12nm Polaris 30 in Q4 2018, 7nm Vega 20 in Q1 2019, 7nm Navi Between 2019/2020

wccftech.com/amd-gpu-rumors-12nm-polaris-30-7nm-navi-7nm-vega-20-radeon-rx-gaming/

I know it's iranftech but whatever

Attached: Screenshot_20180701-095343.jpg (1045x1087, 391K)

Other urls found in this thread:

images.anandtech.com/graphs/graph11278/86530.png
tomshardware.com/reviews/amd-radeon-rx-580-review,5020-6.html
twitter.com/SFWRedditGifs

POO in GPOO

good post

I just got a GTX 1080Ti, how fucked am I? How much better will Navi be?

Don't worry it will probably be trash

I don't see Navi being anywhere near decent
not when AMD still hasn't figured out mcm for gpus

>POOLARIS

ENJOY YOUR AYYMD HOUSEFIRES

Attached: power_peak.png (500x690, 34K)

Oh sweet just bought a Vega 56 3 days ago.
Gonna go get comfy and wait a year for navi to hit actual shelves with any real numbers anywhere near retail

Attached: 1526746660774.gif (500x500, 800K)

Lol my undervolted 56 ocd to 1.6ghz only uses 220w about the same as my 1080 oc did

This it will be pretty bad at launch just wait
If it's big boi navi it will be around 1080 level like Vega and pascal just cheaper... Like polaris was to Hawaii xt

Navi is going up against the 1080, not the 1080TI, so don't get your hopes up, and don't fall for the fanboy hype.

The high end people have had 1080+ power for about two years, Navi only serves to bring that kind of power down to the 250~300 dollar mark

It won't beat the 1080 ti, but probably cost less and consume less (assuming 7nm)

1080ti market is very small they shouldn't care about them anyway.

Gtx 1070ti fills that niche but yeah my next screen will be a freesync anyway cbf paying extra for gsync

Jesus Christ, anybody actually believe these rumours? There's no 12nm GPUs and 7nm Vega is enterprise only. Navi is going to be scaled down Vega with lower power consumption, bringing GTX1080 level performance to the 200-300$ range. Meanwhile, nvidia will milk you dry for anything they can with their next cards. Expect 1080ti level of performance with lower power consumption at 700 bucks. So you'd be paying 1080ti money either way, but you'd save that sweet 40$ on electricity in a year.

>Navi is going up against the 1080, not the 1080TI

This is just false, AMD already has a 1080 equivalent. A simple shrink to 7nm, plus architectural refinements, will automatically bring 1080 Ti performance. The real problem is that by that time the gtx 1170 will also reach that performance.

hmm, im considering buying a vega 56. i actually dont really game only eroge mainly but its got to be an amd card. should i maybe just buy a 570 or something? its like $350 vs $700 AUD for the 570 vs 56 for strix cards. i really cant tell which one i should go for though. i want the 56 because i figure it will be quieter

Go for the 570

>bringing GTX1080 level performance to the 200-300$ range
$200-300 MSRP
that will be like $400-500 actually
and that will be 600-700€
which is like $800
just fucking kill me now

Attached: 1417555472697.png (696x1053, 272K)

This
Get something cheap now I had to wait over a year to get my cards at 719aud shipped
As a 56 owner it's not worth it
Titan v is a year old on 12nm shut your dumb fucking pie hole

GPU prices are coming down since nobody is buying and stores are overstocked, overestimating the mining craze. Up to 20% cheaper within the next month. I expect prices to fall even further the next months. The mining boon will repeat in a 4-6 years time, but for the time being, ASICs are whats driving miners and GPU hash rates are underwhelming. We'll get good prices soon and new cards will be at MSRP.

>As a 56 owner it's not worth it
what issues do you have with the card and are you using the gigabyte one by chance? im looking at a strix

Attached: GTX.jpg (791x835, 270K)

>Titan v is a year old on 12nm shut your dumb fucking pie hole
Are you implying nvidia are going to sell a card that is Titan V levels of performance for 700$ as a 7nm 1180?

Wait, Polaris will obviously be mid-range (basically the same as now but on a smaller process to reduce power and/or bump clocks), so why also release a mid-range Navi GPU?

It will be probably a bit more powerful.
You're underestimating the 7nm node.

>It will be probably a bit more powerful.

Attached: 1529767249937.webm (450x472, 1.44M)

Polaris will be low-range in the next generation

Hmm I see. So the RX 680 will basically be Vega/Navi based instead of Polaris based, but with GDDR5? Or maybe even GDDR6?

