/w4v/ - Daily Wait for Vega Thread - 3 Week Countdown

can you post that black with red leds render? it looked really good

RTG will be sold off for pennies on the dollar once Volta releases and BTFOs Rajeet once and for all.

File: VEGA3.jpg (72KB, 1500x943px) Image search: [Google]

72KB, 1500x943px

>>61285627
Here ya go.

>>61285662
this thing is pure sex, hope it looks like this

thanks anon

the wait is getting to me, summer is dead season and I want to replay a few games
but I wanted to do that with new videocard not to worry about fps drops
can't buy 1080 or 1070 would be a waste with current prices, also lack of cards in the stores
390 runs 1440p somewhat decently, but damn

Why wouldn't you just buy a 1080 TI?

File: AIjIH9d.gif (679KB, 400x400px) Image search: [Google]

679KB, 400x400px

>>61285394
Whats the esimated hashrate of the vega cards?

>>61285867
This question btfo all amd. 1080 even

Who's hoping AMD drops out of the consumer GPU market?
I personally can't wait.

>>61285867

Freesync monitor
Price/performance
Not wanting to deal with nvidia control panel.
Not wanting to deal with g sync
Not wanting to deal with shitty nvidia software and drivers.

Thats why I'm getting Vega.

Xth for sold out because of miners in the first hour

>>61285914
the FE performs like a turn on mining, doubt RX will be any better in that regard

File: MHs-1030x639.png (51KB, 1030x639px) Image search: [Google]

51KB, 1030x639px

>>61285874
its 20% or 30% faster than a FuryX results.

Please don't buy.
Prices are bad enough.

>>61285896
Then what?
1160 for only 1160$
Stupid gaymoar GPUs are ridiculously expensive already due to the eth mining boom. You wouldn't want that dirty Jew tricks company holding a monopoly on top of that.

>>61285965
>1160 for only 1160$
That's exactly what I want.

>>61285943
I don't just mean DaggerHashimoto, I mean all hashrates.

If I just wanted to mine etherium, I wouldnt be asking this question. Still going to preorder though.

>>61285968
This might seem like a stupid question, but why would you want the prices to skyrocket even outside of the mining crazes?

>>61285394
Vega is DOA

>>61286014
It wasn't obvious? Because I want these gaymen niggers to not play games because the hardware is so overpriced, the tears will be delicious.
We need another 5 years of Bulldozer on the GPU front, with Intel prices(or worse) to match.

>>61285968
Actually, that's a good point. Game companies would have to optimize their code again or go bust due to fewer people that could afford 300w bitchin fast 3d graphics cards.

>>61286028
>wanting overpriced GPUs because of hate towards gamers
Why? Who hurt you?

>>61286040
These /v/ spillover apes are shitting up the board and pretty much any graphic related forum.

So yeah, they hurt me, they make my free time less enjoyable.

>>61286040
iGPUs are good enough for daily use. Only gaymers are being butthurt and that's always nice to see.

>>61286054
>iGPUs are good enough for daily use
>>61286047
>/v/ boogyman
I'd like a non-bait answer, please.

>61286071
that Nvidia troll does this every day the stupid the poorfag hasn't bough a GPU in years


he just wants it too expenive so that he can keep up with us.
Honestly just ignore dumbasses like this.

>>61286037
or you know, gaming companies simply stop developing for pc and you'd have to play 30fps 900p for all eternity

>>61286104
oh and the funny part on AMD hardware because nobody wants to deal with nvidia bullshit in corporate world

>>61285943
no they aren't, the guy who tested it live got ~33MH/s

>>61285867
Not buying anything from a company that ships spyware drivers with $700 pieces of hardware.

>>61285874
Dog shit on Lyra2REv2, the only algorithm that matters.

>>61285662
Looks nice though it's a shame the reference blower sucks.

what is the gpu after vega's successor?
I don't think I will upgrade for the next few years because of miners

>>61286645
Navi. /w4n/

>>61286867
>Sodium Industries
I kek'd

>>61286645
>what is the gpu after vega's successor?

It depends on who's gonna buy AMD once it goes under after Vega RX's upcoming blunder.

>>61287293
>AMD only makes GPUs
3/10 made me reply.

>>61286116
>because nobody wants to deal with nvidia bullshit in corporate world

Dumb fuck, the consoles use AMD because they want an x86 APU and only AMD can do that. Intel can't into GPUs and Nvidia doesn't have an x86 license.

>>61287358
Fucking Intel fucking hates dealing with nVidia. They are fucking cunts.

I am going to crossfire Vega on an i9 7900x rig.

>>61287360

>Intel says

So now Intel is good because it fits your anti-Nvidia AMDrone bias?

>>61287361

>Shit GPU and shit CPU

You got it backwards, you SLi 1080Tis on a Threadripper.

>>61287358
Lmao theres gpu in console not apu kek

>>61287366
No, even fucking Intel, the kikes themselves, hate dealing with nvidia.

>>61285905
Don't kid yourself. The only reason anyone buys AMD is because they're poor.

https://www.youtube.com/watch?v=txFdcAlf2Yc [Embed]

AYYMD IS FINISHED & BANKRUPT

AYYMDPOORFAGS CONFIRMED ON SUICIDE WATCH

File: Screenshot_2017-05-02_14-37-21.png (9KB, 451x86px) Image search: [Google]

9KB, 451x86px

You still here buddy? Its not looking good for you.

>>61287838
Hey that's me

>>61287847
That was fast

>>61287873
/g/ is not a very large board and these threads are in 90% cases always found by filter

>>61287432
Not an argument.

>>61285905
I gave up, caved in, and ponied up for huang's wang again.
Nvidia is currently top dog for 4K 60.

>>61287992
>60Hz
cuck

>>61286071
iGPUs or even low end (GT 1030) are more than enough.
>/v/ boogyman
You don't need to believe in crossboarding scum for them to shit up the board. Personally I want to see AMD back out of the consumer AIB space alltogether and leave gaymurs to get cucked by leather jacket man for all eternity.

