Vega - Air Cooled Model Costs $1200 US, Liquid Cooled Costs $1800 US

>currytech

>>60917900
>$600 for a chunk of copper, a shitty pump and a radiator

Wew lad

>300W & 375W HOUSEFIRES

AYYMD IS FINISHED & BANKRUPT

AYYMDPOORFAGS CONFIRMED ON SUICIDE WATCH

>>60917960
Dunno what's in those $600, higher clocks? Better bins? Some other disabled features?

Either way, it's implying Vega can overclock, AMD hasn't had good OCing products since Tahiti

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

80KB, 1352x524px

>>60917988
Would be a problem if it wasn't faster :)

>>60917993
This is a workstation card, not a consumer part.
The price is comparatively very low for the market segment.

>>60917988
Aircooled P100 is also 300w. So?

>>60918008
>titan xp
Where is the Quadros?

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.

>>60917993
I'm betting AMD counts TFLOPS using base clocks, which should be the same on both cards, but the liquid one has higher boost clocks.

>375w TDP
It's beyond PCI-E spec. Somebody is fucking retarded.

File: 2017-06-16-00-04-24-1251672671.jpg (4KB, 144x144px) Image search: [Google]

4KB, 144x144px

Any don't they put Ryzen into Vega?

>>60918008
Fury-X had 30% more single precision FP32 FLOPs than Titan-X, but Titan X performed 10% better in games.

Vega only has 9% more FLOPS than the Titan XP 2017. If the same FLOPS to FPS scaling in assummed as last-gen for each company, that puts the Titan XP 2017 31% faster in games than Vega, which puts Vega on par with the 1080, but not the 1080-ti.

>>60918227
>If the same FLOPS to FPS scaling in assummed as last-gen for each company,
But that's stupid, especially considering both companies have different architectures.

>>60918227
Vega has much beefier front-end. You can't apply flops scaling from Fiji.

So is this using workstation or RX drivers?
Or is it just AMD's version of Founders edition?

>>60918322
It's more like OG Titan (the cuck one).

>>60918340
But all Titans besides the first Titan are cuck Titans.

>>60918010
AMD is always pulling shit like this, cherry picking what there competition is. Makes them hard to trust.

>>60918350
The first titan is THE cuck card, since novideo nerfed it's FP64 perf.

>>60918388
Wasn't the first Titan (Kepler) the only Titan with FP64, all the shit later, Titan X, Black, Z whatever all had gimped FP64

>>60918408
>Titan X, Black, Z

Why are your prosumer cards named out of some DBZ fanfic?

>>60918431
>company founded by chink
You got your answer.

>>60918070
They'll be used for gaming benchmarks. Kek amd never change.

Very non-aggressive pricing, like they're implying Vega is faster than Titan XP

>>60918010
>This is a workstation card, not a consumer part.

Where's the double for, and does it have ecc? Will xfire benefit industry standard software yet?

>zero sauce on TDP
>make up numbers based on power connectors
>people actually fall for it
Fury X had dual 8 pins and that was 275 watts.

>>60918499
>275 watts
nah, moar

>>60918499
P100 PCIe has dual 8pin and that was 250W

>>60918499
Vega is watercooled tho

>>60918532
And? So was Fury. And blower version exists.

>>60918532
Why are you ignoring the blue version that's clearly aircooled with a blower?

>>60918515
hello /v/*ddit

>>60918539
There was no blower or fan-cooled 64CU consumer Fury card, though. I really hope Vega fixes that for those who like silence at idle.

>>60918588
Fucking /v/eddit can't even answer me when did 50% more transistors in Vega go. Fucking hell this board is getting worse and worse.

>>60918588
>blower
>silence
Gonna have to wait for the MAX GAMING ULTRA PWNKILLSTREAKHEADSHOT with ZERO(0*2) dB QUADRUPLE FANS(bearing not included)

File: 1444602501945.jpg (65KB, 638x357px) Image search: [Google]

65KB, 638x357px

>amd
>efficient

literally WHO GIVES A FUCK

WE KNOW VEGA IS SHIT

NVIDIA WINS ONCE AGAIN

COOL

>>60918618
Zen says hello. TeraScale2 says hello and shits on Fermi all day long.

>>60918070
Titan xp uses the same die as p6000 bit perform worse due to greedy drivers that make it a Quadro and cost a lot more.
It also means this hot furnace can't compete with p6000 but at least is cheaper.

>>60918672
Not the same drivers>>60918075

>>60918408
Both Pascal Titans are same die, but the first one, same as ti, are the ones binned with fewer cuda cores. Titan xp is roughly equal p6000 die witch is the full die.

I've gsync screens anyway, fk amd

I'm seeing double rate FP16 support there, not even quadros have that, that's probably where the price comes from.

>>60918352
That's called segmentation and it's basic marketing.

