Qualcomm to announce Snapdragon 835 at CES

>manufactured on 10nm Samsung process
>40% performance improvement over Snapdragon 821
>30% longer battery life over Snapdragon 821
>HTC 11 will be the first Snapdragon 835 device to ship

How do you feel about this?

>30% longer battery life
Which somehow makes companies think 30% smaller batteries

If they make it to 10nm before Intel I will be impressed.

But 1.3^–1 = 0.77, user. That's only a 23% smaller battery.

You have to keep in mind that a mobile ARM chip is a few mm^2 big, a 4 core x86 CPU oder 200mm^2

[grain of salt intensifies]

They already did.

I just ordered a new phone :(

With how little progress Intel has been making due to AMD being terrible
it's only a matter of years until mobile chips are fast enough for 99% of computing uses

I didn't want it to be like this

Depends how hot it runs

MediaTek has some 10nm parts coming as well.

14/16nm high end ARM SoCs were around 70mm2 to 100mm2.
Mainstream desktop Skylake has a die size of 122.4mm2.

They are not that far apart.

Thats the real problem...

>intel doesnt make progress because AMD is bad.

I don't give a fuck.

No competition equals no reason to spend funds on innovation.

This hasn't been the case. Intel has been heavily spending in X86 core development, and they've been utterly dumping millions into their foundry business to try and maintain a perf/watt edge.
They're not making substantial gains in IPC because they can't. What they're putting out is the best that their engineers can muster. The core architecture has been evolving since the P6. They've plucked every low hanging fruit on the performance tree, now all that is left is few and far between. Every 1% performance uplift costs more than the last. Its diminishing returns, not a lack of competition.

Intel's position at the moment has nothing to do with a lack of competition from AMD. Coffee Lake will launch and add 2 more cores to the mainstream product line up. Cannonlake HEDT will eventually launch, still no large IPC uplift. Then intels 10nm refresh chips will launch, again with no substantial IPC uplift. The time frame for that is 2019-2020. 2.5 years after Ryzen launches. AMD will have 7nm SOI FinFET Zen+ parts on the market, and intel has nothing in their lineup which promises major performance gains.
It isn't going to happen.

Can't wait for the Surface Phone containing the SN835 to come out.

It's not a true 10nm process tho.

Exynos 8 is going to shit all over it and then eat it for breakfast. It won't be anything remarkable since it'll be yet another overheating steaming pile of shit.

Stop regurgitating this retarded fanboy nonsense.
It fits what ASML is calling a 10nm class node.

It's not nonsense when the only part that is 10nm is the transistors. At least Intel is not being deceptive when that call a 10nm process 10nm.

Again, stop regurgitating retarded fanboy nonsense.
The public facing process name is named after the relative gate length, this is always the case. It has never had anything to do with back end scaling.
You don't know a single thing about the topic at hand. You're blindly regurgitating bullshit from a schizophrenic on Anandtech.

Fuck off samshill.

Intel has always labeled their process accurately. This has nothing to do with anandtech or whatever nonsense you think I believe.

Stunning rebuttal, tech illiterate retard.
Take your childish fallacies and put them back up your ass where they belong.

>HTC 11
will this one be +10 deg c hotter?

Fuck off samshill.

...

gate length has nothing to do with the naming scheme, retard

>Intel has always labeled their process accurately.
This is a ridiculous, and factually incorrect statement which shows you don't understand what the fuck you're saying. For instance intel's 22nm Trigate process creates gates which are actually nominally 26-27nm wide.
Intel chose to call it a 22nm for marketing purposes. It had considerably bad end scaling advantages over 28nm planar nodes from other foundries, but was still considerably behind yet to come 20nm planar nodes.
I wouldn't call a 26nm~ gate being marketed to investors as 22nm "accurate."

If something fits ASML nomenclature, if the resolution on some of the masks aims for a 22nm structure, but that doesn't quite materialize in silicon, its perfectly acceptable to call it as such. This has always been the case. The transitional FinFET nodes from TSMC and Samsung fit ASML's definitions for a 14/16nm node. Their respective 10nm nodes also fit ASML's definition of a 10nm class process.

You're a moron, clueless, regurgitating someone else's opinions.

Once again, you are the clueless one regurgitating someone else's "opinion"

metal pitch and poly pitch are back end. Good job.

Yes, it does.
You couldn't say otherwise unless you are totally taking a shot in the dark. Bluffing like the ignorant autistic subhuman you are, desperate to argue for the sake of argument.

You're such a clueless tech illiterate retard.
Try again.

wtf am i looking at here haha

Everyone knows Intel has better density and a more accurate representation of the process node in their naming scheme. They just don't plug in a finFet into a 20nm framework and call it 14nm like GF.

>Everyone knows Intel has better density
That isn't relevant and no one is disputing it, tech illiterate retard. The issue at hand is your fallacious bluffing, saying other foundries aren't offering a "real" X process. You've amply demonstrated that you don't understand any basic process metrics, you don't understand why ASML sets the guidelines they do. In fact I would bet good money that you have no idea what ASML is, and you'll probably google it in a panic as soon as you read this.

