https://videocardz.com/71984/first- review-of-radeon-rx-vega

>>61897022
Whether Vega will be bad or good depends on its price, for 4k benchmarks it is just a little better than 1080. If they sell it for a less price then we have a winner.

>>61897115
For 500 bucks it's not bad
Too bad i couldn't wait and got a 1080 for that price

I'd like to see how the 56 compares, and honestly performance is where i expected it to be... the only real issue is the power and thermals.

File: pawnstar11.jpg (41KB, 525x297px) Image search: [Google]

41KB, 525x297px

$300 is the best I can do.

>>61897115
Mr Poo was promoting vega as a true 4k60fps card.

Buts it's not even bad performance, am I missing something?

>>61897188
To be honest it probably can be, ultra settings are typically retarded and dealing a few things back like AA at 4k could accomplish that 60fps.

>>61897156
Use Vega only in winter.

>>61897156
Right now i'll estimate the 56 as using 290-310 in gaming situations. Much more than the 210 we were promised.

File: amdloodie.jpg (162KB, 633x758px) Image search: [Google]

162KB, 633x758px

>>61897022
>500W

>>61897410
If you consider firestrike a game I guess.
Actual gaming is never covered by reviewers, only extreme cases.
Turn on freesync, set a frame target, turn on radeon chill and you'll never see half that consumption while getting the ideal gaming experience of no screen tearing and low input lag

>>61897507
Did you just paste some amd marketing material?

>Name your architecture after 5th brightest star in the night sky.
>A literal blue giant
>it turns out to be energy demanding as fuck
Hmmmm

>>61897529
No, I just own a modern GPU and monitor while knowing how to use it.

>>61897022
>videocardz
Nope
I'm waiting for GN, HU, OC3D, Adored and Level1 to quite likely say the exact same.
But I cannot trust any of these rumor mill sites.

>>61897410
>Right now i'll estimate the 56 as using 290-310 in gaming situations

how exactly do you do that? recent leak shows 64 300w on firestrike, it's heavier than gaming and exactly on the mark of 295 tdp

>>61897209
Not using AA for 4k can actually give you a 30fps (console experience) gaming on a mid-range card. For high-end cards disabling AA @4k will easily give you 60fps.

>>61897555
>not trusting the only reliable and unbiased source computerbase.de

Gn goes in the trash after recent outburst and constant nitpicking, while ignoring intel/nvidia faults next day
HU is even more incompetent than curlyhair, not to mention his admission of fear losing free samples from nvidia, still didn't delete that video, hilarious
OC3D is alright, he is way too focused on OCing and less on proper testing methods, but has right mindset
Adored - no.

one of the less biased sources is guru3d, but I don't really like his methodology he is honest though

>>61897655
>HU is even more incompetent than curlyhair, not to mention his admission of fear losing free samples from nvidia, still didn't delete that video, hilarious
Link?

Well, my dreams of an all red team build are dead..

Looks like the best option is Red/Green?

>>61897655
>OC3D is alright
He's on amd's payroll since ryzen. His vega "benchmark" about feels in doom on 100Hz locked ultra wide meme monitors was the lowest of the low.

>>61897682
https://youtu.be/jqEnwwHC5n4?t=9m26s

9:25

>>61897731
didn't watch that, he is out as well then
he is one of the more tolerable youtubers though in line with wendel who is also on amd payroll

>>61897693
I remember terrible ryzen leaks right before release day from some turks or iranians or whatnot
turned out to be not very correct

4 hours

>>61897804
>4 hours
Is when the NDA lifts?
I was wondering where benchmarks were this morning. But I suppose it's "the morning" in the US in four hours.

>>61897804
werent they basicly spot in with the bioses given though?

considering amd already sent vega out, they have final day 1 drivers, so unless they are lying about performance, or got it through other methods and could not use the drivers, they will pretty much be correct.

>>61898015
about 15% lower than launch ryzen reviews, which weren't good either for gaming part
iranian one was bad also for production part
at least that's how I remember it
ryzen gaming perf got good only in May or so

>or got it through other methods and could not use the drivers
I think they say it in review, it's not sent to them by AMD meaning they used FE drivers, also that is why numbers are exact copy of FE numbers

So where's the bottleneck?
OCing the core by increasing the power target does fucking nothing.
At all.

File: 5667503+_c7f2146a1678f370b9f1dab032d3ed8e.gif (49KB, 200x200px) Image search: [Google]

49KB, 200x200px

>>61897115
64 air is the same price as a 1080

Performs -5-10% in games
40% more power in balanced
Aymd drivers
jet engine cooler

>v-vega is not a gaymin card, it's a mining card. just wait for n-navi

>>61897410
but the 64 is 300w???