>>61285652
>muh big die pascal

Kill yourself, shill.

>>61285905
This. AMD drivers don't have telemetry shit and other bullshit. Cards don't get gimped when new ones come out. Freesync is miles cheaper than gsync.

honestly after 2+ years waiting for a 290 sucessor i just gave up and got a 390x, now amd releasea fury x with 8gb of vram and clock for clock its the same as the new "vega"

give me a fucking break, im done, gonna get a 1080 used

>>61288091
Fuck disgusting TN panels, I hate them with a passion.
I'll get into 144Hz once 4K OLED becomes a thing and GPUs are powerful enough to drive them.

>>61288341
Me too, I'm waiting until DP has single cable support for 4k@120Hz (uncompressed) and GPUs have the I/O for it, I'll buy when a single GPU that can do an average 120Hz on 4k

Before that, I'm sticking to my mobile GPU and catching up on my backlog games with an emulator

>>61288341
>I'll get into 144Hz once 4K OLED becomes a thing and GPUs are powerful enough to drive them.
not this decade

gpu dev has stalled this gen

1080ti = bigger oc'd maxwell 980ti and still cant handle 4k

vega rx = fury x oc with 8gb of vram or 16gb or whatever meme amount, still cant into 1600p barely

>>61287432
I have jewvidya and I'm waiting for Vega to get off this jewish piece of shit. Kill yourself you disgusting shill. And have fun with your forced telemetry built into driver.

>>61288385
ayymy shill here

the vega rx can barely match the fucking 1070 ffs

>>61288459
RX Vega is not even out yet.

>>61288459
I actually don't give a fuck. I will gladly take under performing GPU over israeli data miner.

>>61288459
Reminder to report shitposters.

>>61288115
>wanting less compitition in any product
>still defending iGPUs as a main source of displaying desktops
You know how I know you are a cuck?
You know how I know you are a bait?

>>61288490
Yes, I want less competition soley to spite the consumers in said market. Said consumers are also braindead retards who kept buying NVidya even when they were slinging hot trash like Thermi, so they deserve to be fleeced.

Also, I said low-end GPUs are more than enough.

>>61287564
>Runs cooler and faster than a 1080 with potato drivers

RX Vega is going to match or even surpass a 1080TI on non-Gameworks games.

Wait for rx. Wait for navi. Wait for drivers. Wait for everything. Never deliver. The rtg motto

>>61288589
>I actively want the only competition in x86 to fail despite being several times smaller than their competitors

>>61288362
Even then not a total loss, at least back light bleed will be completely annihilated and we will be back were we started granted burn in is resolved.
For older games should be able to run high frame rates at 4K just fine, and watching videos shot at 24fps at 144Hz refresh rate without any post processing should look smoother since it is a direct multiple.

>>61288526
>wanting to spite the invisible hand
Just like the "invisible (black) cock" that is fucking your wife. and if you think that the death of compitition doesnt fuck you over too as an iGPU user, you are a limp dick retard, just like a true cuckold.

And to your other point, what If I wanted to watch 4k content? or hell, 1080p? Cause I can tell you from first hand experience, an iGPU does not rise to the occasion, just like you.

>>61288628
>I infer made up realities of other posters because I huffed too much street poo

>>61288650
>or hell, 1080p? Cause I can tell you from first hand experience, an iGPU does not rise to the occasion, just like you.
Kek. You're one hell of a retard if you can't even get 1080p content to work. You don't need a $500 GPU to watch your tranny porn.

@61288650
>black cock
@61288658
>street poo
Nothing worth listening to here.

>>61288477
then you are a dickhead.

>>61287358
They could still use an external NVidia gpu, except that both console makers got extremely burned when they did that, due to NVidia hoarding their patents like kikes.

Nintendo is about to learn this too. Don't expect the Switch successor to be backwards compatible like all their other consoles were.

File: 1496766003028.jpg (26KB, 552x437px) Image search: [Google]

26KB, 552x437px

>>61285662

How expensive will the vega be when it comes out?
And why do these threads exist? Its just going to be a new next gen card isnt it? Whats the big deal this time?

I'm building a new rig, should I wait for it to come out or just settle for an nvidea 1050 since everything else is so expensive thanks to crypto fags

We already know it's going to preform at at least 1080 levels and will probably undercut it by 100 bucks.

Solely for that and >muh freesync
its worth the wait.

Any thing in top of that is just extra. Also, i'd rather wait 3 weeks and buy a gpu that gets at least a couple of years of support rather than buy an 1080 and get fucked over in a couple of months when Nvidia drops support or sabotages drivers in favor for their next gen cards. People who bought the 980ti where hella screwed when Nvidia launched the 10x0 line.

>>61288916
Because of the hbm memory and die size don't expect cheap cards. If anything it will be fury prices

>>61288745
Switch can't play Wii U games
BC is already dead

>>61288916
As far we know now Vega RX will launch with three cards costing 399, 499 and 599 respectively.

Vega RX is Vega 10 which is the big dick line-up from AMD, Navi which is Vega 11 releases further up 2017.

If you're not inclined to spending more than 300 bucks on a gpu just buy the 1050, it'll run everything fine in 1080p. Look around for an RX 570 first, see if you can snag one on the cheap, but don't count on it.

>>61287366
https://www.youtube.com/watch?v=_36yNWw_07g

Gas yourself nigger

File: 1317726561959.jpg (11KB, 250x250px) Image search: [Google]

11KB, 250x250px

>50% larger die than the 1080
>much higher power consumption even with HBM2
>might not even beat it when both are oc'd
>it's a fucking year later

AMD has really outdone themselves this time around

>>61289019

Nvidia are cunts to everyone they deal with - they fucked sony on the PS3 and will (more than likely) never get another console deal from them again. In fact all that bad blood is why Nvidia declared that they didn't want the console contract ("b-b-aka, its n-n-not like I wanted it anyway") right after sony went public with the tech specs of the ps4. MS is now buddying up to AMD due to ryzen (also consider of the 3 consoles MS have released 2 of them have had AMD/ATi gpus in them).