>They just don't plug in a finFet into a 20nm framework and call it 14nm like GF.
The laughably ignorance here is too much. You must be here straight from Anandtech to be this dumb.

Mentioning Global Foundries here at all is just icing on the cake for proof that you're parroting things you don't understand.
Samsung and TSMC created transitional FinFET nodes. Smaller node FETs built on the BEOL of their prior 20nm planar processes. This is the "framework" you're referring too, but lack the IQ and education to understand what any of this means. The idea for these nodes came first from ASML.
Intel having better back end scaling at 14nm doesn't preclude them from being "real" 14/16nm processes. They are in line with a set of guidelines set forth by the industry body who creates the tooling that everyone uses.

Global Foundries is only offering Samsung's 14nm process. They had nothing to do with its development at all.
Please, keep embarrassing yourself. Its hilarious.

22nm Intel

You are the one sperging out at deceptive naming schemes. And yes, they just plugged in FinFets into a 20nm framework and called it 14nm. Keep defending inferior korean process nodes; it's truly hilarious.

>I can't even try to make an argument anymore so I'll just blatantly shitpost from now on

Thanks for admitting you're outclassed, retard.
Maybe go to college and actually learn something for once in your life.

>damage controlling this hard

your own meme image shows that the 14nm processes from samsung and tsmc are shams. we won't see 'true' 14nm until their 10nm processes launch and intel will continue to maintain their process advantage.

>desperately grasping at straws
The industry tends to follow ASML's naming conventions. You don't get to pick and choose which you think are valid and invalid.
The 14/16nm processes offered by TSMC and Samsung are 14/16nm class.
The 10nm processes they're offering are 10nm class.
You trying to say they aren't "real" is pure fanboyism. The best part is that you're a fanboy for something you don't have the education to understand. You have no EE background, you're just a dumbfuck shilling for a company you don't even have a financial stake in.

The desperation of an intellectually cornered tech illiterate autistic retard is a sight to behold.

HTC 11 will feature this and yet it won't even be released to America because fucking no one bought it here except me and one guy in the general

Well you explain it, since you are sooo smart and educated. What I have said is so well documented, it can only lead me to believe that you are the one that has no idea of what they say. All you have really done is post how smart and educated you are. GF's 14-nm chips aren’t any denser than- and therefore cost just about as much as the previous generation. Naming something smaller implies the whole node shrinks, not just part of it. At least Intel is not deceptive in this way. This new way of naming nodes was the Korean's hope at catching up to Intel's manufacturing lead in name only.

“It is quite a controversial move,” says William Arnold, chief scientist at ASML, the world’s largest maker of semiconductor-fabrication equipment. “The customers of the foundries, the people who are making cellphone parts, are very skeptical of not being able to get a shrink along with a performance improvement. They’re pretty vocal about saying that they’re not happy about that.”

----Talking about the Korean's fake names 14nm process (14nm finfets only)

>The industry tends to follow ASML's naming conventions. You don't get to pick and choose which you think are valid and invalid.

nobody cares what a random company sets as the standard

>The 14/16nm processes offered by TSMC and Samsung are 14/16nm class.
>The 10nm processes they're offering are 10nm class.

not in comparison to the process tech from the industry leader.

> You have no EE background,

resorting to projecting when people don't gurgle the marketers cum like you do? good goy

>Well you explain it, since you are sooo smart and educated.

he isn't educated. he's an angry NEET who gets mad at anyone who criticizes his favorite companies (amd and glofo). just search any archive for posts with 'tech illiterate retard' in them and it's guaranteed to be one of his.

How much are you paid to post FUD here?

>bring up Global Foundries yet again
You're now fallaciously trying to make claims about cost per transistor, and transistors per mm2, and total cost of a design. Stop trying to argue about things you know nothing about.


Thats a nice completely irrelevant quote. Highlighting how butthurt and desperate you are.
A speculative statement from someone years ago about prospective customer's perceptions has nothing to do with factual process metrics.
I can tell you're desperately googling things right now. Because you're a mentally retarded autist. The type who has a compulsion towards arguing for the sake of arguing.

>nobody cares what a random company sets as the standard
ASML is not "some random company" and yes, literally everyone in the industry uses their measure. Foundries competing against one another uses this bar and try to surpass it by a certain percentage in various metrics so they're better positioned against the competition.
Yet again, this has nothing to do with your pisspoor argument. You made a number of factually incorrect statements to deflect away from how weak your position is. Saying something isn't a "real" node is retarded fanboy nonsense.

This projection is amazing.
Yet again I have to remind you, since you're a tech illiterate retard, that the two parties primarily being discussed are TSMC and Samsung. This has nothing to do with Global Foundries. Global Foundries did not develop the 14nm process they're running at Fab8 in NY. They are running Samsung's 14nm process, entirely developed by Samsung.