It'll be about 210, drop the clocks and voltage and it'll drop by a lot.

File: aSDA.gif (1MB, 245x245px) Image search: [Google]

1MB, 245x245px

>>61897507
>>61897529
lmao

>>61897022
>Literally hotter and power hungry 1080.
And fags are gonna wait another month for aib cards when they could just get the 1080.

>>61898110
Memory, just like every other recent AMD card.

Tonga, Fiji, and Polaris all bottleneck heavily on memory speed, and overclocking the memory gives massive performance boosts.

As far as I can tell, GCN has been developed around the idea that GPUs are going to be used in applications with enormous datasets, which will require massive data throughput. What AMD has done with GCN(NCU) is make a core that is dependent on data throughput, gambling on this large dataset idea. This is why they developed solid-state graphics with HBCC, so that they can feed their cores from terabyte sized datasets without bottlenecking at the data pipe. We saw this start with Hawaii, where AMD had to put slap enormous 512-bit bus on the memory, and literally duplicated the memory controller on-die, blowing up the power consumption.

The downside of this, is that gaming GPUs that don't require the insanely large datasets don't benefit from this, and so we see these GPUs with higher power consumption, that are good at high resolutions, but don't get fast framerates.

>>61897555
they are reposting another site though, its not their review.

>>61897731
you're thinking about hardocp, not overclocked3d

>>61898149
GOD DAMN IT, I WAS PLANNING FOR A FULL AMD 4K BUILD. When will the time come when AMD unJUST their GPU department like what they did with CPU?

>>61897764
he's confusing kyle at hardocp. tom for oc3d is pretty cool.

>>61898212
When they will have the money to waste on funny toys aka GPUs.
>>61898194
They could've bothered with developing half fucking decent DCC for Vega but NO, LONG LIVE BANDWITH STARVATION.

>>61897824
is that when you are going to apologise?

>j-just wait amd was sandbagging FE
>j-just wait for drivers
>tile base rasterisation was disabled!!!!

Face it anon, this is the performance of the card

>>61897804
the 1080ti, will last you a looooong time. If you dont have £700 to spend they vega might still be a good option for 1070-1080 perf

>>61898237
Vega is not a gaming GPU. It's a GPU that works well enough for gaming to keep people interested in AMD, but its strengths lie elsewhere.

The markup on gaming GPUs isn't large enough for AMD to justify the cost of developing gaming specific silicon, especially when we're on the precipice of fundamentally new computing architectures, flash memory, and process nodes. No, it's better for them to ride the Ryzen wave, and focus on the enterprise where real money can be made, before coming back to gamers in a few years when the technology advances settle down a little.

>>61898280
Visualisation (Vega focuses exactly on that) is extremely niche.
Hell, NVDA posts some $200 million revenue from visualisation.
The money lies in gaymen toys.

>>61898280
Makes sense, Big Meme™ and DeepMeme™ and AI™ are all the rage nowadays, AMD would be insane not to ride the hypetrain

>>61898300
This. Radeon Instinct coupled with Epyc provides compute density that allows for ridonculous performance in smaller form-factors, which massively increases the ROI on capital investment. We're talking a single rack server instead of 3-4.

>>61898300
But AMD completely ignores meme learning. Vega is all visualisation and large datasets.

>>61898352
Let me remind you that Radeon Instinct is where 90% of Vega GPU dies are going.

>>61898262
>the 1080ti, will last you a looooong time
after 780ti i'm not going to buy nvidia card with 80 or ti in number ever again.

>>61898262
>the 1080ti, will last you a looooong time.
until it gets gimped when Volta is out in order to force you to upgrade to GTX 2050ti

>>61897022
1080 performance in DX11 and 10% faster in DX12? Not bad if the price is right. However that thing must be a fucking housefire.

>>61898389
>gimped
Literally has never happened
nice meme though

>>61898403
>buy new videocard every generation goy, what are you poor goy?

>>61898389
Volta's not coming out for a long time. you canbuy a used 1080, use it for a year and sell it for the same price

>>61898373
Nah. Instinct is DOA assuming NVDA has at least SOME volume on V100.

>>61898413
ti

>>61898409
What did he meme by this?

>>61898403
someone show him how the R9 290x is still relevant today while GTX 780ti suffers in modern games even though when they both released the 290x was worse than the GTX780ti.