>>61289116
Nvidia is making a shitton of selling consumer gpu's and much more than they would from low margin gpu sales in consoles.

If you think they are the least bothered you are delusional

>>61289065
Remember where they've come from. If they make that same leap again within the next couple years they'll be pushing far ahead of NVIDIA. This is still very good for AMD, even if not great for the market nor timing.

>>61289152
that's not how consistent, reliable business is built
read up on 3dfx
they were selling shitton of consumer gpu and king of the hill for a while
nvidia doing the same thing - proprietary apis, pissing people off with patent trolls, price gauging
they gone bonkers only recently and already slowly losing share they gained

>>61289170
> If they make that same leap again within the next couple years they'll be pushing far ahead of NVIDIA
How fucking delusional do you have to be to believe this.

>>61289065
Latest benchmark leaks where probably non-oc, and the higher power draw should only bother poorfags right?

Besides, spending 700+ dollar on a 1080 or ti when Volta is just around the corner is just dumb garnering that in a few months you could've bought Volta for the same price. AMDfags hate on Nvidia for fucking over consumers rightly so.

>>61289199
>that's not how consistent, reliable business is built
Nvidia's market share tends to disagree with you

>inb4 it's shrinking

It is, due to some good low-mid level gpu's from AMD, but they still hold the vast majority of the market

>>61289204
>Latest benchmark leaks where probably non-oc

Wrong
https://www.youtube.com/watch?v=lgweXM5LZGQ

>in a few months you could've bought Volta for the same price

Volta isn't a few months away and Pascal cards have been on the market for almost a year now

>>61289201
They did with Ryzen curb stomping Intel, don't really see how they couldn't do the same with Nvidia.

Dialectics of lead and all

>>61289213
again, 3dfx was 80% of the market and died overnight
marketshare is meaningless in a market that changes generations every 1.5 years

>>61289237
>They did with Ryzen curb stomping Intel
Ryzen was a good comeback, but curb stomping Intel, come on dude, do you really believe this?

>>61289232
Have you seen the video?
He's literally explaining to us how the benchmarks appear to not show them testing OC capabilities.

>>61289239
Marketshare only changes if there is a good reason for it, what you are describing is not something why marketshare would change over night

>>61289263
I have, have you?

He is literally saying the top results are most likely oc'd results

>>61289262
They've stepped up to the plate big time right? Intel was solo flying in all the markets for years, Ryzen insured AMD's comeback in both the consumer market and professional market.

Going from zero competition straight to butting heads is a slap in Intel's face big time.

Do we have any more information on the cryptocurrency cards? Are they even confirmed or is it just rumors?

>>61289317
it's not AMD thing, AIBs do that a few are out not sure if you can buy any

>>61289277
sure, thing build up build up build up and BAM you got a 30% swing
because semiconductor market is FAST

>>61289305
Like I said, they made a good comeback, but certainly they didn't curbstomp Intel.

They brought much needed competition and lit a fire under Intels ass to do actualy go ahead and innovate.

>>61289294
Skip to 8.40 in the video.

@ 9.17
"The point here of course is that's it's not very likely that any of these scores are OC"

>>61289363
>didn't curbstomp Intel.
they destroyed whole line up in one swing, i'd say that's curbstomping
all intel has left is i7 and pentium as legitimately options
two CPU that minority buys

>>61289363
Which Intel didn't, and even could've seeing as they're literally designing processors.

Skylake-x and Coffeelake-x are not going to sell for shit with the enormously overpriced motherboards and more proprietary high costing exclusive features.

They're a step behind now. Unless they're looking towards seriously price gouging their products it's smooth sailing for AMD in the multi-core processor market.

>>61289213
>Nvidia's market share tends to disagree with you
Their market share in a single market?

That's going to look great when the wind changes and gaymur sales plummet.

Why do you think they make so much fucking noise about AI meme shit now? They're desperately chasing a new market because gaymurs are their only real income

Vega RX(full one) will offer performance between a 1080ti and 1080 for $450

Screencap this

>>61289502
Latest benches shows it just shoulder to shoulder with a 1080 and the number $450 hasn't showed up anywhere. It's $600 for the big dick RX Vega. Which will still be a lot cheaper.

Lower your expectations.

>>61289502
the dream?

>>61289538
sounds reasonable though, if it's 10% above 1080 for 450 it is the dream

anyone got power graphs for FE? not that hard to extrapolate gaming card from those

>>61289502
>484 square milimeters die
>8Gb HBM2
>$450

I'll believe it when I see it

personally expecting no less than 600 bucks.

>>61289502
maybe the cut down version. The full featured version will be $700 and outperform a 1080ti

>>61289538
Those benchmarks were done at about 1400Mhz and with all the new GPU features still disabled, effectively operating like an overclocked Fury X.

RX Vega will clock to 1630Mhz and get a big boost from the new architecture features being enabled.

>>61289593
HBM is more expensive but not much more than 10 GDDR5X modules, some $20 end cost difference, barely relevant for $400+ SKUs
Die size is irrelevant unless you know how much AMD pays per wafer or yields are( protip: you don't)
Remember that Nvidia and AMD had no problem selling 600mm^2 dies for $600 last gen.

>>61289363
>but certainly they didn't curbstomp Intel.
Intel aren't the ones with 80% yields and fantastic scaling through their new interconnect and MCM black magic

AMD set things up for a server-room stomping but only time will tell if they get to deliver it

Hello good day sirs, is vega out yet?

>>61289654
July 30

>>61289453
Man that 815mm^2 die reeks of desperation.

>>61289502
No, it'll be $700 and on the level of a 1080. Slightly faster in 4k.

They can get away with the cost due to miners buying up all cards.

>>61289673
That's when they announce the next release delay

>>61289619
It was already released that the full version of the Vega RX will be 600 bucks and it's not going to trump a TI. It doesnt need to.

