Jesus Christ

You know that those measurements are mostly pr and probably have fuck all to do with the actual gate size?

And then you woke up.

It's not like their 3nm is actually 3nm though, it's all marketing bullshit

>tfw china's (umm ackshyually Taiwan's) approach to education will make the west irrelevant and poor in your lifetime

It's not fair, bros. I wish I had been forced to study properly.

What approach ? You mean stealing, cheating and lying ?

Who did they steal 3nm from?

Based. Fuck the Soilomon islands and fuck Kiribati.

Major cope

genuinely curious as to how you can mitigate critical errors though with quantum tunneling happening

just stop investing in smaller silicon processes and invest into materials research and compact cooling for those near-room-temperature superconductors

reminder the node names is all marking bullshit
TSMC 7nm= Intel 10nm.
TSMC 5nm=Intel 7nm
TSMC 3nm=Intel 5nm

Xi Jinping is Winnie the Pooh and the whole world is honey to him

It's better than 14nm+++++ anyway.

Intel's hypothetical 10nm maybe. They couldn't make it so they made some parts bigger.

Why are Americans always such jealous cunts ?

No. Just no. There is no "quantum" anything, this isn't poorly understood near magic effects of some mythical theoretical particle.
This is simply electrons being so small they can move through any material at the path of least resistance, because nothing can exert 100% perfect electrical control over them. It is current leakage. It is nothing but current leakage.
It is current leakage in short channel devices, and it happens at literally every feature size, it is not exclusive to small FinFET devices like upcoming 5nm EUV FinFETs. Even planar devices have extremely high degrees of leakage through their channels, directly under the gates, electrons still leak out.
Yet despite this the transistors still function.
Quantum tunneling is a meme regurgitated by people who know nothing about the field of FETs.

Fucking based

I know this is pasta but can someone refute him?

BASED

This guy gets it.

Process node does not matter when Intel is still kang

Why is Xi holding the flag of Taiwan? This is treason.

Are you up tho?

because they are white trash now

Zen 10, any day now bros

he's right, quantum anything is just bullshit for the scientists to get funding

This guy has quantum delusions. Go take your quantum meds.

always heard this, why do they lie? why not put the real number?

>meanwhile Intel is STILL researching 14nm

Future chinks stole a time machine made by white people and came back with stolen future designs made by white people

I've heard some scientists want to embrace the tunneling and make use of instead of fighting with it.

The leakage that these dielectric structures exhibit is entirely a quantum phenomenon: there is a distinct probability for any charged particle to tunnel through the barrier described by the change in the potential going from the conductive substrate through the dielectric. The two scales that play into this are the applied electric field and thickness of the dielectric (noting the relation that E ~ V/t). The leakage issue worsens as features in the transistors shrink, and this is unavoidable. Decrying what is a textbook case of quantum mechanics as the usual misuse of the word is just as bad as its use as a buzzword by retards.

Intel whatever nm is marketing bullshit, too.

>all these tech illiterates falling for marketing terms
Never ceases to amaze me.

based

>TSMC 7nm = Intel 10nm

Attached: Behind.jpg (491x491, 42K)

Why not just make the darn CPUs bigger?

Attached: 329854854_9ad6ef54.gif (347x285, 278K)

Just use the blockchain bro

>There is no "quantum" anything
brainiac

I thought it was impossible because of physics?

I have news for you, kid:

Attached: One-waffer CPU.jpg (335x314, 76K)

Quantum everything is a fucking meme if you believe that kike garbage you've probably been drinking too much kool aid.
Believing in quantum garbage and relativity is the basedest thing you can do.

And they're making money from it too!
Maybe AMD could ,you know, make some chips for laptops. my Dell xps13 runs for 10 hours on a 4c8t i7 and that turbos beyond 4ghz. Surely AMD should have that matched at least

aww yeah

soiest*
Get fucked mods

Cringe
Based

>he leakage issue worsens as features in the transistors shrink, and this is unavoidable.
Except you're 100% incorrect and talking out of your ass. Leakage current increases with smaller xtors as part of the short channel effect, which is mitigated with different transistor topology. FinFETs today have *less* leakage current than any planar transistor ever made. Future GAAs will have less leakage current than any FinFET. This is not just proportional, it is ultimate.

Quantum behavior is fake and gay.