>>61898280
>especially when we're on the precipice of fundamentally new computing architectures
I can't think of any advancements that could be coming soon, especially not ones that would matter for general consumers. Did you have anything specific in mind?

>>61898423
That's because AMD's GCN architecture hasn't changed for literally 5 years.

>>61898425
AMD & Intel have a very neat bags of patents of heterogeneous mix-and-match designs.

>>61898149
>Performs -5-10% in games
source? I saw the exact opposite, except for retard games like civ VI

Hopefully AMD will have enough money to make separate gaymen chiplet for Navi.

>>61898423
It's true that nvidia drivers don't deliver equal upgrades for older GPUs, but they don't "gimp" them, that would imply that newer drivers decrease performance, which is completely false.
see
>>61898430
This is why older AMD cards get better performance down the line.

>>61898445
>>61898149
Not to mention 99th percentiles / minimum framerates are better

>>61898416
considering how jensen pushed v100 in engineers thraots "just take it, you will like, me promise"
they probably have volume, they don't have the ask because it doesn't do what people what from gpu - more flops
tensors cores are sued in asics and nobody cares for combo deals with questionable use case

on the other hand instinct box looks interesting because it's read to sue complete package of EPYC performance
they took idea from nvidia, when they sold their titan boxes and made them HUGE
gimping titan was the worst thing nvidia done to themselves

>>61898456
when i buy a +$600 GPU I expect at least a five year driver support.

>>61898459
which one is better? the 1080 or vega?

Something I haven't seen much commentary on yet is how much better a RX 56/64 will be when aged 3 years plus, compared to a 1070/1080. It's a different consumer segment above them (1080Ti, exotic solutions). I'm looking to buy a card that will get me to 2020+, having bought a 7970 late 2012, and I have a feeling a few are sitting similar to me; 290 or 780Ti owners looking for the next to buy something to sit on til 4K is properly mainstream. Would I get a 1080 now? Hard to make that call.

>>61898445
We're talking air 64 here, its behind the 1080 in *most* games.

did you read OP's link?

>>61898475
Vega

>in some games slower than a 1080
>price already confirmed to be $100+ above msrp
this shit is not going to sell unless it's really good at mining

>>61898476
>7970
Amazing how that card is still relevant. I'd wait for Volta or Vega 2.0 but that's just me. That card will still get you through med-high settings for another year or so. I had a 7850 2gb before getting a rx 580 8gb under the noses of miners.

>>61898488
nice, i will wait for data on frametimes though, that is all what matters when it comes to smooth gameplay. for the love of me i don't know why no website ever does it except for very few. avg framerate means fuckall if it jumps constantly between 300 and 150 fps.

>>61898464
Well with nvidia cards you get functional drivers day 1, rather than 5 years down the line.

>>61898495
even if it was shit at mining it will still sell due to shortages

>>61898519
this actually makes no sense at all

THIS IS LIKE FURY X ALL OVER AGAIN EXCEPT THIS TIME IT'S EVEN FUCKING WORSE. AT LEAST FURY X COULD COMPETE WITH 980 TI IN SOME TITLES. THIS GPU IS FUCKING 20% SLOWER IN BEST CASE SCENARIO AGAINST ITS TI COUNTERPART WHILE BEING MORE POWER HUNGRY THAN FURY X. FUCK YOU POOJETS, YOU HAVE TO FUCK UP EVERYTHING.

>>61898545
This uarch is made to accelerate large datasets.
Calm down, kiddo.

File: 1483630004872.png (470KB, 456x600px) Image search: [Google]

470KB, 456x600px

>>61898545
it's not a gaming card, it's a mining card, Carl.

>>61898538
go to >>>/biz/ and you will understand what i am talking about.

>>61898556
not him but i go to biz and have no idea what the fuck you are talking about. if' it's bad people will just buy something else, like a 1070.

>>61898463
Ye, P47 box is literally nvidia's DGX station on some very EPYC (pun intended) roids.

>>61898573
But if the 1070 is out of stock and miners want to get that juicy benefit before the difficulty jumps up then they will be stuck overpaying for Vega or losing on profits.

is it likely there will be after market coolers compatible with the blower models?

>>61898592
>1070 is out of stock
but it isn't
it's basically the same price it has been before the mining bullshit happened
geforce is bad for mining and that's a good thing for normal users

>>61898592
Thank god the miner craze isn't as retarded in Europe. 1070 are waaay overpriced that's true but at least our 1080s are sold at MSRP. From what I've seen on newegg etc. a 1080 costs literally more than a 1080 Ti.

