Vega almost confirmed DOA, AMD want you wait Navi

wccftech.com/amd-taking-the-covers-off-vega-navi-may-16th/

>Navi is AMD’s third generation graphics architecture designed under the umbrella of the Radeon Technologies Group. The graphics architecture will build on the power efficiency principals of Polaris & Vega. Navi is believed to be the first GPU microarchitecture conceived entirely under graphics guru Raja Koduri. It will employ “next gen” memory, likely third generation stacked High Bandwidth Memory or GDDR6 depending on the product to boost performance and efficiency

>Navi is the first graphics architecture from the company built on Globalfoundries’ upcoming 7nm FinFET process. Based on what Globalfoundries has so far publicly disclosed regarding the 7nm FinFET process Navi should land sometime between 2018 and 2019. You can expect more details regarding Navi’s timing on May 16th.

Other urls found in this thread:

guru3d.com/articles-pages/amd-radeon-r9-fury-x-review,25.html
twitter.com/NSFWRedditVideo

>Vega almost confirmed DOA
How so, Ramesh?

>navi will be even better than vega
>that means vega sucks
you're a fucking idiot, OP

>7nm FinFET
crazy

>talk about Navi
>Vega is not even out
>not Another Massive Disappointment

It's GloFo (and by GloFo i actually mean IBM) and they've been developing 7nm process for quite some time.
NVIDIA talked about Volta ages ago. Where's Volta?

>NVIDIA talked about Volta ages ago. Where's

Pascal is released since 2016, Vega not.

Pascal was not supposed to even fucking exist back then on their roadmaps. So fucking what? Major changes take fucking time, Vega's not a fucking dieshrink.

AMD Analyst day, you'll just see roadmaps for Navi and Zen+ benchmarks will be E3 or Computex, but apparently computex is more likely

It's 100% Computex.

They will talk about Navi because they know Vega will not beat the 1080ti.

How does that even make sense you stupid fucking indian? They will talk about Navi because it's on their fucking roadmaps you stinking poointheloo.

Wait(tm) and see

Oh yeah i waited™ for Evergreens. They were so fucking good.

Speaking of Volta, GTC is right underway now, Is Nvidia planning another P100 "launch"?

We need moar wait™

No, later, Volta is not even taped out yet. Probably some fresh roadmaps, maybe some teaser.

>navi
AND THE AMDRONE WAIT (r) HYPE TRAIN BEGIN

If it's launching in Q1 it's likely to be taped out recently.

Then i waited™ for Hawaii. Still using my 290.

It's "launching" in Q1. Knowing novideo it'll be another GP100-tier paperlaunch with consumer SKUs coming Q2, maybe even Q3.

We talking about V100 or the 1180? Because naturally the former will launch much latter

That's exactly what it means

7 nm PaJEET = 45 nm intel

IBM are pajeets now? Kike pls you fear IBM like fucking ovens.

kek

IT'S NOT FAIR GUYS

First of all can we even get a source on this or is this another "wccftech internal source"

Because I really wanna see what Navi and Zen+ are about instead of the newest Machine learning or car meme

POOR VOLTA

>Machine learning
They will talk about it, because muh Radeon Instinct.
>car meme
They won't.

No cars? Thank fuck AMD hasn't lost its marbles yet.

They have no time and money to waste on meme cars. They also like to do actualy R&D like new architectures or new, better memory like HBM or heterogeneous compute or anything but the fresh how memes.

IBM are kikes though

More like the bane of kikes

Are people expecting it to beat the 1080ti? I've been expecting a gpu aimed at the 1070, one aimed at the 1080, then maybe a dual gpu solution to compete with the 1080ti like what they've pulled in the past.

Based on its die size, clocks and maxwell-like improvements, yes.

>wccftech

Dude it's 530 mm^2 die, while STILL having 4096 ALU. Guess where ALL the free die space went to?

i've been saying wait for navi for a long time now
it's going to be for amd what maxwell was supposed to be for nvidia
hopefully they don't relly on false advertising and shitty marketing tactics

Interesting trivia, HBM2 chips have a part of the controller/PHY integrated into the base die itself instead of the GPU, plus it's 2 stacks instead of 4 like Fiji so I imagine there's a lot more room for shit in there.

HBM1 apparently wasn't like this.

>it's going to be for amd what maxwell was supposed to be for nvidia
That's Vega.