The benchmarks where run on 1630Mhz, watch the video.

if RX Vega has a place, it's as a $500 chip. die size is big but it's actually a good chunk smaller than Fiji so that price is realistic IMO.

once the drivers are optimal (by releasing or a month or two in) the value proposition will be :
>performance ranges from 1080 (TW3 etc) to above 1080 Ti (Doom etc)
>costs 150$ less than the latter
>supports freesync and doesn't turn itself off in VMS

>>61289647
>Die size is irrelevant unless you know how much AMD pays per wafer or yields are( protip: you don't)
all we can do is speculate based on the data that's currently available to us. anyone here is dealing in pure conjecture. I hope we're on the same page.

with that said, you're correct. I take it back.

>>61289705
they never delayed since there was never a release date

>>61289705
No, it's when they will launch the card*.

* availability will follow a month or two later.

>>61289750
>performance ranges from 1080 (TW3 etc)
What? Vega is no GCN, it's supposed to have much better geometry throughtput.

>>61289751
How convenient. Just make everyone wait.

>>61289721
>It was already released that the full version of the Vega RX will be 600 bucks and it's not going to trump a TI. It doesnt need to.
>
There's no such thing

>>61289647
Speaking of Yields, if V11 is so late that would kinda imply we're looking at pretty good yields since it will take them months to amass enough big defective dies.

>>61289721
>it's not going to trump a TI. It doesnt need to
that's what you think as reasonable person that knows what price/perf even mean
this is the reason why AMD has only 20% of market, not enough reasonable people

reminder:
>nvidia owner killing amd owner over gpu argument this year

>>61289773
call it a conservative estimate, then. I personally don't expect AMD to be hiding a miracle switch that just needs to be toggled on.

it's plausible though.

>>61289721
There were no price leaks and the only 1630Mhz benchmark that leaked were not in video form, were a synthetic benchmark and still show overclocked Fury X performance with none of the new architecture being utilized.

>>61289798
Vega11 is not defective Vega10.
Its another die.
>>61289817
> I personally don't expect AMD to be hiding a miracle switch that just needs to be toggled on.
Then you're fucking retarded or can't read Vega's high level details.

AMD

>now we unleash the FULL PERFORMANCE
>OF VEGA
>AMD shows screenshot of driver software having all new features "ON"
>it actually performs worse with them enabled

>>61289834
why do you expect TBR (and whatever the fuck else) to magically make up the current measured gap between Vega FE and the 1080 Ti?

right now, all the big unknown shit is looking like it might put AMD on par. not ahead. and that's being optimistic.

fucking fanboys.

>>61289876
>why do you expect TBR (and whatever the fuck else) to magically make up the current measured gap between Vega FE and the 1080 Ti?
Hmm lemme think what made Maxwell good at graphical/mixed workloads.
Also it's not just TBR.

>>61289876
There's an entire internet full of flailing retards claiming RX Vega will be shit performance permanently and AMD is doomed finished and bankrupt etc.

>>61289826
https://www.pcgamesn.com/amd/amd-vega-gpu-specifications
>The top-end RX Vega Nova is rumoured to be released at $599.The rumours also put a $499 price tag on the middle-order RX Vega Eclipse and the RX Vega Core at $399.

If you watch the video it shows that the benchmarks where run on 1630Mhz, which will probably be the stock core speed for the RX Nova. It's scoring 30k rivaling the stock 1080. And with a 600 bucks price tag is definitely beating Nvidia in the bang for buck range.

Don't get overhyped. It's not going to be a $450 TI killer. Stay sensible please.

>>61289876
TBR and "whatever the fuck else" are the entire fucking reason the architecture was made, that's some 5 billion transistors unused compared to Fiji.

Also I'll remind you, Vega has actually WORSE geometry throughput than Polaris, let that sink in for a moment.

>>61289918
Yay, thet disabled even primitive discard accelerator for whatever reason.

>>61289899
it's plausible and I legit hope you're right. I'm done letting my expectations go wild though.

>>61289904
even my "pessimistic" expectations make the RX Vega look worth buying. I'd buy this >>61289750 for sure

>>61289918
there's definitely something wrong with Vega currently. my point is everything that currently doesn't work still needs to compensate for a massive gap.

anyhow, getting the thing to clock so much higher than Fiji is an architectural feat in of itself, and it might cause regressions in other areas. that's not unheard of.

File: 1468524100821.gif (2MB, 480x317px) Image search: [Google]

2MB, 480x317px

>>61289924
It's the ultimate sandbag. The sandbag to end all sandbags. Like bags .. of sand.

>>61289988
>is an architectural feat in of itself
What?
Just make longer bibeline.
Two stages is enough.
>>61289991
Like, it's Polaris tech and it works, why would you disable even it?

>>61289687
It does, like Intel announcing they'll sell us an 18-core unlocked cpu for probably half or less of what it'd go for in Xeon form

Both are backed into this corner of making bigger parts on massive dies because they don't into MCM yet. Lucky for nvidia and intel, money's not a problem and they can just throw the stuff around to keep things going while they come up with something more scalable

>>61290010
>Like, it's Polaris tech and it works, why would you disable even it?
You do not understand 4D sandbagging

>>61289876
Okay let's see.

TBR is a mainly and primaly a perf/watt and bandwidth reducing feature, both are important for Vega(the FE one at least) because both are bottlenecking Vega FE at the moment, you can easily clock FE to stay at 1600MHz but it doesn't benefit from it at all, BW bottleneck.
perf/watt is self explanatory, it would reduce power draw at current clocks or simply keep the same power at higher clocks.

Secondly is the geometry units, which are behaving like Fiji, which is slower than Polaris in this regard, and I refuse to acknowledge Vega having an older geo engine than Polaris when AMD has said a million times they can use any module IP they want on different arches (see PS4 Pro which is a mix of Vega and Polaris), it wouldn't be much effort to dump the geo engine of Polaris into Vega, yet it's still performing like Fiji.
As for why this is important, geometry performance is a major AMD bottleneck, in games where it's not a bottleneck (and it moves to a shader bottleneck) you can see Fiji perform really fucking impressive, getting close to a 1080.