>>61898592
if it pulls FE mh/s at 300w and cost $599 then no amount of difficulty bump will make the card a good choice.

>>61898505
That's my point my dude.

Once you remove the buying from interest ("epeen") phenomena, and consider the slow adoption of panels over 1080p, there's not been all that much need for change in the market. Consumption trends are odd for graphics cards. At least with CPUs/motherboards you're seeing enhanced I/O the further we migrate from southbridge driven functions. But really, a 2500K and 7970 pairing will still drive a 1080p exactly as needed.

>>61898434
Eh, GPUs are already super specialized. Heterogeneous design methodologies are something to look forward to for CPUs, but not really so much for GPUs.

>>61898294

Money doesn't lie with gaymen toys for big chip designs.

It is general compute and professional graphics. Nvidia has been going general compute and professional graphics first with its big chip design ever since Fermi. (It is part of the reason why GF100 was a flop at gaming).

It is mid-tier and reject shit that provide the bulk of gaymen toy revenue. Nvidia it is the 1070 and 1060 but they pale in comparison to the revenue made by the 970.

File: 1497932281413.jpg (7KB, 250x250px) Image search: [Google]

7KB, 250x250px

>>61898592
Here's something that's actually happening and will continue to happen instead : People will just delay and not upgrade their whole computer system including cpu/motherboard/ram/etc because there is no point with video cards at the price they are. Instead, they might reconsider consoles.

This mining faggotry has gone on long enough and is distorting the prices of everything in the computer hardware market. RAM/PSUs you name it and people are getting tired of this shit.

It's literally beginning to look like it was created as a (pseudo carbon tax) to slow and leech off the proliferation of advancing computer hardware. Also to tax the electrical grid (an actual carbon tax). When you look at asinine shit like difficulty inflation it all makes sense. They literally track the advancing computational power of computer hardware and ensure there is price pressure (inflation) along the way.

I'm still trying to narrow wtf is the purpose and intent of these shit coin systems and the above is what I keep circling back to. Hopefully the launch of these overpriced housefires from AMD sparks an outcry and a pronounced movement against these shithead miners.

>NDA breaking reviews
>trustworthy

>>61898686
Pretty smart post. I fucking hate crypto currency.

>>61898683
Money literally lies in gaymen toys, just look at nvidia's quarterly revenue.

>>61898688
Yes pajeet. vega will annihilate 1080Ti any time now.

I'm getting a 56 100% unless there is some huge surprise. I play at 1440p so it should be perfect, plus I always want to support the underdog so AMD is my go-to.

Going to build a rig with 1600x cpu most likely.

>>61898686

Gaming consoles have been eroding any at PC gaming in USA/Europe.

The masses don't care for $599+ GPU solutions.

AMD already has most of gaming console market under its belt for the time being. It is possible that console vendor might switch to Nvidia for the next-generation.

>>61898715
>buying an overpriced gpu only because it's made by amd

>>61898699

Not the high-end chips, you retard. It is the mid-tier shit that makes the bulk of that revenue.

Nvidia has been trying to move away from gaming as their bread and butter since it knows its days are numbered. The low-end portion of discrete GPU market has been destroyed by SoC solutions from Intel and AMD. It is just a matter of time before SoC solutions start to eat away at the "mid-range" market by being "good enough".

>>61898717
t. retard
The most cancerous games like LoL and dota have more players than all shitbox consoles combined.

>>61898686
>Instead, they might reconsider consoles.
I for one will never downgrade to a console as my main gaming platform. 30 fps is genuinely nauseating. If however you are talking about new PC gamers then you have a point.
As for shitcoins, their only purpose is to scam people, most of these coins die out in few months to a year.

>>61898748

The bulk of that LoL and DOTA crowd are in areas outside of the EU/USA.

File: 1501607587120.png (577KB, 570x567px) Image search: [Google]

577KB, 570x567px

56 WHEN

File: Vega-64-BF1.jpg (197KB, 719x988px) Image search: [Google]

197KB, 719x988px

so at new games the RX Vega 64 liquid performs almost exactly the same as a 1080ti, at a lower price and at higher power usage. this means that almost anyone without gsync should buy a Vega, unless mining niggers inflate the prices. I literally bought a 1050ti two weeks ago since every single AMD card in that performance range was overpriced.