Huh? If anything Vega has the Maxwell features + some of its own shit

Not him but Vega will be Fiji 2.0

Anyone got a feeling Navi will be another big arch change? Not like Vega of course but still pretty big?

How fucking so? You can't read?

Wait for Navi ™

Me right now

Navi is supposed to bring advanced packaging methods to GPUs, read MCMed GPUs. Which is awesome indeed for everyone but true horror for NVIDIA.

I'd like a good explanation why you think that, what you're implying is that Vega is Tonga x2 +14nm

That would be Fiji with 6000+ cores, it's obvious the entire point of Vega is increasing performance per individual ALU

No, the point of Vega is making a frontend that DOESN'T SUCK past some 2.5k ALUs.

Pascal is a Maxwell rebrand at 14nm and it is ~60% faster, why not the same with Vega?

Because Maxwell has non-shit frontend that can actually process geometry without choking? Also pre-Vega GCN has no tiling of any kind.

This meme should stop, Pascal aren't rebrands

Because it's not? Pascal has both more cores and higher clocks than Maxwell, Vega keeps the same amount of cores.

Okay, Pascal is dieshrink with clocks upped to 11.

I'd wait for Navi if I had a Polaris, but since I don't I'm getting Vega.

Read the whitepaper. You couldn’t be more wrong.

It's really impressive that we're 3 weeks from the Computex reveal and we know nothing yet.

>bane

the masked man?

No, bane.

It's Maxwell 2.0: more clocks edition (with VR memes).

You don't know numbers, you know plenty about the arch already, you knew it 5 months ago.

Fuck wrong post.

>poojets
>reading anything
They only spew marketing ayymd buzzwords like async and hw sheduyler without a modicum of actual knowledge and understanding.

Ramesh pls stop.

There are Fury(nonX) Firestrike Extreme benchmarks out there at 1450MHz and it scores around 10500 graphic score.

Which is hella impressive for baseline Vega performance that's gonna be clocked both higher and with new arch changes + 500 more cores

AMDegenerate tears are delicious

Are you ok?

...

I can't believe LN2 scores have ever been useful until now.

For comparison a stock Fury X gets around 7600
guru3d.com/articles-pages/amd-radeon-r9-fury-x-review,25.html

just w8 guys, you just w8

GlobalFoundries is headed by an Indian tho.

As of now they are IBM's buttsluts.

Can we get a thread about this that wasn't started by and infested with retards trolls and shills pls. I guess that's too much to ask for the most intelligent technology board on the internet.

...

Maybe IBM will go back to old IBM and gas the filthy, slimy kikes

>mfw I waited™ for Tahiti

fucking poojets can't shit out a proper OGL driver if their subhuman lifes depended on it

Why do you a need a proper OGL drivers (read collection of hacks and workarounds to make it work)?

Tahiti was godly if you managed to get a 1300+ OC, my friend's GHz edition(I think MSI) managed to hit 1370MHz

What a beast.

Tahiti was so good at killing kikes it still does that.

Do MSi still make the Twin Frozr cards? I remember during that time they were all the rage

they are godtier and Twin Frozr is still kicking

I think they rebranded it as Gayming X.

Gross.

Everyone likes shiny RGB 'gaymen' things, right?

What more of Vega can they show us? I do recall Raja saying there's still more architectural features they didn't mention last time.

I wholly wish I could disagree, but I can't.

The architectural thing they have showed us was HBCC working: they launched and played DXMD with 2GB VRAM.

Not that, I meant the slides where you had primitive shaders, L2>ROP association, TBR, configurable precision rate, 2x geometry throughput per clock, etc etc
Which are all pretty big changes.

It was in a interview slightly after that event, he mentioned that those changes aren't all the changes that are going into Vega, so if there's more big stuff we may be in for a surprise

>primitive shaders
What's that? The rest is simple as shit and basically is Maxwell: AMD edition. But what is primitive shaders?

JUST WAIT

Primitive culler

It's another feature to boost geometry performance in its most basic form, we know really nothing about it though.

Ah fuck Raja give us at least SOME details about Vega. What's the point of being SO secretive about it?

Because any info just makes the hype train run wild and creates unrealistic expectations.
Better to just keep quiet, this has burned AMD quite a few times.

But i want techincal details. I want to see Vega, not wait for it eternally.

WaitTM forTM ComputexTM

Get a 1080ti unless you can't wait 3 more weeks