>>61290031
Cucksumer GPU market is not the place where you sandbag. At all.
You show your products shiny & chrome to make all the (man)children happy. Like the way nVidia does it.

>>61290063
>Cucksumer GPU market is not the place where you sandbag. At all.
You're thinking 3D sandbagging. This is 4D

>>61290072
4D sandbaggin would be RTG showing a working MCM GPU prototype.

>>61290063
>>61290072

AMD patented sandbagging do not steel

>>61290090
Worked very nice for EPYC.
Who the fuck knew it has a fucking AES-128 engine built into the memory controllers.

Is it too much to hope for to hope for 1070 levels of performance at $400-450?

>>61290108
We don't know.

File: AMD-VEGA-CUBE-2-740x334.jpg (29KB, 740x334px) Image search: [Google]

29KB, 740x334px

"This isn't even my final form!"

You people don't know anything about Vega, where the press slides, presentations and release announcement?

>>61290126
Well they demoed HBCC.

>>61290025
AMD beating Intel's very best silicon by glueing 4 80% yield dies together and selling it for half the price... that's history in the making

File: Screenshot from 07-07-2017 22:38:09.png (438KB, 1625x898px) Image search: [Google]

438KB, 1625x898px

Just a reminder AMD posted this a few months ago.

>>61290146
Remember when people were surprised as hell that Vega has 8GB less than a month ago?

>>61290144
Magny-Cours was that already shame Stars was much worse as a core compared to Zen.

>>61290146
>4x power efficiency

>>61290160
8GB stacks*

I'm happy with Vega if they just reach 1070 levels of performance at a competitive price desu, I barely need more than 580 levels of performance anyways but I might as well go for the next step up if the price to performance ratio is good.

>>61290146
But how do you know AMD wasn't aware that all those features were not working when they started mass producing Vega?

t. generic Nvidiot's understanding of tapeout to production timeline

>>61290199
What matters if perf/watt. Vega FE sucks as that.

>>61290216
>What matters if perf/watt.
Unless you ask Intelfags

>>61290226
BAD GOY.
REMEMBER TO BUY OUR XEON SCALABLE FAMILY PROCESSORS.

File: blr_3840.png (198KB, 548x629px) Image search: [Google]

198KB, 548x629px

People seem to think vegas will only reach 1070 levels of performance - AMD has had such a card on the market for a long time already - the fury x.

>>61290126
I want a Vega cube

>>61290244
Right now it performs like slightly better housefire edition of 1070.

>>61290244
How much did it cost at launch?

> GlobalFoundries has only access to 14LPe and 14LPP. The Globalfoundries 14nm+ node is HPP. Not to be confused as a better version of 14HP, which is SOI FinFETs. 14HPP which is Bulk FinFETs and is basically 14LPP with 7LP-styled transistors, MOL, etc, and such has higher performance.


Vega 11, Raven Ridge and Pinnacle Ridge seem to be using this process.

Very interesting

>>61290287
650
3 years ago.

>>61290216
>>61290226
high power draw doesn't matter if the performance is there. just ask anyone who owned a 7970GHz

>>61290325
I'm an AMDfag and that is plain bullshit, Zen is ultra fucking efficient, there's no reason Vega shouldn't be the same.

>>61290325
>just ask 1080ti owners
heck even 1080 is above 225w

>>61290325
The die is big so the performance is not there right now.
At all.

>>61288977
>1050
>run everything fine in 1080p
Yeah on minimum options with 40 fps dips

>>61290330
GPUs are glorified parallel calculators, you can't compare them to CPU

>>61289909
>even the person that wrote the article doesn't believe the price rumour
>6 months old Doom benchmark
Please try harder next time

>>61290336
The moment you start throwing moar ALUs at the problem "efficient GPU" meme goes down the drain.

>>61290311
is Pinnacle Ridge Zen2 or is it the 14nm Zen refresh?

>>61290330
if you run a single GPU setup it's honestly not a big deal

I'd rather have a fast hungry chip than a slow sippy one

>>61290337
sure. if Vega is neither fast nor efficient then it's going to be a Bulldozer tier blunder. thing is it only really needs to be fast.

>>61290392
>is Pinnacle Ridge Zen2 or is it the 14nm Zen refresh?
It's Zen+ (aka refresh) and it's cancelled already anyway.
GloFo is too sanic with their 7nm LP node.
>efficient
"Efficient" GPU is a meme. The moment you start throwing moar ALUs at the problem "efficiency" goes down the drain.
What matters is perf/watt.

>>61286054
>iGPUs are good enough for daily use
>>61288115
>iGPUs or even low end (GT 1030) are more than enough.

Some truth right there. I use a 3 monitor 1440p setup on my desktop and I've got a RX 470 with 8GB RAM on it. It's fine for a desktop.

My laptop has a i7-5500U and no dedicated graphics chip. It really is more than enough for one laptop screen - I don't play games on the laptop, so there's that. And I can't with the iGPU. But I can do everything I want to do on the laptop. It's totally fine.

>>61287376
are you retarded?

>>61289294
You should watch the video again and pay attention this time.

>>61285394
Volta will be out before this pile of shit.

>>61290994
>volta
Wow bigger dies thank you leather jacket man.

File: vega-v-furyx-firestrike-ultra.png (18KB, 824x394px) Image search: [Google]

18KB, 824x394px

>>61290146
>4x power efficiency
>2x peak throughput per clock
>2x bandwidth per pin

What card are they talking about here

>>61291050
The one with working drivers.

>>61290146
>"Primitive Shaders"
AMD BTFO by AMD

>>61292082
What? Do you know what geometric primitive is?

>>61290417
>Zen+ is cancelled
>Zen 2 is on its way sanic fast

Source?
And awesome if true.

>>61292116
Anon, 7nm LP enter HVM H2 2018. Risk production is H1 2018.
Lisa said they are doing tapeout of "7nm products" (guess them) at the end of 2017/

>>61292100
Not a clue (what are they?), but they have to be worse than Advanced Shaders, it's part of the name.