>>61898757
I had a plan for two vega 56 cards. However, with the faggot pricing of the 64 I am seeing on newegg/amazon whereby AMD said they'd be MSRP but are now showing $100 markup, I am literally searching for alternative ahead of 2 weeks from now. I've never bought a Radeon card and with shit like this happening :
> Uses fucktons of energy
> performs worse
> More expensive

There's no way in shit I'm going to.

>>61898730
I don't think it's overpriced, it should handily beat the 1070 at around the same price. And power draw/heat should be acceptable since it's no the top end part.

But yes I'm willing to pay a little extra for AMD. I'm just very much about supporting the underdog, supporting competition.

>>61897022
hmmmmmmmmmmmmmmmmmmmm

looks like it gets beaten by the 1080 in the vast majority of those games

>>61898688
>Being in this much denial

and in a few hours it's going to be

>j-just wait for drivers guys, remember finewine. p-please don't leave..... they are just sandbagging

"get woke niggah"

>>61898857
In new selective games with heavy optimisation, it'll perform well, if all you do is play doom, aots and dice games. sure, go ahead and spend $700 on the vega liquid.

>>61898903
"heavy optimization" it's literally only decent use of dx12/Vulkan features.
1 year from now and every graphical engine will behave like this

>>61898876
newegg has not released prices yet, amazon will always have the price gouging marketplace that they have no control over.

you'll get msrp at those two places, they'll always be out of stock though so in reality you wil have to buy elsewhere or wait for 4 weeks with an amazon preorder.

Pretty much what I expected, Vega 56 is the most worthwhile if you can find it at the same or lower than 1070 price. Vega 64 is meh, Vega 64 Liquid is trash tier.

>>61898927
>1year from now
no

if that was the case all dx12 games will perform well with vega [spoiler]it doesn't[/spoiler].

>>61898927
>just wait
Please just don't come to /emugen/ to bitch about problems with emulation. I've had enough of retards with amd cards shitting the treads.

>>61898944
Are you implying that all games ported to dx12 actually use new features like async? [spoiler]they don't[/spoiler]
Time spy is a good example of a shitty implementation with only legacy features, in order to not piss off Kepler owners.

Reminder that HBM overclocking barely gives any extra performance, definitely not 1:1 scaling that you'd see if Vega was really bandwidth starved. Don't believe the AMD shill lies

>>61898930
Newegg has released prices for 64 air..
It's $599.
Not sure if its with gay game bundles or just the card. Same goes for amazon.

>>61898932
I agree with this analysis .. The power usage on these things is getting retarded. I thought the 1080ti was bad.. Good fucking lord :
https://cdn.videocardz.com/1/2017/08/Radeon-RX-Vega-64-2.jpg

File: ehh.png (1MB, 3458x1784px) Image search: [Google]

1MB, 3458x1784px

>>61899022
>j-just wait for them to implement full dx12 spec
>this will surely happen within a year

/sigh

>>61899039
us newegg?

AMD gpus are 1 generation behind again. Nvidia is not intel.

>>61897022
>>61897022
delet

>>61899039
that'll be bundle price, amd specifically said newegg and amazon(not marketplace) will have them for retail.

File: radeon rx4999.png (721KB, 1024x576px) Image search: [Google]

721KB, 1024x576px

not that bad if they can pick a good price point, but the kind of person that would buy a vega would probably just spend more on a 1080 just because they can

AMD told reviewers to focus on the 56, no surprise why really. Nothing to see here.

>>61899094
Nice heating electric

>>61898857
>picking bf1 to compare anything to amd
there is no way you're not a shill, it's the most amd optimized game out there, in any other game the vega 64 is similar to the 1080 or slower

>>61899101
but /g/ told me amd was sandbagging and 1080Ti is finished when they enable all the super hardware features in the drivers

>>61899149
>it's the most amd optimized game out there
Eh? I though that was Doom 2016

>>61899165
they're both almost the same

>>61899149
>amd optimized
Another way of saying it uses open standard correctly.

>>61899185
>literally grasping at straws

>>61899185
This desu

>>61899203
Look at the standards AMD introduced, Intel and Nvidia see performance gains with them.

Look at the propriety stuff Nvidia introduced, everyone loses performance but mostly anyone that isn't Nvidia.

File: 1491221061481.jpg (208KB, 1000x1000px) Image search: [Google]

208KB, 1000x1000px

>>61899185
I recently have been investigating GPU computing and its clear that Radeon pulls the same shit as Nvidya. All of the advanced features and non-play code APIs are only available on their pro line. The issue being that Nvidia is more up front about it whereas you have to dig around all over the place to find this out w/ Radeon. DESU, open standard lately just means your software stack is amateur hour. I really feel the need to separate RTG from AMD proper at this point because their GPU group is clearly fucking up from a hardware/software and even marketing stand point.

