between card like the Titan or anything...

What the fuck is that power draw?

Why is everyone giving this card a hard time when it's nearest competition is almost $2000. It's just like the RX580 vs GTX 1070: a little slower but a bargain.

Half of you are like OMG AMD BANKRUPT THEIR $1300 CARD CANT BEAT A $1900 CARD!!!

>>60921864
It's a competitor to Titan. Radeon Pro SSG (actual Vega-based Pro card) is god knows when.

>>60921878
I'm confused then because they have labeled it a workstation card. I would thing it is meant to compete with the lower end Quadro cards.

>>60921894
No. It's like me ebin original cuck Titan.

>>60918782
Also:
>Cuck FP64 and use the transistor budget on moar FP32
>Move a lot of scheduling and stuff to software to save transistors and power (cpu power draw isn't counted in 'muh efficiency' autism contests)

>>60917972
>whats the matter goy? Are you a poorfag?

>>60921931
And P100 has FP64 back and it's hot as balls.

>>60920587
nvidia want $1200 for the titan and this is amd's answer to it

>>60920587
If you're this retarded you deserve getting shitty eBay thinkpads for the rest of your life

>>60917936
pretty cheap for enterprise card, quadros are what? $2k?

>>60921954
Isn't it also the only one (of the pascal range) that has legit FP16 capability too?

>>60918595
>Even so the rated tdp is way beyond reasonable.
except AMD slides say 225w
videocardz bullshit is meaningless for now

>>60922015
Yes.

>>60922035
Not just AMD slides, 64CU Vega is in the iMac Pro. No way you can put even 250w card into AIO chassis.

>>60922086
I forgot about that
plastic would melt at 370w

>>60922103
And it's still 11tflop, so it's not even heavily underclocked.

what are the chances stores even get a few cards before it gets pre ordered to hell by miners?

This is 290x all over again. At least we will see AMD profiting this quarter first time in decade.

>>60922129
>profiting this quarter first time in decade.
Mainly for EPYC.

>>60922137
in whole complex - ryzen, epyc, polaris,vega, last year they were VERY close just on 480 sales

I'd buy a bit of stock now, it's gonna fall after vega launch a bit, but will jump after financial report

>>60918703

FP64 isnt half FP32 peformance on vega. It does 8-16-32FP in that scheme. They just started calling FP8 half precision for some reason, that us were the confusion came from.

FP64 performance is much lower than you might expect.

>>60918634
They're sandbagging just like they did with Zen.
AMD realized that hyping up their products while releasing as much information as possible is a bad business strategy, since the opposition can prepare easily.

>>60923462
Mostly, yeah. Who the fuck did expect near perfect core scaling and fucking 65w octacore? Hell ANUS put one into laptop.

File: amdvega.jpg (156KB, 633x758px) Image search: [Google]

156KB, 633x758px

>>60917936
>slower than quadro in CAD
>slower than Ti in games
>$1200

>>60923475
>98/99% scalling
I still cream whenever I think about it.

>>60923483
It's a Titan. From AMD. What did you expect? Vega-based Radeon Pro SSG comes later.

>>60923508
>7.90 SPECINT2006 core scaling across 8 dies and two sockets
Absolutely marvellous, sasuga Clark, Keller and Papermaster.

>>60923524
CERTIFIED

>>60918251
Only if they're drawing the power from PCIE. Chances are it will have dedicated power support for the cooler, so they will still pass it.

>>60923528
SHIT
Also IBM nazi scientists will also deliver on new nodes. The future is bright and devoid of jews.

File: umadlet.jpg (17KB, 368x311px) Image search: [Google]

17KB, 368x311px

>amdrones start talking about cpus in gpu thread to shift the topic away from this poojet embarrassment
every single time

>>60923558
WRECKER
>a timeline where Intel will be the inferior choice in industrial/consumer/professional/scientific workflow
Feels good, based, absolutely based.

is it true that their gay man cards are coming out this July?

Why are you stupid fucks comparing it to a quadro p5000/6000?

Those cards are twice and three times as expensive

>>60923599
Leakers said June/July, but it's most likely going to be July/August, just like it was with the Fury line.