You're childish, and your lack of education is woefully apparent.

I can't believe how much two people would spend arguing with each other over measurements for products they had no implication in developing.

>ASML is not "some random company"

they are, retard. being a large company does not make them the law. even your own favorite company ditched their standard with their supposed 7nm process (actually equivalent to 'real' 14nm).

>Global Foundries did not develop the 14nm process they're running at Fab8 in NY. They are running Samsung's 14nm process, entirely developed by Samsung.

nobody was claiming otherwise, dipshit. the only thing in dispute is the naming.

Intel: We developed a smaller node with better density and efficiency. We'll call it 14nm

AMSL: Ok, cool

Garlic Eaters: We made part of our chip smaller with nearly the same density as our old process, but now its a bit more efficient. We'll call it 14nm.

AMSL: ..ummm...OK guys, whatever, just keep buying our shit and you can name it whatever

>I keep backing myself into a corner so I have to blatantly lie to keep arguing
No, ASML is not some random company. They are the biggest driving force in the semiconductor industry because they make the overwhelming majority of the tooling used in photolithography. They don't hold industry conferences on the future of the market for their own sake.

You keep trying to bring up Global Foundries for no reason. Like the child you are, you tried to paint me as if I was a GloFo fanboy when no one here mentioned them save for yourself. Ridiculous. Your petulant childish autism knows no bounds.

More like:
>ASML: These are the guidelines for what defines an X class node
>ASML: Oh, it looks like cost per transistor is reach an enormous inflection point beyond 28nm planar. We need to do something about this.
>ASML: With short channel effect entirely out of control at 20nm planar, and the financial investment made here, if this isn't salvaged the entire industry will feel a ripple effect and costs for everything from micro controllers to smartphone SoCs will sky rocket. We need to contact everyone in the Common Platform Alliance and TSMC.
>ASML: Basing a smaller FEOL on the prior backend will decrease cost to deploy in addition to increasing time to market while giving the performance adders of X,Y, and Z.
>IBM: We're getting rid of our foundry business so we don't really care. We were entirely SOI anyway.
>GloFo: We cancelled our own 14XM process.
>Samsung: Fine we'll do a 14nm Transitional FinFET node.
>TSMC: Fine we'll do a 16nm Transitional FinFET node.
i>ntel: We spent enough money and jumped on FinFETs before everyone so this isn't a problem for us.

Its like you really don't know anything thats been going on in the industry for the past 10 years.

ASML: buzzword buzzword buzzword
IBM: Fuck this
GF: We suck
SS: We suck
TSMC: we suck
Intel: Always first to cumm

I didn't know Lisa Su worked for Samsung

>They are the biggest driving force in the semiconductor industry because they make the overwhelming majority of the tooling used in photolithography.

being the biggest does not make them a driving force, in fact it's telling about their status and quality of products when all of their customers are behind the industry leader to the point of redefining their naming scheme and universally pushing back their roadmaps as they struggle to catch up.

Just give up. You can only bluff for so long. I've known you were lying and talking out of your ass from the very start.
Intel too uses ASML tooling. You not knowing this is proof that you're doing nothing but frantically trying to construct nonsensical arguments on the fly.You're clueless. Outclassed. Utterly lost.
Its pathetic. Just give up. I'm near blind drunk and I'm still far more coherent and informed than you'll ever be.

>Intel too uses ASML tooling.

[citation needed]

And Android will still find a way to make it lag.

It's 40% performance improvement OR 30% more efficient, idiot

Intel is ASML's largest dedicated customer you fucking retarded child.
At least gain a cursory understanding of these things before you try to argue with someone leagues more intelligent than yourself.

>le ebic android lag meme XDEEEEE

The difference is that when these problems occur on iOS they actually get fixed and fixed for everyone instead of the small subset of Android users who actually get device updates for more than a year after purchase. No, I do not care about your custom ROMs, as this is about what everyone else does.

Can't wait to see their keynote.

>he is not generation mobile

if that shit doesn't play 2160p or 1080p 10-bit HEVC then I'll shit on it

>this butthurt samshit shill itt

It will sell like crack if

>5.2 - 5.5" amoled or OLED screen
>3600 to +4000 mha battery
>Same sexy design
>Stereo front speakers
>Headphone Jack
>Capacitive buttons + Fingerprint reader

They know the dick size of the average /g poster? How is this possible?

if this is even remotely real then the next generation will be quite good.

That's almost what my HTC 10 is

This isn't hard. Normies are clueless on specs.

>Garlic Eaters:

Confirmed for butthurt Jap.

When is Renesas getting a working 28nm fab?

>Intel has always labeled their process accurately.
>not a single feature on intels cpus are 10nm

What did he mean by this?

What the fuck are you on about?
.3 is only 22% of 1.3?
Shit, you're right...
Also, do they not understand how much extra money I would pay for better battery life and open firmware? It's like they don't want my money.