Could Intel’s 10nm Node Face Another Delay?

>>59291211
Time to ADD MOAR BIBELINES :DDDDDDD

Intel is like the tortoise in the tortoise and the hare fable. Everyone is always like "Intel may be behind". But Intel always takes its time to optimize, it never lies with node title like almost everyone else, and when it says they have 10nm it actually does have 10nm not 14+.

File: 1483737017981.png (973KB, 801x1500px) Image search: [Google]

973KB, 801x1500px

>>59291238
Ebin :------D

>1.7% yields
wew

File: 1352197404494.png (89KB, 399x382px) Image search: [Google]

89KB, 399x382px

>>59291571
>Ryzen manufactured on equivalent of Intel 22nm
>still smaller core-for-core than Intel's 14nm processors
>beats or keeps up with Intel's 14nm processors on most multicore tasks

Really gets the synapses firing.

>>59291765
AMD doesn't have foundries at all. They are so shady we don't even know what they use, GlobalFoundries or TSMC (if you don't believe me, google it).

I believe the main reason they run behind Intel like a dog is exactly that, they don't have their own Foundries, so they have to pay fees to others so they have to be behind and/or more expensive for the same performance.

Notice the AMD downfall came exactly at the time they lost the control of their own Foundries.

>>59291831
>AMD spins off GloFo
>"gee we dunno what they're gonna use"

http://hothardware.com/news/amd-shipping-zen-in-limited-volumes-q4
>Zen is being built on an advanced GlobalFoundries-sourced 14nm FinFET process

They also have a sourcing agreement with Samsung for Zen, who has the same 14nm process as GloFo.

>>59291211
There was already word that they've been having persistent trouble. Cannonlake desktop may not happen. It could turn into another Broadwell, but cancelled much sooner this time around. It may be the case that the first mainstream 10nm desktop part from intel is a later Ice Lake. That could mean no 10nm desktop parts until late 2018.
There have been some patents which depicted a modular Cannonlake Xeon. They were planning on fabbing logic and uncore separately and stacking them. Trying to keep costs down by no longer fabbing big monolithic dies. Though now apparently intel has pulled all Cannonlake Xeons from their roadmap.
Things definitely aren't looking good for them.

>>59291765
>implying 14nm LPP is only comparable to 22nm
Intel's 22nm process was far behind industry 20nm nodes in feature size.

>>59291831
This is one of the most ill informed retarded posts I've ever seen here.
We know exactly what process AMD uses on all of their parts.

>>59291889
Do you even know how to google?

https://www.overclock3d.net/news/cpu_mainboard/amd_is_rumoured_to_be_fabricating_ryzen_cpus_with_both_globalfoundries_and_samsung/1

https://vrworld.com/2016/01/19/amd-to-offer-finfet-chips-with-both-tsmc-and-globalfoundries/

http://wccftech.com/amd-contracts-tsmc-produce-zen-16nm-woes-14nm-process-troubles-globalfoundries/

>>59291946
>rumors from 2015

Wow, good job, little kid. You proved that you're the only clueless one here.

>>59291932
>We know exactly what process AMD uses on all of their parts.
How? Did you look at their tiny little stamps with a microscope? You are full of shit.

>>59291571
Jeb pls go.

>>59291946
>rumors
>as opposed to AMD literally saying that they're using GloFo with Samsung on backup in case of supply issues

Why the fuck would they use TSMC anyways? Samsung/GloFo literally share a 14nm process, you'd have to do a lot more work to transition to TSMC's 14nm process.

>>59291932
>Intel's 22nm process was far behind industry 20nm nodes in feature size.

I don't doubt that. I was just citing the pic in OP that claimed that Samsung/Glofo 14nm was equal to Intel's 22nm.

>>59292000
>AMD literally saying that they're using GloFo
And you believed them. Good pet.

>>59291982
The die and package are marked where they're produced and assembled. The Zeppelin die is a 14nm LPP part fabbed at Global Foundries Fab8 in New York. It is a product of the USA. Packaging is handled in a facility in either Malaysia or China.
Stay in /v/, little kid.

>>59292016
Not him but you are need of medical attention. Why do you call everyone a little kid here?

>>59292033
He likes talking to little kids.

>>59292049
I see. Aha, aha. That explains everything.

>>59292013
Wow, let's totally believe some faggot anon on /g/ who claims that AMD decided to qualify Zen on two completely different processes, instead of dual-sourcing from Samsung and GloFo and just using the same PDK for both.

>>59292062
I think its some meta intel fanboy trolling
>intel's 10nm node is shitting the bed harder than their 14nm did three years ago?
>W-w-well you don't even really know what process Ryzen is made on!

>>59292127
Wait. I wanna troll as much as the next guy but that's pure AMD fanboyism. Intel is obvious having the upper hand in manufacturing.

I'm not saying they're having the upper hand on the process necessarily, but they sure have it on chip making and chip pricing.

Unless you want to convince us Ryzen destroys everything Intel has up its sleeve.

>>59292127
I wonder how butthurt Intel fanboys will be when GloFo delivers Zen+ on 7nm SOI node.

>>59292328
And he is the AMD fanboyism again.

>>59292347
What the fuck is that English, Pajeet?