>>60917936
Will these get nabbed by crypto miners as well?

>>60923648
The high end ones won't, not worth it.

>>60918761
Apple,as usual, fucked in the choice of processor, and put on it a fucking housefire Xeon instead of a Epic or Threadripper.

>>60923613
Shitposters, retards and nVIDIA shills.

How many MH/s can it do?

>>60923743
megahashbrowns? probably a lot to replace your green deep fryer

>>60923702
t. amdrone
>why do compare "professional" card to professional cards
>i-it's cheap

>>60923930
>AMD is the budget option this time
Woah so surprising!1!

>>60917936
Is this the one with the m.2 slot?

>>60923944
except it's embarrassingly slower in professional work and people need cuda

>>60923743
Not enough to make it commercially viable vs 580's

Will this thing make Avatar 2, 3 and 4 release earlier?

>>60918045
>Titan
>GeForce
But that's wrong. Titan is, and always has been, a poor man's Quadro, supporting a lot of those same pro features.

>>60923959
>2 times cheaper
>less than two times weaker
>embarrassing
OK

>>60923985
It's a fucking failure for its HUEG die size.

>>60924002
How do you think your mom feels?

>>60924017
>amdrone out of arguments
wew

>>60924046
You already admitted defeat with >>60924002 though.

>>60924048
>it's twice slower on big die year later
ok amdrone

>>60924065
I don't give a fuck about die size. Price and performance is all that matters.
It has better price/perf ratio but worse perf, as per usual for AMD. Nothing to point out after all those years.

>>60923414
8 bit stuff is integer, not fp (half precision is fp16)

it might prove useful for a lot of image and data processing stuff so hopefully it's enabled on the mainstream/gaymer cards too (don't jew us like nvidia pls)

>>60923698
No memebolt. Apple is retarded though.

>>60924065
>big die
Vega10 is about as big as gp102.

>>60924279
and slower

>>60924298
Could you please show me benchmarks?

>>60924248
>memebolt
It isn't a issue. The controller doesn't need to have a intel processor attached to it to work, and, as I know, the licencing doesn't forbide to use it with AMD processors.
I guest that Tim Cook is just an idiot.

>>60924333
Or Intel gave them Xeons with severe discounts, like the good old days.

>>60924345
I suspect this, as it helps Intel push that "we're the premium brand" marketing meme

>>60917936
>blue and yellow
nah thanks, i only do my graphic cards on stealth , red or white.

>>60925150
i'm sure that people who buy workstation cards care what colour they are

oh wait, they don't

>>60924345
This too. Probably Intel has Apple grabbed by the balls.

>>60917936
>Thanks for waiting

>>60925255
>workstation card
why do they compare it to titan then

>>60925516
Titan is also ebin pseudo-prosumer cards for people too poor to afford a Quadro.

File: tidus.jpg (30KB, 475x343px) Image search: [Google]

30KB, 475x343px

>>60917936
>watercooled "workstation" card
I see where lincuck took his "watercooled OC servers"

What's the market for this? Buttcoin? Machine learning?

>>60925538
They compare it to titan because quadro shits on it so bad it's not even funny, hence the price and the vague marekting spins.
Nvidia never marketed titans as workstation cards.

>>60925538
But p4000 is 900$ and faster than Vega FE.

>>60925574
>They compare it to titan because quadro shits on it so bad it's not even funny
Quadros are no different from usual gayforce cards, just speshul drivers. Vega FE is literally AMD's Titan, anon. Actual Pro WX cards and SSG one are coming later.

>>60925577
Because Vega FE is not a fucking Quadro competitor you moron. Pro WX cards are coming LATER, along with Pro SSG one.

>>60918045
So, the Vega equivalent to the Titan is roughly 25% faster while being priced similarly?

Can we extrapolate from this that the gaming GPUs will also be faster than their Nvidia counterparts while being priced similarly? If so, AMD are fucked. Nobody's going to buy their GPUs if they cost the same as Nvidida GPUs. Even if they are slightly faster.

>>60925662
>Nobody's going to buy their GPUs if they cost the same as Nvidida GPUs. Even if they are slightly faster.
And that's totally okay, no one bought Evergreens, why would anyone even buy anything but nVidia? It's The Way It's Meant To Be Played, goyim.

