How can Intel even compete with this coming in around 12 months?

How can Intel even compete with this coming in around 12 months?

Meanwhile in 10nm land

digitimes.com/news/a20170920PD207.html

Other urls found in this thread:

twitter.com/PatrickMoorhead/status/910544559955861504/photo/1
tomshardware.com/news/amd-ryzen-vega-12nm-lp-2018,35502.html
globalfoundries.com/news-events/press-releases/globalfoundries-introduces-new-12nm-finfet-technology-for-high-performance-applications
twitter.com/AnonBabble

They can't. If anything this ensures AMD will finally get a foothold in the server market again.

THIS IS ANTI-SEMITIC

Its going to be interesting to see just what materializes in the Zen2 based parts. Unless transistor count significantly increases in Zen2 then power should plummet. We could have a higher performing Ryzen 7 1700 at just 35w.

>7nm APU with Zen2 + Vega/Navi
MAKE IT HAPPEN

What the fuck Intel.
How the fuck do you JUST your fabs so much?
They're about to become GloFo's personal designated shitting street.

I-i'de like to see thread ripper beat the 18-core $2000 i9... Goyim

6 gorillon in diversity. oy vey!

NOOOOOOOOOOOOOOOOOOOO

Transistor count means nothing, there are now transistors solely dedicated to REDUCING power consumption, Zen has plenty of them.

that table is misleading, the paper says >40% more performance at the same power OR 60% power reduction at the same clocks

Theres nothing misleading about it.
The new architecture is going to have higher IPC and performance at any given clock will increase.

twitter.com/PatrickMoorhead/status/910544559955861504/photo/1
twitter.com/PatrickMoorhead/status/910544559955861504/photo/1
twitter.com/PatrickMoorhead/status/910544559955861504/photo/1

I TOLD YOU NIGGERS "zen+" is a THING I FUCKING TOLD YOU
PEOPLE SERIOUSLY THOUGHT AMD WOULD GO FOR AN ENTIRE YEAR WITHOUT SOME KIND OF PERFORMANCE REFRESH WHILE THEY HAD THE INITIATIVE HOLY SHIT

Papermaster saying something (well, anything) has nothing to do with hypothetical Zen refresh.

a72? so amd will finally bring arm cores along with zen?

k12 is finally here?

You're just massively confused.
There is no Zen+.

Zen+ was a placeholder name AMD used for the next generation of architecture when they were first talking about Zen. Lisa Su confirmed in an investor call that successive generations in the Zen core family are called Zen2 and Zen3.

Global Foundries here is just continuing to offer Samsung IP here. Its nothing new.

K12 is not A72.
Its fully custom, Aarch64 core.
Maybe they want to kill any chance of Qualcomm ever entering the datacenter market.

What do you think they'd use 12nm LP for if Zen2 is 7nm?

They are not going to use TSMC's nodes for Zen.

Embedded parts. Their ARM line. Semi-custom.
Stop trying to read too much into things.

literally nothing since they both start risk production at the same time

you do realise that k12 is exactly that eh................

We followed up with Papermaster in person and confirmed directly that the company will transition both Vega GPUs and the Ryzen line of processors to the 12nm LP process. However, it’s still not clear whether or not he meant that 12nm LP will be a shrink of Ryzen in 2018 (a "tick" equivalent if you will) or if Zen+/Zen 2 will also be using the 12LP process. Previously, AMD has implied that Zen 2 will use the 7nm process. The company has used both "Zen+" and "Zen 2" to refer to its next-generation die.

K12 is not related to any reference ARM designs you moron.
It's fully custom, AArch64-compliant core.

tomshardware.com/news/amd-ryzen-vega-12nm-lp-2018,35502.html

Wait, so GloFo was making a custom node for AMD A G A I N?
*looks at 28sHP*

so RR and Vega 20? both stuff that was supposed to come in 14nm+ before?
dumbass

Anyway YES, mobile Vega11 can't come soon enough.
Fuck nKikia and FUCK Optimus.

Improved nodes are used for new SKUs all the time. They might fab raven ridge on these

And Vega11, and Pinnacle Ridge, and probably refreshed Vega10 because porting it would be trivial.
They are probably already fabbing it.
12LP will murder any node advantage KBL-R has.

neither is a72 retard

whitepaper when? I want to read

Holy fuck it's the first time rambling schizo that is seronx was right.
Cortex A72 is a reference ARM core you braindead moron.

A72 is a ARM Holdings reference core you retard.

14nm+ renamed to 12nm ?

Is it a derivative of a 14nmLP or a custom non-samsung node ?

Anyway, Vega 11, Vega 20 and probably Raven Ridge are fabbed on this.

Its like just a tweaked Samsung LPc.
Samsung has had higher density variant of 14nm LPP for a while now.

Guess so. TSMC calls their improved 16nm, 12mm so must GloFo. Marketing business

It's GloFo's own node.
>Anyway, Vega 11, Vega 20 and probably Raven Ridge are fabbed on this.
Vega20 will still probably be 7HPC UNLESS they pushed the reticle limit forward.
14LPC is cost-optimized solution.
It's still incremental gains in performance or power.