>>61292162
What the fuck are you even doing on /g/?

File: AMD-Zen_3.png (2MB, 2560x1440px) Image search: [Google]

2MB, 2560x1440px

>>61292116
I guess this

>>61292190
>The B-team is working on Zen 2
It's going to be shit just like Dark Souls 2. B-teams never deliver.

>>61290417
There was never a refresh. Zen on 14nm+ was referring to the APUs. GF doesn't even actually have a 14nm+, they never configured Samsung's new 14LPU

>>61292207
You need to go back.

>>61292160
>(guess them)
Navi (Vega 2.0), Ryzen R5-2600X, R7-2800X, R9(Threadripper)-2998X
Are they still going ahead with R3, haven't heard of that in a while.

>>61290206
Because the Vega hardware was completed nearly half a year ago and the only reason there's a delay is because of HBM supply and drivers. At no point was there any indication that there was a delay due to Vega hardware itself, only supply chain and software delays.

>>61292240
Ohhh boy, Navi is far and away from being a Vega 2.0.
>Are they still going ahead with R3, haven't heard of that in a while.
Yes.

>>61291050
>>2x bandwidth per pin
HBM2

>>2x peak throughput per clock
double rate half precision

>>4x power efficiency
Very specific compute benchmarks show that the Vega outperforms Fury X by 650% with similar power consumption

>>61288977
>Navi 2017
Navi won't release until q4 or 2018 q1.

>>61285848
Yeah my 390 isnt too bad with my 1440p ultrawide, but its performance in the game I play the most (FFXIV) is terrible at high settings, like 30-40 fps on max settings.

>>61292483
>Q4 2017
They release gayming Vega in Q3 2017 and you honestly expect them to do Navi in Q4 2017?

>>61289750
>costs 150 less than 1080 with sane performance.

I would hope so, it has been almost a year since the 1080 was released.

>>61291009
Long as it is faster it will fit in my full sized case. All I care about is performance for my 4k gaming.

>>61292916
Don't expect much from it, it's kekler refresh tier shit. Also incoming housefires since you can't just add moar ALUs without suffering consequences.

File: ryzen_pro_-_june_19_press_deck_-_final_17.jpg (199KB, 1918x1080px) Image search: [Google]

199KB, 1918x1080px

>>61292240
>Are they still going ahead with R3

>>61292190
the socket will stay the same right ?

>>61293280
Yes.

>>61293289
nice
i'm more ZEN™ now

Vega will be 25% faster than Titan.

Screencap this.

>>61293406
t. Raja Koduri

>>61293413
Raja has better things to do than shitpost here, his minions shitpost here.

>>61293441
The hardware dudes are making Navi, yes. The software dudes should chock themselves.

>>61293441
Yeah right. When White House staffers and senior officials are spotted browsing 4chan in Starbucks, that's when anything's on the fucking table.

>>61293509
But those don't do anything of value, compared to guy leading a small understaffed team without budget trying to compete with nVidia at least.

Will Vega be available short after the announcement or month after it?

>>61293545
We don't know.

>>61293543
Spicer's the fucking White House Press Secretary, one of the busiest guys on the roster. He's a fucking /m/-fag and closet weeb.

>>61293545
Why not just wait and see?

>>61293595
>He's a fucking /m/-fag
That's why /m/ has no moderation.
>>61293678
Everyone's tired of waiting.
It's hilarious though, everyone waits™ for flagshit Radeon and no one will buy it anyway.

>>61293280
IIRC AMD stated that AM4 will be supported at least until 2020

>>61287361
Are you building a bomb?

>>61293793
No, you need 7980xe and Vega (or GV102) for that.

>>61293804
>not a 1200W Vega cube


You like little baby

>>61293816
>not new hot 2U Instinct rack from Inventec
Pleb.

>>61293856
We're looking at heat/mm2

>>61294000
Nothing will beat 550mm^2 (give it or take it) GV102.

>>61294031
4x vega 10 on a pcb the size of a hand sure will

AMD got caught out by the 1080Ti. They had VEGA ready for launch several months before that but HBM2 supply and drivers held them back. Now they have a 1080 beater but highly unlikely a 1080Ti beater. It will have to be priced accordingly and expectations lowered.

>>61294766
AMD got caught out by cutdown GP102?

>>61288354
>grandmaster of waiting
teach me senpai

>>61285394
>not /w4n/

get with the times gramps

>>61295803
Don't think, feel

File: 14325.jpg (26KB, 239x268px) Image search: [Google]

26KB, 239x268px

>>61285683
>>61286474
>WHOOOAAAAA LOOKS GOOD
it's a rectangle

>>61296824
with brushed aluminum beveled edges and lightan cube with a nice color scheme.

>>61296824
Rectangles look bad?

>>61296849
Not at all...
It's just fucking rectangle though
>>61296844
wow edges black and red
design team deserves a raise

>>61296859
>design team deserves a raise
No shit they do. The card looks better than any gaymen turd in the past 5 years.

File: 17-GF-TitanXp-Key_Visual-copy-3-blog[1].jpg (74KB, 1080x608px) Image search: [Google]

74KB, 1080x608px

>>61296859
this ford transformers industrial POS looks better to you?

File: CardTop.jpg (548KB, 3327x2160px) Image search: [Google]

548KB, 3327x2160px

>>61296879
It's just a rectangle
All of AMDs reference cards are just rectangles

>>61296911
And it's fucking great! Modern consumer cards lack simplicity. They look like fucking toys.

>>61296904
No it doesn't

>>61296911
not enough edges for you?

>>61296928
I thought the old 980tis by evga looked pretty tasteful

>>61296911
>>61296904
>>61296928
>Giving a fuck what the internals look like
Fucking ridiculous. Saturates the market with faggots that demand pay to design shit that nobody ever actually looks at at the cost of performance or in a more poignant light MSRP. It's fucking absurd and people who jack off over the shit are 10/10 the worst people involved in the system sales.