>>60925689
I'll buy a Vega if it had slightly lower performance than a 1080ti for slightly more than a 1080ti because I want a faster GPU but without giving nvidia any money.
I've simply been let down too much by them and I didn't even get caught with the Fermi or 970 shenanigans.

>>60925721
You NEED to buy nvidia. We all want for AMD to leave consumer GPU market, anon.

>>60925725
Because that worked out so well for us in the CPU market didn't it?

>>60918045
It's not a pro card. It's a content creation card just like Radeon Pro Duo. It gets handiy beat in any real "workstation apps" by GP104 Quadros because it doesn't have application profiles for CAD yet. The WX Line is intended to be workstation specific cards.

nVidia rates their board at 250W; AMD's is just spitting out the PCIe spec, 2x8-pin for the water cooled = 150W+150W+75W (PEG)=375W and 1x6pin+1x8-pin air cooled = 150W+75W+75W (PEG) = 300W. Vega may very well be a piece of power hungry shit but it's pretty misleading to use charts like these, especially when the author doesn't make a proper distinction between FP64 and FP16, states that HBM2 is HBC and compares workstation applications on two cards that don't even have workstation profiles on them.

There won't be an affordable Vega until end of July anyway.

>>60925689
THIS.
For the past 12 years I've just been refreshing my Nvidia GPU yoy with no issue whatsoever.
I don't even look at benches or prices I just order whatever Nvidia launches as their flagship

>>60925823
It's a fucking irony, pal. Trying to sell hardware to gaymen faggots is hard. Like, fucking THERMI outsold Evergreen.

>>60918251
you keep saying that phrase

i dont think it means what you think it means..

>>60925840
Aircooled Vega FE is also 2x8pin.

>>60925840
>It's a content creation card just like Radeon Pro Duo

How appropriate, because it is a paper release with 0 availability like the Radeon Pro Duo.

>>60925896
>0 availability
>preorders just started
Wut.

>>60924240
>8 bit stuff is integer, not fp (half precision is fp16)
>it might prove useful for a lot of image and data processing stuff so hopefully it's enabled on the mainstream/gaymer cards too (don't jew us like nvidia pls)
Can you raytrace render on fp16?

>>60917936
Is that 2x Fury X with glue like threadripper?

>>60925968
No, Vega has almost nothing to do with Fiji besides SIMT width. Scalable GPUs are coming in Navi though.

>>60923483
Wow, did they already released gaming benchmarks? Where can I see them?

>>60926302
I'm an expert circuit architect, trust me when I say it'll barely perform over a 1080

>>60917972
It's a workstation card you fucking mongoloid.

>>60926401
Not even that. It's a Titan. From AMD.
>>60926361
Mr. Expert IC architect, where did 50% increased transistor count go?

>>60926361
I'm an expert gay detector, you gay.

>>60926361
Oh, okay then.
Thank you, not buying this shit.

>>60926418
>Mr. Expert IC architect, where did 50% increased transistor count go?
On useless jokes like HBM

>>60918046
>Who are workstation GPU's actually for and why do they spend that much?

They started for graphics things like cad or 3d programs or even rendering. Then the other people saw how much power they had and started moding their code to use them.

Password cracking engineering science number crunching neural nets or ai stuff. Basically anyone who needs a shit load of numbers crunched and can mod their process to use the kind of math a gpu handles.

>>60926361
hi thanks me too
I think electrons in AMD GPUs have a more negative charge than the electrons in NVidia GPUs
this is really bad :S:S:S:S:S::S

>>60926418
Titan isn't a workstation card, it's a consumer card for manchildren who think they're cool for buying a """pro""" card to play overwatch on max settings.

>>60926551
>memory
>anything to do with chip transistor count
Also was GDDR5 also a joke when RV700 came?

>>60926579
Vega FE is exactly that. It carries no Pro branding, neither it's WX series.

AMD need to get their GPU house in order. Start again from scratch just like Zen.

>>60926676
>start again from scratch
Really bad idea and GCN is actually good. Also
>consoles

>>60926602
What the fuck?

http://pro.radeon.com/en-us/frontier/

They clearly outline for what purpose this GPU is for.