Tweaked 14nm, as expected.

But this will be enough to shave off Coffelake, if AMD can at least get 4.5GHz and some small IPC improvements it's more than enough.

>15% more performance than 16FF+
Oh boy, this will clock pretty high.
Just look at Pascal.

>It's still incremental gains in performance or power.
Of course, it makes quite a difference for mobile so they are labeling it for intelligent, embedded applications etc

a few sources indicate 15% higher density and 10% higher performance compared with 14LPP, seems like a really marginal improvement not worth of a refresh

No, they are labeling it that because of their RF and SerDes libs inherited from IBM.
Exactly enough to fill product stack for a year.

It's more than enough than sitting on their ass on Ryzen's released in early 2017 until Q1 2019

If it's built off 14nm LP then it shouldn't be hard to port, and that's another year of even better Ryzen chips until 7nm

And since Intel is supposedly shipping Cannonlake only late 2018, AMD won.

%10 higher perf, so 4.0GHz base clock ryzens?

risk production of 12LP and 7LP start at the same time, this is not an argument

The yields.
7LP DUV will have atrocious yields for a year at least.
15% higher perf with trivial porting.

Okay, then there's not a full 7nm in 2019, most likely due to fab capacity or yields.

>full 7nm lineup

Intel will compete by releasing coffee lake and letting the perforemance speak for itself instead of hiring poo in the loo influencers
Fake coffee lake leaks won't make Ryzen IPC, clocks, or memory compatability any better, and having an extra 4 threads won't save FX: Ryzen edition from being btfo in every category

OY!
JUST
FUCK
MY
FABS
UP
Noyce is literally rolling in his grave.

>glofo can't into updating their own website so we can read about 12LP
fucking slow people

They've literally just announced it.
Have patience.

“We are pleased to extend our longstanding relationship with GLOBALFOUNDRIES as a lead customer for their new 12LP technology,” said Mark Papermaster, CTO and senior vice president of technology and engineering, AMD. “Our deep collaboration with GF has helped AMD bring a set of leadership high-performance products to market in 2017 using 14nm FinFET technology. We plan to introduce new client and graphics products based on GF’s 12nm process technology in 2018 as a part of our focus on accelerating our product and technology momentum.”

>graphics products
Oh look, a Volta competitor!

>“It builds on 14nm,” said Sanjay Jha, CEO of GlobalFoundries. “It pushes new designs and constructs.”

>He noted that shipments are expected to begin in the first half of next year.

>pajeet is literally making Intel his designated shitting street

lol

>LP
Nope.

7nm is also called LP you know?

Leading Performance, yes.

>seems like a really marginal improvement not worth of a refresh
Tell that to Intel

LP is low power, Rypoo struggles to go past 4ghz so Intel will eternally be the superior choice. Deal with it, poos.

shoo shoo pajeet
LP is Nazi (i.e. IBM) speak.

SOPA?

[Hebraic howling intensifies]

>people complaining about a small perf uplift
Reminder: A 10% increase in clocks would let a low end Ryzen 5 or Ryzen 3 with 3.2ghz base clock get boosted up to 3.5ghz.
A 3.4ghz base clock goes up to 3.7ghz.

Thats significant enough to keep an edge in benches.

doesn't matter without any IPC improvements and the voltage wall still there

It's actually a 15% performance uplift.

Every category? You mean 1080p and 720p gaming. That's the only place Intel wins. But nice try madfag

Even better.

it's 15% density, 10% performance

Higher clocks mean higher performance in general.
Oh, sorry.

>Higher clocks mean higher performance in general
Pentium 4 pls go.

Not everything has a 31 stage bibeline you know.

fx 9590 pls go

You don't think the voltage wall won't change at all with this new refinement? I'd say being able to push at least another .300ghz to 4.4~ on the top end chips sounds pretty good.

the newer Stepping already hit 4.1ghz regularly, with 4.2ghz being a golden chip.

We could see 4.65 regularly and 4.8ghz golden chips.

I won't say shit until I read the paper (which glofo didn't release yet, fucking damn)

I HATE AMD
DIE AMD DIE DIE DIEEEEEEEEEE

That would bump Ryzen and Threadrippers clock ceiling to >4.5Ghz if true.

now I want to see the intelfags saying that 7nm ryzen won't be able to reach 5GHz on air

Any day that 300 million they invested in feminism is going to pay off.

amds roadmap has a 48 core cpu in servers, that would mean an extra 16 cores total

unless amd makes a 6 die mcm, they are adding 2 cores to a ccx

also, the node is targeting 5ghz, so that will play some part in increasing power use too.

yeah, amd will benefit a lot from that actually

RYPOO

globalfoundries.com/news-events/press-releases/globalfoundries-introduces-new-12nm-finfet-technology-for-high-performance-applications

>still no paper

Serious question

How does cpu architecture work and differentiate?

Ask Google