>>61297064
I don't care what it looks like
I'm just responding to the people who are circle jerking over it's design when it's a fucking rectangle

they'll just commend AMD for anything, although I really liked Ryzens design, I still keep my box on my desk because even the box looks nice

So should I sell my 1070 now and buy vega later? If I sell now, I could take advantage of the gpu inflation.

>>61297088
You should sell it now, save up a little, and buy a 1080ti or wait for whatever AMD is going to release that isn't Vega because it's nothing

>>61297088
Wait for Navi

>>61285905
>Price/performance
People are delusional if they think Vega will be significantly better in price/performance then Pascal. 1080's are $500, $400 used and 1070's could be found for $350, $300 used before the mining shitstorm.

>>61294766
How do you get caught off guard by an x80Ti card? It came out like 10 months after the 1080 too, Nvidia probably figured they'd save it as a competitor to the top tier Vega but even they got tired of waiting.

Vega will be 20% faster than a 1080Ti.

Screencap this.

>>61299167
>XDDDD

>>61299208
u ma dbro

>>61299167
After 5 years of drivers vs GTX 6080?

>>61286054
You fucking idiot, if prices continue to skyrocket and GPUs become inaccessible to consumers they'll stop producing the whole gaming line which means you'll be stuck shitting out 4k for a quatlo or dealing with iGPUs or otherwise shitty and antiquated GPUs. The same can be expected for CPU sales if they ever skyrocket.

>>61293692
>flagshit Radeon and no one will buy it anyway
I will, if it's under 650

>>61298641
Operative phrase being: "before the mining shitstorm." unfortunately the market won't be flooded by the 1080 Ti, just fuckloads of 1050, 60, 70's that literally nobody wants because you can grab a 1080 for the same price as a 1070.

>>61299624
You'll buy it no matter what

>>61299659
>"before the mining shitstorm"
1070 is $670 here. It was $400 in march.

>>61299667
I don't know, 280w TDP might stop me, kind of worried 650w PSU is not enough

>>61299667
yes

t.Drone

File: question-questions_answers_5.jpg (184KB, 1000x756px) Image search: [Google]

184KB, 1000x756px

will the mining faggots make the vega rx cost too much? is there any chance it will cost around $500?

File: Reaction - Chihiro !Goddess.jpg (61KB, 397x393px) Image search: [Google]

61KB, 397x393px

>No good virtualization from Vega
Waiting for Navi desu, I can live with 1080p/60fps on a cuck 970. I don't even play that many new games and most of the ones I pick aren't disgustingly unoptimized shit.

>>61299909
I think miners will just be grabbing the used cards and trying to hock their shit before buying the new stuff

>>61299909
Very unlikely. That's the delusional price, I'm betting it will be at least $600.

It's ok, I didn't even want these digits.

File: shamefurdispray2.jpg (158KB, 319x480px) Image search: [Google]

158KB, 319x480px

>>61299978
You deserve a ban just for that mang

>>61299919
What are you waiting for exactly?

>>61300067
Virtualization nigga. I want to passthrough part of my GPU to a VM instead of needing two GPUs.

>>61299494
kek

>>61300095
then get a pro card dummy or that VEGA : FE or Titan Xp. High end isn't going to cutt it you need the top dog for that shit.

>>61300568
>VEGA : FE
>having it

>>61300554
2070 makes Vega outdated for half the price

>>61300780
The VegA FE has pro features that why the TDP is out the fucking window.
Just a reminder that the Titan Xp has the near
same TDP as a VEGA :FE +/- 20%
>>61300799
At half the comput.
Yup sounds about right.
AMD fail can't into software.

>>61301133
Are you fucking retarded?

>>61300799
>2070
>10 generations ahead

File: mD6yAju.jpg (113KB, 1883x1047px) Image search: [Google]

113KB, 1883x1047px

https://www.youtube.com/watch?v=6FzOJTNRaQI

>>61300799
GTX 2070 is just going to be rebranded GTX 1080

>>61301454
Where the fuck are people getting this bullshit from?

>>61301483
Common sense?
Volta is nothing but bigger dies. And the last time nvidia was simply making bigger dies the lineup basically consisted of rebrands.

>>61301483
because that's exactly what Nvidia did last time. GTX 680 became GTX 770, GTX 670 became GTX 760ti and so on.

>>61301538
lolk
They're reporting a substantially lower TDP for the same amount of power (50%) they'll likely uptune it due to enhanced thermal performance. They're bragging it up that they're pushing 5x and 15x pascal and maxwell performance respectively with their GV100 release, and it shows.

+5.8b transistors for the same wattage, in addition they're doing a die shrink to 12nm and a bandwidth bump.

>>61301607
Wow they got less leaky node! Bravo leather jacket man!
>enhanced thermal performance
Maybe in their wet dreams.

>>61301607
That's all deep learning workloads in data centers.
There is 0 talk of consumer Volta.

I wonder how well meme shit is gated in V100.

>>61300799
Wow! Cool! Where can i see 2070 benches and price, anon?

>>61301607
that's for tensor cores, anon
you know the ones that google makes x20 faster at 65w on outdated node

raw GPU teraflop goes from 12.5 to 15, that's it
no wonder jacketman was so evasive about actual GPU components and tried so hard to sell this shit from stage
on 7 month preorders no less

>>61301954
>we make 815mm^2 die with memes
>pls buy
Reeks of desperation.
Like really 815mm^2 die is fucking retarded.

>>61299919
>>No good virtualization from Vega

Either the PRO or SSG cards will have SRIOV.
Instinct has SRIOV confirmed though, the others not.

>>61302228
WX Vega will surely have SR-IOV.

>>61302228
What's preventing someone from having a modded RX driver that enables SR-IOV if AMD doesn't decide to flaunt it in RX just to spite NVIDIA's virtualization greed?

>>61302394
Firmware

>>61302440
What's preventing a BIOS flash then?

>>61302524
signing keys, which I assume would be pretty important for critical data.