>machine learning
>real-time visualization
>physically-based rendering
>VR
>video editing
>animation
>post-production
>Game development

It's not a fucking Titan.

>>60926724
Also at the bottom you can clearly see they're segregated into 3 distinct brands.
RX, Pro and Instinct.

Guess which one FE is?

>>60926749
Instinct?

>>60926760
...No.
Those are without video output cards, like Tesla.

>>60926760
consumer =rx
pro = FE
instinct = pure compute

>>60926724
Why the fuck it's shit in CAD though?

>>60926786
The fuck if I know.
Software support? No AMD profile yet?

>>60926800
Also why the fuck it's rated at 300w when it's in fucking gaymac? Whoever made that chart (that animal also listed FP16 as FP32) should kill himself.

>>60926786
isn't it worse than current gen? meaning something is wrong

>>60926724
>>It's not a fucking Titan.
The original Titan was actually quite usable for all those, besides ML, that wasn't important then.

So either it is a titan with normal drivers, but is trying to bring back the same market back to <$2000 it was before Nvidia gimped the titans.
Or it's a very fucking terrible card with pro drivers.

>>60926820
because retard counted x2 8 pin as 300w

>>60917936
So what does this mean for the average consumer gaymer pleb (i.e. me)?
How much would a card that could match or better a 1080 TI cost?

>>60926831
raja said that consumer vega will be better at games meaning it's pro drivers

>>60926820
They used pins lol, apparently LCS and air have the same clocks, but the other one is 75W higher? Makes no sense, and some sites don't even have TDP listed, still set to TBA

>>60926840
It means nothing and buy 1080ti you gweilo scum.

>>60926840
300=>450=>600-650
they will undercut nvidia in top bracket with slightly better performance

>>60926820
because 2x 8 pin = 300W and 75W over pci-e
it doesn't mean that it's going to use that, it's just the board design
it probably means the air cooled card has 8 pin + 6 pin since it's listed as 300W total

>>60926845
Well that Pro card was quite impressive in Sniper Elite and RoTR.
>>60926847
What? Who the fuck measures TDP like that? What, every 2x8pin AIB card is now 300w?

>>60926847
what it means is that those benchmakrs do not mean anything yet
when is release date?

>>60926845
Then the terrible performance in CAD compared to cheaper quadros is all down to software efforts.
There's a lot of pro software out there, hopefully AMD didn't just port 3 pieces of software for their GPU

>>60926874
well, radeon pro duo is also impressive if you count only avg fps

>>60926862
The 1080 Ti currently averages at 690 freedom units on newegg. How much would a consumer-vega cost? About 600-ish?

>>60926868
Air cooled card clearly has 2x8pin as seen on renders. Looks like the dude that made that list is fucking retarded.
>>60926877
Late June for FE card. Gaymen one launces on SIGGRAH but you'll have to murder thousands of cryptocucks to get one.

>>60926874
>Who the fuck measures TDP like that?
Someone dumb, and apparently the aircooled card should be 8pin+6pin + 75W PCIe.

File: IMG_1576.png (860KB, 841x720px) Image search: [Google]

860KB, 841x720px

>>60921781
You're right. It's going to be even hotter.

>>60926900
Raja said every FE is 2x8pin. 8+6 one was a devboard.

>>60926895
probably fury x launch price for 1080ti performance

>>60918251
Source please...

>>60917936
blunder of the century

>>60926918
he also said it doesn't actually need that much, it's for headroom but will know later for sure

>>60926896
>you'll have to murder thousands of cryptocucks to get one

It will be too expensive and no one is dumb enough to grab loads of them without knowing their hashrate.
480/470 get very close to the HBM1'd Fury cards for significantly less money, that's why its popular with miners

File: Screenshot_20170616_164659.png (41KB, 1280x336px) Image search: [Google]

41KB, 1280x336px

>>60926874
Meanwhile


Overclocking to the moon confirmed?

>>60926933
It if actually can overclock Tahiti-style it's gonna be a monster.