Two threads for this. Poojeeta is running amuc.

>>60917900
>375W TDP on a watercooled card
That's outside the PCIe spec.

>300W TDP on a blower card
Impossible.

I'm assuming 'TDP' is a typo and they meant maximum possible board power limit here.

Otherwise, we have to work out exactly how AMD halves the process size, keeps the shader count the same and GAINS 100W of TDP. That would be a legendary failure.

>>60918974
>legendary failure
Nothing beats 5800 ultra.

File: amd_radeon_r9_295x2-100259790-large.jpg (34KB, 580x433px) Image search: [Google]

34KB, 580x433px

>>60918974
>That's outside the PCIe spec

Not the 1st time, pic related

>>60918974
I'm looking at both sabrepc and scanuk and neither specify TDP

300W for a blower is near impossible to cool.

>>60919019
That's a dual GPU board.

>>60919045
Yes and it is outside the PCIe spec

Anyway, 375w is pure bullshit.

>>60919088
If it's overclocked to the moon it wouldn't be.

>>60919134
How the fuck would you overclock it to the moon with longer bibeline?

>>60919149
The fuck are you talking about, a longer pipeline would imply higher clocks.
Thing is, we don't know what Vega's maximum clock is.
We know the chips are currently around 1600MHz, but what is its wall? 1700, 1900, 2000?

>>60919170
Okay we actually know fuckall about Vega.

>mfw AMD ends up trolling everyone and Frontier Edition comes with consumer drivers and not pro ones

File: Herc_is_Disappointed.gif (950KB, 320x234px) Image search: [Google]

950KB, 320x234px

>>60917900
>1800

pls be retail price gouging

RX Vega
-FP16 gutted to 1/4
-Higher clocks
-aircooled, only AIB models available
-8GB HBM, no ECC
-275-300W
-$600

>>60919298
That's cheap for workstation chips. "Workstation" GPUs are the Xeons of the GPU world.

>>60919339
No they're not, "Workstation" GPUs are the Quadros and Firepros.

THe Xeons of the GPU world are Teslas and Instinct cards, the ones without display outputs

>>60919408
>workstation Xeons are not a thing
End yourself.

>>60918352
It's obvious to see what parts are for what

>>60917900
... what's up with that GPU render? where is the blower fan? or the attatched radiator for liquid cooling?

>>60917900
>misleading title
>>60917988
>obligatory BTFO post

Left out the crying wojack and falseflag post. 2/10

>>60919019
I just sold mine but damn did I love it. Everything on ultra always over 100fps at 1440p.
Doing an ITX build and it is bigger than every case I could use so...

>>60918184
Nice idea frodo

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

156KB, 633x758px

>>60917900
wait

>>60919327
>275-300W
will you stop memeing it? the thing is in AIO Apple product, AMD slides show 225w

worst case it's 250w.

It might be worth it for some users depending on what it can do. If the new memory controller can let it do gpu raytracing without being limited to the size of VRAM it will be very useful. At the moment GPU renders need to be able to fit the entire scene and every texture inside the GPU VRAM. Creates quite a big limitation for large scenes.

>>60919327
the 300 and 375w is just 2x8pin+75w from the mobo max spec. Who really thinks AMD will go with another blower 300w 290x style?

>>60923159
just like poolaris slides showing <100W

>>60923208
they were 120w, recent batch 480s are under 100w

>>60923206
raja on fin. analyst stage said it's just to give headroom

>>60918227
fury x had a big proble of under utilized rops
vega with the bigger front end and the proper gating wont have that

>>60918008
>190 GFLOPS FP32
>Titan XP
Why is it so garbage?

>>60923208
You won't believe it, but there is full 95w Polaris10.

File: poo.png (312KB, 595x654px) Image search: [Google]

312KB, 595x654px

buy my gpus

Soo... They are speaking of AI and machine learning on the official site. Where are the Tesla vs Vega benchmarks?

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

78KB, 500x500px

>>60918618

>>60925151
>amdtard resorts to 7yo meme

File: 1482004865624.jpg (17KB, 251x249px) Image search: [Google]

17KB, 251x249px

>tfw have to buy Vega because RX 580s are nowhere to be found

>>60918008
So id be paying about a grand more for roughly 30% better performance? nah

>>60925171
[something something EVGA 1080]

File: lmao.gif (353KB, 500x357px) Image search: [Google]

353KB, 500x357px

>>60918008
>titan xp
>CAD
amd, poojets and poojettesses

>>60925280
>nvidia is responsibe for evga cutting corners
kys

>>60925282
This Vega is Titan competitor.

>>60925296
>designed so badly that even evga can fuck it up
holy fuck nvidiots

>>60925319
evga didn't put thremal pads on VRMs you retarded streetshitter
literally every card has them

>>60925333
VRMs can handle these temperatures. Shitty novideo PCB design is shit. Deal with it.