>>61302529
Oh right, buildzoid mentioned he ran into that bullshit. Even so it's a matter of time before it's defeated, and it could be a Pro/Instinct only thing because businesses care about it whereas plebs don't.

i can't wait for a card that might compete with the 1070 over a year later that won't be in stock for a long time

>>61302859
t. leather jacket man

so what even is a current, upper high end+ amd offering?

>>61302879
why would an nvidia user make that complaint

i'm a disgruntled long term amd user who's been wanting to upgrade for fucking years

>>61302953
Buy nvidia, gweilo.
Yours, JHH.

Will definitely wait until after launch since I can already predict Vega is going to have issues with PCI-e passthrough like every other AMD product.

>>61302964
not an argument

>>61302964
>>61303401
not arguments

>>61302228
But those are all pro cards, it's expected for them to have it. Vega FE and the latter RX Vega don't have any of that juicy shit.

>>61304619
For FE it might be enabled sooner or latter, but RX will never have it.

>>61303360
>I can already predict Vega is going to have issues with PCI-e passthrough like every other AMD product.
What are you talking about?

>>61305063
Because it seems to be a common theme that people having issues with passthrough on AMD GPU's especially. Nvidia is a massive dick and tries to stop you from doing virtualization on GeForce GPU's, however this is easy to work around. AMD support for pre-Polaris cards is wonky and even Polaris has issues with system hanging and crashing that I've never seen on Nvidia. Also Ryzen virtualization is a shitshow right, Wendell of Level1Techs said it's "fine" but I'm really wondering what level of performance he deems acceptable. Intel still destroys AMD CPU's in virtual machine performance, I guess that's what I get for being an early Ryzen adopter. Not making the same mistake with Vega.

Pretty sure GPU passthrough in general is a pain, but I've not heard or seen any major issues passing through AMD GPUs, Linus was about to passthrough 7 nanos without too much trouble given the task.

>>61303619
>not arguments
quite literally not an argument

>>61305125
You usually have to wait about 2 years after release before drivers for AMD hardware have matured to the point at which Nvidia drivers are at release.

I was intending to wait for RX Vega, then the FE performance turned out to be shit, and now I can't buy an Nvidia card because of the mining bubble.

Seems like I'll be rocking my Fury for a bit longer, after all.

File: Screenshot from 10-07-2017 18:05:45.png (47KB, 1131x430px) Image search: [Google]

47KB, 1131x430px

Vega is ready guys, really FE is the final performance


>AIBs get final BIOS for RX from AMD August 2

https://www.3dcenter.org/news/der-release-fahrplan-der-herstellerdesigns-zur-radeon-rx-vega

>>61305330
Vega faster than Volta confirmed

>>61305330
>Mass production only starts in August
>Vega GPUs will be available by October

Fucking kys, poo in loo.

>>61305330
The fuck.

>>61305330
Vega FE

TL note: FE means Q2 edition

>>61305330
So we'll be waiting end of august for the features to be enabled.

Hopefully it's something, because Bega on paper should be more powerful than a TitanXP

Man RTG is surely taking their time.

>>61305330
I think I know what's going on!

AMD is pushing the release back to get drivers in gear/build up stock so it can avoid having RX all be sold out to miners.

>>61305452
ETH is crashing.
So miners will hang themselves looooooong before RX Vega launch.

>>61305452
>avoid having RX all be sold out to miners.
This might actually be part of the reason, if it's actually good price/perf they want gamers to get their hands on them, not miners.
The added time will allow them to fix more platform bugs before release.

Problem is Nvidia isn't staying still, be it Volta or Pascal refresh, it's coming sooner or later
Though if AMD is planning on doing a limited run of Vega(sub 1 million units) then their job is already done.

>>61305468
>ETH is crashing.
You can say that when every AMD and Nvidia card aren't costing two to three times their MSRP

>>61305476
Bega will have a very short lifecycle since 7nm LP is coming 'erry soon.

>>61305330
Based AMD

I bought a rx 580 nitro 8gb on on backorder for $260. Sold my rx 480 nitro at mad profit.

>>61305125
>Wendell of Level1Techs said it's "fine" but I'm really wondering what level of performance he deems acceptable.
Wasn't that only when running the latest bios/agesa update?

The one where AMD made a bunch of improvements to iommu and vm support?

>>61285394
Should I try selling my rx480? I got a new monitor that can do 1440p 144hz.

>>61305330
Who's trolling who anymore?

>>61305452
They're pushing the release back because they're incompetent and fucked up.

By the time Vega is widely available and we have working drivers you might as well wait a couple of months longer and buy Volta instead. At least then you can be certain that you'll get the card when it's supposed to come out with working drivers on day one.

>>61306076
>Volta
Wow bigger dies thank you leather jacket man.

>>61306018
IOMMU support is fine, gaming performance is still pretty bad. The issue is with the nested paging tables bug, which has been present since Bulldozer and never fixed. You get to choose between good CPU performance and 50% GPU utilization, or 100% GPU utilization and vastly lower single core performance and I/O, which causes stuttering and slow down and also dramatically longer loading times. What's the point of buying a fancy new 6 or 8 core CPU if you're getting Pentium performance and frametimes in games? If you just need to run Windows in a VM for compatibility reasons, Ryzen is fine, but as a Ryzen owner I wouldn't recommend it for gaming or GPU-intensive tasks yet. The entire point of GPU passthrough is to get native performance, and that's not what you're getting, so what's the point? Just dual boot for now.

>>61306076
>Volta
Mid 2018 called, that's still almost a year away

>>61306114
It'll take AMD that long to figure their shit out.

>>61306140
Do you mean hardware or software dudes?
Navi will be taped out by then.

>>61306140
Vega is a first gen arch, when GCN started out it took AMD a while but the drives came along and you ended up getting the 7970 which is the GPU equivalent of Q6600 and Sandybridge in longevity.

I'm not expecting much from Vega, but from Navi where the drives will be mostly complete I'm expecting a lot, even without the whole MCM thing

>>61306930
Reminder to report shitposters.