Is Brian Krzanich the most incompetent Intel CEO ever? First they got BTFO by AMD Ryzen...

Is Brian Krzanich the most incompetent Intel CEO ever? First they got BTFO by AMD Ryzen, now everyone seem to be getting an upper hand in the fab process.

Attached: Screenshot_2018-04-28_14-06-40.png (709x719, 544K)

Other urls found in this thread:

en.wikipedia.org/wiki/Intel_Ireland
asml.com/products/systems/twinscan-nxe/twinscan-nxe3400b/en/s46772?dfp_product_id=10850
asml.com/products/systems/twinscan-xt/en/s46772?dfp_product_id=10272
twitter.com/NSFWRedditVideo

>yield issues
in other words intel's dies are too big

Exactly. From a defect density perspective they are hosing themselves with monodies.

>Is Brian Krzanich the most incompetent Intel CEO
No.
He keeps stock holders happy.
He keeps intelavivdiots busy with rebranded sandy bridges
Intelaviv has profits
Everything is fine.

Also in the ER there was talk of quadruple/quint/hex patterning. I do not see how this is even remotely feasible without an idealized average step defect rate, which at hex patterning basically requires a perfect clean room, 100% flawless tools, and the purest consumables in existence to minimize incremental defects over the huge number of implied process steps.

Most other fabs bullshit on size, Intel 10nm is smaller than everyone else's 10nm

Smaller, and also impractical seeing as they're unable to bring it to mass market.

He's somewhere there, yes.
At least he made Intel swallow its pride and bring in some external talent.
They can't yield full 2C CNL.
Die size is not the problem.
But can Intel fab it?

>10 nm here is less than 10 nm there

Don't forget that 10nm from Intel was planned for 2015.

And that in late 2017 a major share holder in Intel publicly denounced the engineering department, and stated his massive disappointment in Intels lack of progress. and now it's delayed yet again, to 2019. they're going to be severely behind very soon. AMD is on track with both Global Foundries 7nm and their own Zen 2 and Zen 3, it's being verified every 3 months or so, that it's on track. And VIA was stating a few months ago, that they're developing their own x86, and Intel can't do anything about that, they inherited Cyrix's x86 license.

He is right. There's no standard. Maybe because no European company is engaged in wafer production, so there's no legacy German Industry Standard (DIM) or a Austrian ISO standard (International Standardization Organization)

I know i sound euro-elitist, but there's no standard.

GloFo's 7nm is likely similar to Intel's 10nm++ (But Intel is not of the hook, neither 10nm or 10nm+ turned out to be functional, and 10nm++ is delayed yet again)

this is like 2017 march news why is it making the news again?

14nm+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

Why don't they just use the same damn size transistors and just make bigger CPUs? Why do are they cucking themselves and us to the
>muh smaller is better
crowd?

Because Moore's Law is not dead mumble mumble Intel rules mumble mumble

you can fit more transistors on the same die. More transistors = more processing power.
you can shrink the die size and still have the same processing power. Smaller die = more die per wafer = cheaper to manufacture
smaller transistors consume less power.
Smaller transistors have a shorter delay, so you can operate at higher clock frequencies.

>Why don't they just lower the yield per wafer to Fermi percentages
I dunno.

because smaller is better (in certain aspects).

Isn't the diameter of an atom around 0.1 nm?

Imagine if Intel spent $22 Billion on their 10nm and 7nm process instead on wasting it on McAfee and Mobileye?

Why do you post this stupid shit in every thread?
>just make CPUs bigger hurr

Die already.

This is about the first time I posted it, so you must be thinking of someone else.
If the point though is to get more transistors why can't they just make a bigger CPU instead of the transistors smaller?

Because the yields would tank (even more). The point is: smaller transistor = smaller dies = more dies per wafer = more shekels = goyim happy

>Everything is fine.
Everything was fine while amd had fucking bulldozer. The only thing brian was good at is abusing intel monopoly, at first signs of competition he has no idea what to do except for throwing money on everything in hopes it'll workout somehow.

This and hiring Jim Keller :^)

Keller is doing fucking SoC integration, the cores are still either Hillsboro Downs victims or da joos with unrelenting itch to strap more shit to P6.

The P6 cow is going to dry soon.

Attached: intelcoresb7jai.png (884x408, 376K)

See the image for why small dies are better than big dies in terms of yield. Its why AMD can make Epyc cheaper than intel's high end server dies despite having more total silicon area.

Attached: Wafer_die's_yield_model_(10-20-40mm)_-_Version_2_-_EN.png (3000x1000, 163K)

Why can't they just make a square or rectangular version of those? Also what does the ones coloured yellow represent?

wont AMD have 7nm in 2019?
jesus... intel is really fucked

Attached: 1375567284935.gif (208x156, 1.23M)

Yes. GloFo will have risk production 7nm DUV in the second half of the year.
AMD already has some 7nm Vega GPUs from TSMC in house for testing.