>>60925333
why are you calling me a streetshitter if i'm running a 1070?
I'm not gonna defend a horrible company just because i bought their product which just happens to be better than what AMD had to offer
why do you have so much brand loyalty?
why are you so militiant for nvidia?

>>60925349
nVIDIA has some insanely fanatical army of fanboys.

>>60925339
>Shitty novideo PCB design is shit
>no problems with orher manufacturers
you streetshitters are utterly delusional
>>60925349
if it was nvidia's fault it would affect all vendors

>>60924183
Someone is getting fired

>>60917900
>ITT: People conveniently forget that AMD literally stated publically that Vega Frontier edition was meant as a workstation card, and consumer Vega would be following fairly close behind
>Who am I quoting?

>>60925349
>"running" a 1070
>check my sweet rig, brah
>my rig runs anything I throw at it my broseph
>come check out how sweet Overwatch looks man-ayys
>I swear porn looks better on it, check it out broman
>fuck, that's hot, isn't it bronii-chan
>w-want to make out...? Bromageddon?

>>60926869
Kek'd harder than I should have

>>60926869
>bronii-chan
Lost

>>60918070
They cost a lot more, hence it would be a dumb comparison

>>60927005
p4000 is 900$.

File: amd2.jpg (52KB, 568x612px) Image search: [Google]

52KB, 568x612px

>>60927183
but I's no HBM

>>60918184
Actually that would be pretty neato but combining a 140watt ryzen and 300watt+ full Vega into a 500 watt house fire is too much.

16gb hbm2+ shared memory plus ddr4 sounds nice tho

>>60918008

This is the very first bench leak, saw it earlier

Vega beating generic Titan XP as it is

Excellent result and meets expectation, not even factoring in HBM2 over GDDR6

>>60918496
>GPU
>ECC
really makes you think

pricing makes this nice

driver optimizations later down the right might help or naw?

>>60928513
If you're using it to crunch big numbers that need to be kept accurate, like for scientific research, it makes sense

>>60926869
>Lebro James
>Brone-plus-one
>xbro 360
>Xbrone
>Brosemary
>Bro J. Simpson
>Strickland Bropane
>Brocahontas
>Game Brover man

>>60928532
Pro driver does work for gaming but is about 3 months behind.
AFAIK you can install Crimson drivers on workstation cards no problem.

This is gone over on AMD's website somewhere if someone wants to look it up.

$1200 isn't bad at all for a card this powerful that supports SR-IOV. Though I'm sure it'll be slower than a 1080Ti without the proper gayming drivers.
$1200 was basically my guess price of "I don't think it'll be as low as $1200, it'll be $2000 probably, but AMD has surprised with pricing before..."
I told myself I'd buy it at $1200 so idk I guess i am. Maybe two.

>>60928983
You must be retarded to think any pro-level GPU has ECC memory. The most they have is ECC support, and HBM2 provides for that.

>>60929266
>SR-IOV
If that's enabled and supported it'd make my "live on linux and do windows work and gayming in a VM" plans a lot cheaper

Don't have high hopes for being able to get my hands on an FE card here in Straya though

>>60930423
Well that's the thing.

It's enabled on the $1200+ WS cards. Which they are marketing to where a big office with thousands of employees could have a few Vega cards in a server while all their employees use ultraportable laptops that can utilize those powerful cards through virtualization. 512 TB of virtual address space per card.
It WILL NOT
>WILL NOT
be on the RX Vega <$800 cards. I already confirmed this a while back, which is why I told myself that I guess if it's $1200 I might just get the FE for that feature...

But frankly, I think the high end gaming cards, like 1080Ti and whatever the highest end RX Vega is, should also have that feature. Enterprise will still buy the $1200+ card for the support and stuff. At least throw hobbyists a bone.
But whatever... it's not like I don't have the money.

>>60930423
>>SR-IOV
What's this?

Some kind of hardware visualization for the GPU? Wasn't VT-d that?

>>60930795
Uhh VT-d is INTEL virtualization for CENTRAL PROCESSING UNITS.
Vega is not a CPU. And not Intel.

It's virtualization that allows multiple people to use the same GPU as if it's one GPU at 95-99% native performance, similar to VT-d.
So, for example, you can have a single computer with this GPU (assuming the software to utilize it is there) and run a whole house of devices gaming. Like a rasberrypi attached to your tv to stream games and controller inputs. Ultraportable laptops. Smartphone. Whatever.

It's interesting. Especially if you have a girlfriend (male) who enjoys gaming. Instead of 2 expensive GPUs in both computers you can really just have the one with minimal i/o latency on the other(s).

>>60917900
I hope they've improved their stock cooler. My case has kind of bad airflow so I'd prefer a blower style.