Raja is crashing Radeon with no survivors

>>60926949
in ruslands even 1080ti are sold out everywhere
cheap electricity offsets the cost

>>60926963
probably not
AMD recently simply pushes chip to the limit as is thanks to new techprocess
there is no need to OC

nvidia OC throttles as hell btw

>>60926971
Wut.
>>60926960
Who knows? Tahiti was a good overclocker.

>>60918045
>Titan is overpriced
>price Vega FE even more
>with slighty better numbers
HEI ITS GOOD VALUE GOYS!

>>60926960
I'm not hoping for 2000MHz, but even 1800MHz would be really nice.

>>60927012
Do you also buy GPU's based on colour?

>>60927007
titan can't do anything besides gaming this oen can

File: 1450670919290.png (6KB, 358x291px) Image search: [Google]

6KB, 358x291px

>>60926920

>>60926992
>Wut.
everything except for 1050s and 460s sold out

What the fuck is Vega? Why can't RTG just show us the arch.
>>60927046
What the fuck.

>>60927029
numbers just shows its slighty faster in those other task, i bet they are comparing cutdown titan as well as using the AIO Vega.
If they compared full titan xp 2017 vs Air cooler Vega Fe, the numbers would be identical while titan would be cheaper.

>>60926971
Maybe they only imported 10 1080ti's into the country so it means nothing if they sold out

>>60921864
>a little slower
The GTX 1060 is about 50% faster than the RX 580. The 1070 is practically twice as fast.

The 1070 is now a low-end card and AMD still doesn't have a proper response for it.

>>60927066
believe it or not GPUs sell well here even for 50-60k RUB
every shop got at least 50

>>60927080
Dude hands off the crack.

>>60926896
>Air cooled card clearly has 2x8pin as seen on renders. Looks like the dude that made that list is fucking retarded.
hmm
cooler rating perhaps?
but the 375W for water cooler seems too little.
yeah probably whoever made that was dumb

>>60927012
I think the headroom is just to avoid throttling, not for overclocking

File: 1490121142888.jpg (21KB, 278x357px) Image search: [Google]

21KB, 278x357px

>>60927080
>The GTX 1060 is about 50% faster than the RX 580.

Pro tip
>gaymen Vega cards will have 8/4gb of HBM
>price will be similar to 1080ti/1080

>>60925721
I bought a 480 even though I had a 970 simply to support AMD and spite Nvidia. I've had to deal with so much bullshit from Nvidia, yet my AMD has given me zero issues.

>>60927099
300 one might, 450-600 bracket will be 8Gb

>>60927120
300 one is late 2017 at best.

>>60927087
>>60927095
It's true though. If I'm wrong, prove otherwise.

NVIDIA has not had any competition to its Pascal lineup, which is sad to say the least since it's been more than a year since Pascal's release.

The 1060 is poorfag tier, the 1070 is low-end, the 1080 is mid-range, the 1080 Ti is high-end, and the Titan X is enthusiast range.

NVIDIA does not have competition, and from the looks of it, it definitely won't have any.

>>60927127
so there won't be 4Gb card then

>>60927147
Hands off that crack.

>>60927147
>It's true though
(you)

>>60927155
No, it'll use GDDR is some kind I guess. Samsung's low-cost HBM is not ready yet.

File: 1483163611712.jpg (16KB, 680x510px) Image search: [Google]

16KB, 680x510px

>>60927080
>>60927147

This is the worst one I've seen in a while.

>>60927179
didn't they start production early june?

>>60927120
Sure they will price the 1080 competitor 50$ cheaper. But thing is nvidia could literally drop 1080 price to 300$ and still make good money. They are really far ahead, while also using cheaper components, smaller die size..

>>60927198
What? Never heard about that.

File: Palpatine_Laugh.jpg (5KB, 300x190px) Image search: [Google]

5KB, 300x190px

>>60927202
>But thing is nvidia could literally drop 1080 price to 300$ and still make good money.
Whoever drops the most price will receive my dollarydoos
>mfw brand loyalty

>>60927202
>NVIDIA
>cutting their margins
Heh.