Msi airboost I'm waiting on a morphius 2 air cooler no issues at all I just meant price wise it's not worth it.
What do u fucking think? God. This board is getting dumber by the day /v/ has better hardware threads
It will hit 2ghz core easily

You know that you can also undervolt Nvidia cards right? My 1080TI undervolts better than my Vega 56. Literally like 200W while still maintaining like 40% higher performance than the Vega 56

Good for you?

There's no way nvidia is going to give you that level of performance for 700$. There's no way they'll give you that level of performance for that kind of cash in the next 3 years.

Attached: 1434265669352.gif (320x240, 1.99M)

40% higher? post benchmark

No shit it's why I'm happy with my 56 and gtx 1080

Probably shit but Navi will power PS4s so it will probably play better.

>same card
Amd rebrand confirmed
AMDrones on suicide watch

Attached: download46e8300a__2fstorage_2femulated_2f0_2fClover_2f4chan_2fg_2fAMDEPYCRome.png (1023x480, 424K)

I've a 280x. Great card, but the old gal is starting to show her age. The 580 would be a decent bump in pergormance, but it's overpriced for the time being. Anything lower and it's more of a sidegrade than an upgrade. I'm hoping to ride it out a little while longer.

It's funny cards are more expensive here used still retarded miners still trying to make a roi after the ass fell out of buttcoins.
Fuck them
T. Gamer

>currytech

Yeah gddr6 is likely for a lower cost, shrinked Vega/Navi. With gddr5 it would definitely be bandwidth-starved.

>pergormance
Shit, I fucked up.

Prices are coming down.

Every damn generation there are retards who say this.
> b-but muh Titan costs 2000$, no way nvidia will do that.

Titan is overpriced crap, unless you work with neural networks.

Navi 10 is mid range Navi 20 isn't coming til 2020/2021

AMD is not going to drop something as complex as the Vega 64 to 200~300 dollars

Complexity means absolutely nothing.
What sets the cost for amd is die size and the type of memory.
A 7nm Vega 64 with a gddr6 controller will be nothing more than mid range next year.

why would it be bad
it's literally using technologies that helped pascal and maxwell get ahead of amd, likely will take advantage of the end of shader engine limit. amd also has raked in more cash from the cpu market to redistribute and they have more than a year to build up their graphics infrastructure to take advantage of more powerful geometry engines; navi will almost certainly use gddr6 controllers, have smaller or equivalent dies (cheaper) and on a superior process to nvidia's which will remain on 12nm for a while. it's very likely navi will be within striking distance of volta cards in gaymen just by merit of clock speeds, not withstanding architectural changes. the real question is how much impact will amd make by then in the hpc and gpgpu market, i think that question is a lot more interesting than whether or not amd will be able to perform on games, i think amd/rtg most certainly has the ability to do so, but it never mattered anyway since it takes amd cards more than just performance advantages to win over nvidia's gaming market.

How many hashes per second can I get with these? Asking for a friend

Are they again going with hbm2 when its price is inflated?

christ WCCFTECH is such trash, based off rumors by some account who previously posted rumors that were loosely accurate, what a load of bullshit

People expect GPUs to get get better over time, not go sideways. Games get more demanding over time as well, most people don't "need" a 1080Ti now but they might in a year or two.

yet idiots visit pajeetfftech and pitch in some page views.

You should be upset about the 1180, not Navi
What's funny is the RX480 is quite reasonable.
the 580 is lol though.
I got a v56 last year for $380
feel bad for all the people getting stuck buying them for $500.

preorder to lock in msrp :^)
what the fuck is wrong with you? just get a rx 460 or 560. maybe even a rx550 would be enough. if you dont even fucking game. eroge is not worth getting a videocard for dude.

>7nm
>1100 series
lol

why is the gtx 1050 so based?

why the fuck would you switch from a 1080 to a Vega56

it's literally a newer node

>i want the 56 because i figure it will be quieter
On what basis did you come to that fuck stupid conclusion.
>i actually dont really game only eroge mainly
then get RX 550

>The GeForce GTX 1050 is a mid-range graphics card by NVIDIA, launched in October 2016. Built on the 14 nm process

that's because those benchmarks from shillpowerup are wrong.
They don't keep consistence between benchmarks and they don't update the benchmarks when they change methodology

Attached: shill_power_up.jpg (711x457, 57K)

english motherfucker. do you speak it?!

and to continue shitting on pajeetpowerup and sh1llzard
somehow peak powerconsumption from tpu is 240 watts...
anandtech measures peak SYSTEM power consumption for 580 250 watts images.anandtech.com/graphs/graph11278/86530.png

how fucking dumb are you?

>2016
>new

>Pascal is manufactured using the 16nm FinFET process.
newer than 16nm, yes
learn to read

>msi 390x

No wonder that thing draws so much juice - its clocked to 1100mhz which makes it the highest clocked hawaii card to be released. Hawaii really shouldn't be clocked over 800-900mhz from an effeciancy perspective.