26 more minutes and I get to finally confirm this shitshow is over and terminate my wait to see what Radeon is going to deliver. Tons of benchmarks are leaking out and none look promising. The only thing that captures me is the potential $50 savings on price and the possibility of some very specific performance differences. That being said, I shudder having to deal w/ AMD's amateur hour drivers/software stack and OpenCL. If I do buy a card, it will be strictly on an evaluation basis.

>>61897022
for that price and tdp isnt bad for everything else but gaming sadly
i mean we know vega is on par with p600 in general computing stuff but i can understand the dillema amd had with miners/light workstation users and gamers..

>>61899267
An example... Box says Linux supported but drivers and all sorts of features are in amateur hour status :
http://www.phoronix.com/scan.php?page=article&item=radeon-vega-setup&num=1

File: Linux.-OpenGL-Vs.-Vulkan-GAMINGONLINUX.png (14KB, 599x311px) Image search: [Google]

14KB, 599x311px

>>61899267
>Intel CPU
>Nvidia GPU
>AMD API

>>61899267
Please don't compare pro market APIs to normal APIs.

>>61899352
>linux for gaming

>>61899352
Please don't take me for one of these brainlets on here who is uninformed and buys into
> 1 : A vidya benchmark
> 2 : A selected vidya benchmark that favors Vulkan over OpenGL

I know the benefits and promise of Vulkan vs OpenGL and I support and am looking forward to Vulkan. Vulkan is a consortium btw. If you weren't a brainlet and actually followed Khronos group, you'd know that. One of the big contributors is an individual from Nvidia's Staff...

And Oh btw, I have an intel desktop (windows/linux), a mac laptop (intel), and 2 ryzen based computers. Nvidia powers them because they've been the top product to do so. I am looking into Radeon's cards but so far it isn't promising. The power usage is shit, as is the price, and is the development and driver support.

I'm not a fanboy. I go with the best value product with the best value support.

>>61899372
I didn't. I highlighted that both of them pull the pro market API bullshit. As such, they're both really on par when it comes to profit taking... and I say this specifically about the Radeon group and nvidia. This has nothing to do w/ AMD CPU proper. This has to do w/ the slimy ass shit GPU manufacturers pull across the board. The cards have the hardware to support it but they gimp the drivers/software stack to drive pro sales. They're both guilty of this shit thus I have no fanboyism or loyalty to either. There's no sense yapping about shit being Open when it performs worse and you have the same API lockouts as your competitor. And btw, Khronos group wasn't created by AMD. Vulkan nor OpenCL are some AMD creation.

http://www.guru3d.com/articles-pages/amd-radeon-rx-vega-64-8gb-review,1.html

review

aaand it's trash
https://www.computerbase.de/2017-08/radeon-rx-vega-64-56-test/4/

So, its nothing unusual about this release
Cheaper, runs hotter, a little worse now, better once drivers matured
Most recent Radeon releases have been like this

HOLY
FUCK
IT IS SO FUCKING BAD.
LIKE, UNMITIGATED DISASTER-bad.

>>61899165
>corridor shooter
>optimized
There's nothing to optimize when there's nothing to render

And it's fucking nothing

>>61899565
It's 486mm^2 die that fails to compete with 314mm^2 ones.
Hell bigger Polaris would be better.

>>61899522
Huh vega 64 power consumption isn't as bad as predicted. Still getting the 56 though.

show me a unity benchmark. that engine makes stuff crawl because its so innovative

It's been fun, guise.

>>61899522
>that noise level under load

WEW lad

>>61899522
>>61899562
Proof that retard going on about bandwidth is full of shit. Overclocking HBM speed NOTHING. Vega is a fundamentally flawed sack of shit and now we know Navi is going to be the same shit. No amount of scaling up will help because it's clear it can't scale up anymore. AMD better rip their architecture up and start from scratch

>yfw when I am ready to buy atleast 40 of these for mining
>yfw underage/poorfag gaymers are still mad

>>61899627
k..enjoy your 30mhs rate @ 300 watts

>>61899646
LITERALLY 40% WORSE PERF/CLOCK THAN FIJI.
GOD BLESS AMD.

>>61899602
>reference card
wew, it's fucking nothing

>>61899664
yeah but that's the cheapest model. if gpu pricing ever returns to normal, do you really think anyone would pay over $399 for a 56 when, based on these benches >>61899522, it's only like 5-10% faster than the 1070 on average which has a baseline msrp of $349?