File: rx-580-performance[1].png (103KB, 616x427px) Image search: [Google]

103KB, 616x427px

>>60927147
Just

File: 1471598081512.jpg (473KB, 1920x1200px) Image search: [Google]

473KB, 1920x1200px

>>60927202
someone must invest 300B into maxtor and save us from this misery

File: Average[1].png (50KB, 1295x1331px) Image search: [Google]

50KB, 1295x1331px

>>60927275

>>60927202
Nvidia can drop prices not because they have been out longer, but because they shifted from selling silicon that should have been 1080's as 1080Ti's. They are just happy to keep slightly ahead of the competition and sell at premium prices.

>>60927306
1080 and 1080ti are separate dies, ergo gp104 and gp102.

>>60927318
he means that 1080ti was ready for a long long

time they just didn't sell it for marketing reasons
they'd make twice as much if they sold 1070 for 1060 price, kill next year upgrade market though

>>60927318
Only because they gutted their consumer cards of compute with the GTX 580 to make up for the terrible efficiency of their architectures. They kept doing it to offer their customers less for similar or increased prices with each generation. No wonder they suck at mining and DX12.

>>60917936
26.2 single precision teraflops for little more than a grand.

Good luck getting one before the minerfags rise the prices to 3k.

>>60927365
I wonder when will novideo go back to dynamic scheduling.

>>60927377
It's 13tflops FP32 and mining uses int8 you moron.

>>60927399
this includes ethereum?

>want to mine eth
>have 980ti
i don't think it's worth it

I WANT YOU GOSH DARN CRYPTO KIKES TO STOP DRIVING GPU PRICES UP THE FRICKING WALL!!!!!!!!!!!!!!!!!

ROOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO!!!!!!!!!!!!!!!!!!!!!

If AMD GPU's go up in value anymore I think I will sell my RX 480 and put it towards VEGA.

>>60918045
> --

Who would buy this shit when you can get an AR 15 for half the price?

>>60927462
What the fuck is AR15?

>>60927476
A gay male escort that performs scat

>>60927476
yuropoor detected

>>60927515
What?
>>60927498
Sounds like shit.

>>60927416
Mine ZCash, then trade it to ETH around the middle of next month, or alternatively when BTC reaches it's low around august, trade it to that.

>>60927528
would i be able to mine off a 770 i have laying around as well

980ti + 770

>>60927528
oh i also don't pay for electricity as it's included in rent

>>60927534
Depends on a variety of factors, electricity, your faith in altcoins, and a few others. I'd check what hashrate the 770 can get you for how many watts.

>>60927552
Well then hook it up, you got nothing to lose except some deprecated hardware.

>>60925689
It's retarded, but nobody bought AMD when they were faster in the past. I don't see how anything should have changed. Consumers are still the same retards they were back during the Cypress vs Fermi days.

File: 20170611_130628.jpg (1MB, 3264x2448px) Image search: [Google]

1MB, 3264x2448px

Will Volta or Vega have any cheap cards for poorfags like myself? Like some 1150ti?

>>60928573
Vega11 is later. Probably Q1 2018.

>>60928588
Yeah but it'll have any cheap cards?

Around 200?

Volta or Vega any of those

>>60928613
We know fuckall about Volta lineup. At least Vega11 was on fucking roadmaps.

>>60928635
Yeah don't know if shoud buy a rx550

>>60928671
No you definitely shouldn't.

>>60928695
So should I wait I'm a Poorfag from Argentina we got good wages but the fucking inflation here is unbelievable Mexicans got them cheap but their wages are lower than Haiti

I could buy a rx560 or a 1050ti

>>60928722
Get 1050ti then, nigger.

>>60928770
OK it's cheaper but isn't the 560 better?

>>60928798
Nah. 1050ti is better on average afaik.

>>60928810
K thanks you sure a 1150ti isn't coming right?

New to /g/ didn't feel like asking plebbit

Also should I buy a 8350 or go for cheapest ryzen or just an i3?

>>60928847
Get g4560.

>>60918014
>>60918039
My 1080ti also has 2x 8 pin
Why is this a problem?

>>60928881
It's not. Some faggy website measured Vega's TDP by count in PCI-E power.

>>60927275
1060 wins 3games, 580 wins 3 games. Performance is literally the same.

>>60928847
it will be called 2050ti most likely.