Daily reminder that Intel's 10nm is currently in full production for Xeon and the smaller mobile chips.
And Intel's 7nm is actually on track.
Apparently the 7nm team wasn't full of incompetent retards like the 10nm was.

The only reason you're not getting another 14nm next gen is because of AMD. You should suck our dicks Intel kike drone.

based EEchad

Attached: CPUguys.jpg (763x1100, 332K)

It's impossible yet they're still doing it

Whats hilarious is that there are annual conferences where all these industry firms gather and discuss methods and materials to keep progress advancing. They go in depth on new types of transistors, what materials need to be used in what metal layers, new tools or design methods to circumvent potential hangups, etc. Every tiny detail down to financial viability.
The typical brainless internet news sites totally ignore all of this and instead run clickbait articles about Moore's Law being dead or who muh tunneling is going to stop all scaling.

We're sitting here now, 20nm MMP is on the horizon, and fags will still be bloviating on how we need graphene to save us. Absolutely ridiculous.

It's about density too. Intel 10nm is slightly more dense than tsmc 7nm. tsmc 7nm is still better though. We know nothing about the density of the planned 3nm as far as I know.

It isn't a "lie." The entire industry is going by back end metal pitch scaling. Gate length being the defining feature of a process has not been a thing since the late 1990s.

It's for machine learning btw

>He doesn't know about transmigration

Attached: 1562029446095.png (872x550, 369K)

>It is nothing but current leakage.
t. doesn't understand how transistors work

Go migrate to the dilation station

>NOOOOOOO YOU CAN'T GO UNDER 5NM, YOU WILL GET QUANTUM TUNNELING ISSUES AND YOU'LL HAVE TO MITIGATE IT WITH ERROR CORRECTION WHICH WILL HURT PERFORMANCE

It is 3nm++++++++++++++++

this confuses the chink

Taiwan? You are just delusional. That is clearly the flag of the People's Republic of China comrade.

Attached: 1567892086885.jpg (640x640, 61K)

cringe

This.
Take him to the infirmary.

Intel already lost the marketing war, every normie knows that lower nm is better.

Not really my friend.
Normal fags* have no idea what NM even means or any it matters.
Furthermore, the server space and business know the numbers are largely marketing.
And the server space is where 90% of the money comes from. Desktop is pennies in comparison.

Meanwhile, Intel's 14nm+++++++++ chips are failing early and being recalled 10nm volume production may or may not start ship before 2020.

>And the server space is where 90% of the money comes from
So we can all agree Intel is finished

>tfw single core performance doesn't matter unless your a gaymer

Attached: 1561769153520.gif (640x358, 1.94M)

> Average Yield : 0%

Apparently its designed and built with a shitload of redundancies in mind. Also, IIRC it has a multi-kilowatt power rating. It makes the 400W monstrosities intel made look like little babies.

based chinese

It's literally "one wafer is one CPU = one CPU is 100% yield", you retards.

Wrong

It's funny because Intel has none of those.

Attached: SAVAGE.jpg (980x1200, 187K)

The future is bright

Attached: Fig.-1.png (997x487, 81K)

(Non-Copenhagen) Quantum is goyish, Relativity is Jewish.

All I care about is how fast 4 cores can go and Intel still has that lead at 14nm. Kek.

Attached: 1564312348244.png (593x445, 105K)

retard

only single core performance matters because video games and software (besides some video rendering shit) are written to use only one core because programmers don't give a fuck about multithreading

Jesus this is just pathetic. A hundred years from now white people are going to have their own Yakub myth

Based bait poster

Reminder: 4T and 3T high perf HD SRAM cells are a reality the entire industry is gearing up for.

Attached: nextgenfig4png.png (1045x609, 243K)

>I have a PHD
kek

The yield isn't that cut and dry. If two adjacent tiles/cores on a reticle/die are defective, they have to throw away the row and column of dies. The part will still work, but there'll be considerably fewer tiles/cores for computation.

t. Cerebro

The fact that TSMC is outperforming the western entrenched companies only proves that they were a monopoly so far and are purposefully not engaging in the needed R&D for their valuations

It's ok, Intel will finally have 10nm by the time they release it. It'll definitely beat a two-node advantage.

Wakey-wakey user

I wonder what the thermal output is.