Said like someone who can't even make a half-adder out of logic gates.
What are you going to do with all of this extra die space you're requisitioning?
Fucking nothing, that's what. The dies are the exact size they need to be to accomplish the sets of instructions they implement.
The primary limiting factor in CPU performance is the distance between transistors, pleb.

Circular wafers are due to how they make the wafers in the first place.

Yellow squares in the image are dies rejected due to defects (little red dots).

You mean the wafers? Because they slice them from huge silicon crystals (ingots), and those crystals are made by dipping a seed crystal into molten silicon and rotating it. So the wafers are naturally circular and they'd actually waste material and effort by cutting them into a square shape.
Pic related.

Attached: growing-silicon-ingot.jpg (641x400, 238K)

>consumers buy nanometers

Attached: brainlet.png (1063x1063, 200K)

>Most other fabs bullshit on size
ALL fabs bullshit on size, 10nm is just a marketing term. It doesn't refer to anything concrete whether you're talking about Intel or one of the other fabs.

Why does it need a seed?

Because that starts the alignment of the larger crystal as it grows.

So the silicone stuff gets misaligned if you just let it grow around a piece of metal or something?

Yeah, but their competitors will produce wafers on their 7nm node in 2019 which beats intel's 10nm.

Why hasn't the industry moved into 450mm wafers yet? I remember reading about it in 2010 and they said it's imminent but we're still on 300mm.

Silicon, and it would otherwise grow into a structure that wasn't uniform. They start it with the seed and it becomes perfectly uniform internally, one consistent monocrystal.

Everyone put it off because of the cost. Most of the industry delayed, until what was supposed to be 2018 at the time, but no one is committed to it fully yet.

What do they do with the dies on the edge of the circle that aren't complete squares? Just throw them away?

>Maybe because no European company is engaged in wafer production, so there's no legacy German Industry Standard (DIM) or a Austrian ISO standard (International Standardization Organization)
*blocks ur path*

Attached: Euro fabs.png (1410x1153, 93K)

They don't print anything that overlaps the edges. An incomplete die is just waste.

oh wow that sucks.

When molten metals begin to solidify, they form sets of crystals. Apparently this is an unwanted phenomenon in some fields of electronics--one property i'm familiar with is that a piece of metal composed of a single crystal has lower resistance than a similar one made up of multiple crystals.

Attached: blocksurpath2.png (1461x669, 60K)

Attached: blocksurpath3.png (1461x468, 35K)

Attached: blocksurpath4.png (1452x664, 58K)

We're talking bleeding edge memory/logic you braindamaged nigger.

Still leaves intels ireland fabs and global foundaries german fabs. btfo'd

Intel is a European company now?

Fucking stupid nigger.

Not to mention the most important company for bleeding edge lithography equipment is euro

Attached: ASML_Holding_N.V._logo.svg.png (1065x300, 26K)

They make tooling, they don't define nodes or final design rules.
Jow Forums, please, stop being retarded for god's sake.

>And VIA was stating a few months ago, that they're developing their own x86, and Intel can't do anything about that, they inherited Cyrix's x86 license.
this must be bait

There ireland branch is a separate holding company

en.wikipedia.org/wiki/Intel_Ireland

dare I say it btfo'd

too bad intels 10nm smaller than everyone else's 10nm doesn't work

the top picture does show that though

asml.com/products/systems/twinscan-nxe/twinscan-nxe3400b/en/s46772?dfp_product_id=10850

>The TWINSCAN NXE:3400B will support EUV volume production at the 7 and 5 nm nodes

asml.com/products/systems/twinscan-xt/en/s46772?dfp_product_id=10272

>The TWINSCAN XT:1460K 193-nm Step-and-Scan system is a high-productivity, dual-stage ArF lithography tool designed for volume 300-mm wafer production at 65-nm resolution

woah

yes because EUV technology is a piece of cake that basically everyone can do.

This is the result of forced diversity.

I bet they hired a lot of women and niggers on their tech department.

the only ones that once fabricated in europe/germany were amd that there is just offices for e.g. advertising

No it doesn't. It shows a grid, but the dies are either dark grey (good) or yellow (bad), while the squares on the edges are just light grey (meaning unused).