File: niggers.png (17KB, 579x238px) Image search: [Google]

17KB, 579x238px

great now i will never get a rx 580
>only 20% faster than a gpu that is half the price

Guys. I figured it out. Make memory bandwitch regressed for Vega launch. Make it 'good enough' to match GTX1080. Make it shit for miners. Get it into gamers hands. Wait a couple of months then release BIOS update unlocking memory bandwidth. BOOM! It's beating 1080Ti! Mining hash rate goes up incredibly. Miners now jump on it but gamers at least had a chance to get in before miners gobble up all the cards.

I can only dream.

File: jpg.png (47KB, 587x633px) Image search: [Google]

47KB, 587x633px

at least it's better for other tasks... well shit

>>61899701
mining clients don't use the new instructions yet.

Is it worth getting vega 56 for freesync? Out of the entire lineup it seems to be the best card

AMD shills are STILL doing lie-hype-disappoint strategy.

maybe you don't understand American culture but dishonesty is one thing people never forget here..

>>61899742
>Slower than GTX 1080 by a huge margin

TOP KEK

>>61899772
b-but muh driver optimizations and locked driver features

>>61899742
> unironically video encoding with a gpu
Summerfags are a thing apparently

>>61899807
>unironically doing stuff where amd sucks instead of cherrypicking the 2 tests in which amd actually doesn't blow

>>61899772
Except everyone forgot that AMD literal shill telling everyone bulldozer was going to have higher IPC than Nehalem

>>61899758
Yes

>>61899565
This, it will just come down to the prices really. The only thing downright bad about the cards is their power consumption, but in terms of performance AMD finally has a fine answer to the 1080/1070.

Looks like the AMD shill marketers are focusing on the 56

Kek, AMD really paid these techtubers

>>61899579
The consumer don't give a fuck about the die size. If it runs better at the same price it's a win, otherwise no.

>>61898389
Is this a thing? I know apple does this wit the iPhones and that's it,

https://www.newegg.com/Product/ProductList.aspx?Submit=ENE&N=100007709%20601301447&IsNodeId=1&bop=And&Order=PRICE&PageSize=36
>more expensive than a gtx1080 and slower
that's poo in loo tier

>>61898475
Vega is already better than the 1080 and I'm confident that in one or two months, with better drivers, it will get better than the ti.

>>61899927
whatever it is i am not paying the early access tax, when the drivers get better i will buy it for sure.

File: 2017-08-14_15-38-23.png (54KB, 320x486px) Image search: [Google]

54KB, 320x486px

>>61899910
>$499
>All the others at $599
Get it while you can boys.

>>61899958
>out of stock
why would you buy it anyway, even captcha says fire exit

At least it's shit at mining

>>61899987
it's shit at everything

http://www.overclockers.co.uk/sapphire-radeon-rx-vega-64-8gb-hbm2-pci-express-graphics-card-21275-02-20g-gx-387-sp.html

http://www.overclockers.co.uk/ocuk-geforce-gtx-1080-reference-blower-8192mb-gddr5x-pci-express-graphics-card-gx-21c-ok.html

Say it with me.

DOA

>>61900155
DOA!
DOA!
DOA!
DOA!

>>61900014
Exactly. At least the Fermi cards were dominating the field while causing housefires. This shit is just disappointing.

File: c3_1920_2[1].png (41KB, 582x408px) Image search: [Google]

41KB, 582x408px

what.

>>61900251
What in the fuck is that.
Why is it so fucking bad.

>>61900305
PEBKC

>>61900251
what in the fuck

>HBCC and primitive shaders still disabled

Where's the miracle driver?

>>61900379
They are?

>>61900401
https://www.computerbase.de/2017-08/radeon-rx-vega-64-56-test/

the only enabled new feature is tile-based rasterization

>>61900468
WHAT ARE YOU DOING YOU FUCKING RTG NIGGERS
ITS CHOKING FULL OF OVERDRAWN TRIANGLES
RYYYYYYYYYS YOU NIGGER

>>61900468
Proof that bandwidth isn't the problem. DSBR reduces bandwidth usage by 30% but no improvements from Vega FE, what a joke

>>61900305
it's a fake benchmark

File: images.jpg (11KB, 228x221px) Image search: [Google]

11KB, 228x221px

>that AMD tax on your electric bill

>>61900468
>Primitive Shader – noch inaktiv
>Primitive Shader – still inactive
this explains A LOT

>>61900546
sure it's fake m8 http://pclab.pl/art75140-7.html

>>61900379
I was afraid this would happen, and it's pretty disappointing to see.
AMD did the same shit with Mantle and Trueaudio, which took several years to be actually publicly available for developers. In that time, we only had a few shitty games use this technology as a marketing gimmick.
Primitive Shaders are looking to be the same fucking thing. They did talk about it ONCE ( https://www.youtube.com/watch?v=XC5Dy_b-kE8 ) recently, but there's fuck-all in actual details as to what primitive shading actually does that's different from the usual pipeline.