>October 2016

Attached: _1417289643805.jpg (550x366, 39K)

That doesn't reflect actual heat.
RX 500 series are cooler than comparable gtx 10-series GPUs, and significantly cooler than 400 series.

>tfw privatized electricity is 25c a kwh

Attached: 1529623360229.png (2296x2100, 230K)

And has put a lot of their money into paying off debt
They had been in a bad place for a long time

Freesync probably.

tomshardware.com/reviews/amd-radeon-rx-580-review,5020-6.html

>Trying to defend this garbage

KILL YOURSELF FAGGOT

Attached: aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS8wL0kvNjY4NzU0L29yaWdpbmFsLzAwLVdhdHRhZ2UtT3ZlcnZpZXcucG5n.png (711x533, 25K)

Dude vega is worse than fury x what makes you think navi is gonna be better than vega lol retard amd is so fucked

>vega will certainly fix all of AMD's gpu problems!
>vega flops
>navi will certainly fix all of AMD's gpu problems!
>???

> iranftech

explain

>vega flops

Only Jow Forums thinks that. Then again Jow Forums was utterly convinced Inted had a new super architecture ready to counter zen and crush AMD - look how that turned out.

>vega flops
where is the vega stomping on the gtx 1080Ti?
thought so

Right because vidya is the only application for a gpu.

they are iranian not poo

Attached: 555.jpg (598x369, 45K)

AMD isn't the one with 300,000 gpus being sent back to their warehouse because they'll never get sold.

now you are changing the subject
we are talking about vega vs 1080ti performance

Nvidia sells MILLIONS of GPUs every quarter

300K is NOTHING

>peak
fuck off retard

>iranftech
What?

As much as it's wccftech, I've been saying that we'll be getting 12nm GDDR5X Polaris refresh in Q4 2018 if Navi cards aren't on track to launch in 1H 2019.

Usually we'd hear AMD talking such a thing up by now, but under The Wang RTG has been much more tight lipped.

Mid-range Navi will be 1080 to 1080Ti performance for half the price.
High end Navi (if it comes, which it may not) would be better, obviously.

>You know that you can also undervolt Nvidia cards right?
No you can't. You can only set the power limit on Pascal cards. That's not undervolting.
>40% higher performance at 200w
LMAO.

Yeah. I had a 7970 since the launch of the high end AIB cards. Got the best one you could get, the Lightning, at the time.
Was a great card, but the heat sucked and the video coder wasn't up the snuff for 1080p@60fps on Twitch at the same time as other videos, or 4k 60fps youtube.

Glad I replaced it with an RX580 finally when I should had earlier. There's a lot of little modern conveniences plus the lower power with it. But I still want a card that I can play anything in 4K on at closer to 90fps.
But there's no 4k, 10bit, 90%+ DCI-p3, 85hz+ monitors out. Freesync 2 HDR, basically. And I want 27" or so. Backlight also has to be good and even...

where do you live user, mine is privatized but i only pay 7 c a kwh

commiefornia

You can still undervolt the RX580 and perf/watt is better than a 480. They clocked them so high stock.
Also the reality is, with a 60fps lock, my RX580 uses around 40-115watts since it's so overpowered for most games.
I even run it at 2560x1600 medium to highest settings, just whatever looks good, and that gets me 60-95fps in most cases. Locking it to 60 with FRTC like that keeps it running cool.

200watt, under max peak load, isn't even that high. Lmao. That's 4 lightbulbs.

Comparing OC models to FE without having both for each model is dumb.
I have OC 1070 cards and they pull more like 170-185 in Furmark.

polaris30 - 48cus/48rops/1600mhz

Vega 10 was rushed so this is a good thing

makes sense, my state sells power to your state
always thought the california seceding from the union was hilarious when they dont grow their own food or have enough power

What's the limit? Can they make 1nm shit?

...

AMD GPUs have been shit forever so unless they pull a ryzen I wouldn't sweat about it.

Original source is actually Chiphell, so this is more believable.
I'd really hope for more than a 15% performance increase, though, fuck.
Just give 40CU Polaris with GDDR5X for $250.

Limit is 3-4nm for silicon.

So that makes them more based than any other site.

yeah because the power they draw magically dissipates into the ether

>AMD GPUs have been shit forever

Attached: 1432515406550.webm (800x450, 253K)

Isn't there supposed to be a new meme material to start making chips out of that they've been hyping forever but hasn't been seen yet?

>Limit is 3-4nm for silicon.
Not same user but I recall an article about hitting a controlled but expensive 1nm limit.