>The Intel Ireland campus at Leixlip is Intel's largest manufacturing plant outside of the United States and consisted of two semi-conductor wafer fabrication facilities: Fab 10 Ireland Fab Operations (IFO), and also the Fab 24 manufacturing plant which includes Fab 24-2. IFO was a 200mm wafer facility whilst Fab 24 processes 300mm wafers using 65 nanometer and 90 nanometer process technologies. IFO ceased operations and fab 10 is now classed as Fab 24, along with Fab 14. The Leixlip facility now only produces 300mm wafers using 14 nm technology.

Why is Intel fabricating in Ireland where they can evade so many taxes?

Attached: 1509213345397.png (1054x526, 209K)

When you scale your designs down, you get a whole slew of benefits like power.

Attached: Capture.png (695x492, 209K)

He's crashing their fabs with no survivors

Why was haswell bigger than ivy bridge?

Because it added transistors.

the OG fermi firehouse
a legend of thermal performance

>We're talking bleeding edge memory/logic

Doesn't NXP make bleeding edge stuff?
Or is it just innovation at larger nodes?

To be fair: all these delays are 99.9% ASML's fault.
They need to get their shit together, the entire world is waiting on them.

Why would they have to fabricate in Ireland to evade taxes?
Foxconn-Apple fabricates in China but evades taxes in Ireland.

>b-b-b-but muh applel!!1

Attached: 1520794112670.jpg (633x758, 62K)

>same thing over and over again with slight differences
lol

TSMC and Samsung also use ASML fabs and they're doing fine.
It's like saying a hammer company is faulty because 1 out of 3 of the customers doesn't know how to use a hammer.

>the entire world is waiting on them.
Maybe it wasn't such a good idea to have only 1 fucking company in the entire world that can make tools for high end manufacturing.

Don't know the size of them but transistors won't get smaller than around 3nm because of the size of atoms

The larger the cpu is the more energy is needed, and the more energy you pump in the more heat is created. CPU's are unable to go past a certain size without necessarily being a certain temperature.

*because the rate that heat goes up vs size is faster than linear

The known velocity and known position of an electron are conjugate pairs (it is impossible to know both completely at the same time). With that in mind, we can model as electrons’ position spatially with a probability distribution. As I recall This distribution is fairly well localized within 3mm.

*nm
You don’t want electrons hopping across the channel at -3 sigma.

No, it's still largely white males working with fabrication and design. The problem is that their workforce in those crucial departments LACK diversity, since they refuse to bring in nonwhite nonmale workers for any position that isn't in the public spotlight. By removing the dinosaur white male employees in the engineering department and ending the classist exclusivity of their careers, Intel would remain top dog for decades to come.

GloFo and TSMC are ahead because less their half their engineering workforce are white males.

Well it's not exactly that easy to do what ASML does. Producing machinery to fabricate chips at 7nm now and even 5nm as a test now really is an achievement that would be extremely difficult and costly to any new competitors if they arise at all.

They're quite good at what they do. But whatever you may think, lookin at whay our politicians are doing to my country, ASML is one of the lasts things in my country I feel proud about so they definitely counts for something to me.

>BTFO by ryzen

Uh when did that happen

>By removing the dinosaur white male employees in the engineering department
They'd lose all their accumulated knowledge, culture and style. This is what happened to IBM in the early '90s. With all their experienced staff gone they started making the sorts of mistakes the oldies knew to avoid.
The new broom sweeps clean but the old broom knows where the dirt is.

What rock have you been living under?

>With all their experienced staff gone they started making the sorts of mistakes the oldies knew to avoid.

sounds like what the Aerospace company i work for is doing.
the best employees are often the ones who will put up with the least amount of bullshit.
i'm currently watching an entire company crumble as bad decisions compile on another.
stupid fucks didn't listen to be and it cost them 30 million in revenue.

They intend to do 3D die stacking at some point.

8700K 5ghz 1.136v Z170

Attached: 13948028943095.png (2560x1440, 633K)

>implying

Attached: 1513914904516.jpg (685x599, 69K)

>in case you're serious
when 1st gen Ryzen made a 1K+ (((HEDT))) Intel CPU obsolete, and then again with Threadripper vs. i9 (which cost double the price of TR but doesn't perform twice as fast)

that's not solving the issue of errors that just consolidates errors to multiple layers. layering is going to a massive game changer when we get it, but its going to demand massive redundancies and low clocks. it likely won't be feasible till a process is 100% perfected, or we go backwards to sub nm parts.

just wait that 300$million on feminism is going to pay off any day now, you will see.

intel drank the kool aid and believes their marketing, the culture at intel needs to radically change but by then it will be to late.

keller may be able to deliver a cpu, but with no process advantage after already making a top end cpu for amd, they will never regain the top position.

haswell was the last gen intel tried to improve and did not rely on process alone.

>risk production
>1Q-2017

Zeno's chip fab.

why not answer the question honestly?