>>61900602
Prim shader requires no dev input.
Driver runs NGP prim shader path instead of vertex one. Vega as is has no vertex and geometry stages anymore in hardware.

>>61900602
Mantle wasn't available because it's a creation of EA DICE and AMD just wrote drivers giving the spec a platform

>>61900628
>Prim shader requires no dev input.
Source?

>Vega as is has no vertex and geometry stages anymore in hardware.
It's a pretty massive difference, as geometry shaders are usually avoided due to their cost.

Also, how the fuck would an indexed draw work with that? Would it just duplicate the vertex shader work in order to allow the merging of vertex and geometry shaders?

apart from AMD power consumption fail,

why THE FUCK would you max out AA on a 4k setting? what is the point of setting it to 16x? There is literally, LITERALLY no difference after 4x

>>61900698
>source
Rys literally said that in Twitter.
And I can't answer the other fucking question since I don't know how prim shaders work.

>>61897655
>Adored - no
Why not? He said Vega was going to be shit almost a year before it launched.

>>61900581
Literally a couple cents more every month.

>>61900773
Even if it was shit, he would still be bias towards it.

>>61900538
GN already tested Vega 56 overclocking and found big gains from memory overclocking and near zero gains from clock speed increases. Which proves precisely that it is memory bandwidth bound.

File: 12.jpg (69KB, 1760x740px) Image search: [Google]

69KB, 1760x740px

>>61900855
Yet he cuts through other reviewers bias and bullshit

>>61900155
DOA
O
A

>>61900155
>>61900175
>>61902050
>inflated britbong preorder
>meanwhile it's sold out absolutely everywhere else it's actually been available

Try harder.

>>61899626
"We halted testingoverclockinguntilAMDcan deliver a final driver where it works properly."

The shipped press driver was fucked for OC'ing. Most reviewers didn't get it to work at all. Go read the GN article where Steve says outright that all the OC'ing gains are from memory clocks.

File: 73fh55ivopfz.png (73KB, 894x442px) Image search: [Google]

73KB, 894x442px

>>61897022

>>61903167
You should add the difference in price to both of those as well.

File: c9a1d51ff3bc11c20fc4e5378bb779d4fdb15870e901182176972e3c95574206.gif (2MB, 174x174px) Image search: [Google]

2MB, 174x174px

>>61903211

This is what further increases the disadvantage of Vega 64.

@

Also

>putting a i9 7900X with a Vega 64 Liquid Edition

File: 1471396486391.png (182KB, 578x691px) Image search: [Google]

182KB, 578x691px

>>61897156
Vega 64 is a turd but 56 is destroying the 1070 and even trading blows with the 1080 at high res.
I am probably going to grab a AIB Vega 56.

>>61903468
>I am probably going to grab a AIB Vega 56.
Good luck, miners will be gunning for those cards too.
If you want one, you needed to buy one last week.

>>61903167
Its barely getting as much as the clock speed increase. I wonder what happened with this dud.

>>61903167
This is misleading.
Fury X and Vega 64 have both 4096 ALU and 64 ROPs and Vega has a bit less HBM bandwidth.
1080ti has 1000 more ALUs and 24 more ROPs than 1080 and double the memory bandwidth.

File: RX_Vega_Effective_Texture_Bandwidth.png (66KB, 1113x865px) Image search: [Google]

66KB, 1113x865px

>>61903681
This. Memory bandwidth is fucked, effective texture bandwidth is fucked, primitive shaders aren't even enabled yet.

File: 1060.png (159KB, 483x800px) Image search: [Google]

159KB, 483x800px

>>61903718
Of course you can't ever compare generation to generation, but you can at least ballpark it to get an idea.

1060 and 960 have the same configuration outside of the ram and the improvement is ridiculous.

>>61903910
nvm I was looking at the 960 ti they never released.

File: 1491936851750.jpg (229KB, 1920x1274px) Image search: [Google]

229KB, 1920x1274px

>>61903468
Amada branded Vega 56 will save us all.