>>59292358
And racist too. I was saying you pretend to know so much and then casually drop some AMD fanboyism out of nowhere. Whatever process node issues Intel is having, you can't seriously they don't have a good hand at manufacturing over AMD consistently and for years.

File: Screenshot_20170307-184924.png (372KB, 1440x2560px) Image search: [Google]

372KB, 1440x2560px

>>59291831
>>59291903
really interesting how amd runs behind intel like a dog when though intel hired you to shill for them. or, seeing as how you put so little effort into hiding it, you might just be doing it for free?

>>59292328
They'll be shitposting nonstop, especially since intel's 7nm process won't reach market until sometime around 2021, possibly later with delays. The facility in Arizona hasn't even finished construction yet, they have years left to go before the building is even fully ready to handle wafers.

>>59292347
LOL look at the broken English street shitter.

>>59292437
Even more terrible grammar. Jesus you can't even hide it.

>>59292325

I don't disagree with that. But that Ryzen was manufactured on an inferior process, at least density-wise, and still comes out smaller than Intel's cores, suggests that there's plenty of room for improvement once the foundries catch up with Intel.

But
>chip pricing
Nigga, you best be joking.

>>59292476
Dude, quit repying to the troll.

>>59291571
wasn't intel's 22nm closer to 26nm?

>>59292464
Hes just doing the needful for his rupees. Very much good on post to intel in favorable manner, yes.

>>59292464
>>59292476
>>59292532
> le I'm a racist le meme
keep sperging racist child
some of us are not American or British
GASP!
what a surprise!
go back to your mommy.

File: FeatureSizes1[1].png (18KB, 872x597px) Image search: [Google]

18KB, 872x597px

>>59291571
>it never lies with node title like almost everyone else
t. nigger

>>59292483
>once the foundries catch up with Intel.
wishful thinking. people always say that. Intel seem to always play a smart hand.
they don't need to rush a 2nm process overnight. they have most people covered.
start by thinking that most of /g/ has still chips as far back as sandy bridge that people don't yet seem to think they must replace

>>59292483
>>chip pricing
if AMD can't compete for 90% of the time for the last 10 years, yes, pricing
if this small window we might be seeing now, it might be closing fast
besides, many people see the new priced i7-7700K an ideal chip

>>59292727
you could at least try to not expose you have cognitive dissonance by not calling people niggers.
I bet you would never say that word in front of your mommy or of course a black person

>>59292820
t. /g/ retard nigger

>>59292772
>>59292820
>please to be stopping of the racist, sir
>I am to be askings with politness
>you vould not saying such things to mommy
>please sir
>making sure to install Microsoft Vindows10 with most latest intel processor of the i7 7700k for very much good perform

>>59292772
>Intel seem to always play a smart hand.
The fuck? Retards tried to compete aginst ARM with their shitty arch.

>>59292934
I don't give a shit what you think racist kid. I'm just pointing out it undermines your argument that you must resort to racist slurs. It exposes you have cognitive dissonance on the manner and therefore you don't even believe what you're saying yourself.

>>59292948
Well, on the desktop. /g/ seems to like to shill AMD a lot.

>>59292986
The fuck? Intel created abomination named NetBurst.

>>59292995
yes yes, they are so useless, that's why even in /g/ a board full of AMD shills 99% has Intel cpus.

>>59292772

>intel has something up its sleeve
>like it did for itanic and netburst

The only thing Intel did was to shut AMD out of the OEM market and wait for AMD to do something retarded like cancel an entire microarch iteration.

And you really think Intel has a fucking 2nm process in the works, when they're struggling to get to 7nm and haven't even worked out the yields for their 10nm process?

>>59293051
What the fuck is that English, Rajeesh?

>>59293057
2nm was a figure of speech, it might never even happen.

the point is Intel has the upport hand, they don't need to be fast at making it into new nodes.

>>59293051
>"intel is of genious and is never wrong"
>remember presshot and lolAtom?
>"WELL EVERYONE USES INTEL CPUS"

What kind of fucking argument is that? He's just saying that Intel isn't the god that you think it is. They've had plenty of fuckups before.

>>59293074
fuck off to /pol/ racist kid.

>>59293091
Oh, they do. Their arch is old as shit, the only way for them to stay competetive is node shrinks and they are already behind Samsung in that department. Shill harder poojet.

There is an actual paid shill ITT posting from India defending intel. This has to be one of the funniest things I've read in months.
What a time to be alive.

>>59293103
fuck off with the textbook strawmans kiddo, nobody told you Intel are saints, the entire subdiscussion was about Intel having a market dominance.

so let's hear your arguments that they don't have a market dominance.

in a board full of AMD shills, even here Intel dominates.

>>59293118
> I'm a racist
> I don't have an explanation why AMD can't beat such an incompetent company like Intel

are you literally in need of psychological attention at this stage or what.

>>59293091
They have _maybe_ a half-node advantage over the foundries on 14nm.
See >>59292727

In any case, GloFo has IBM's 7nm process, they just need to wait for EUV lithography like everyone else.

>>59293173
(you), Ramesh.

>>59293184
I don't know what you're trying to accomplish at this point. To tell me that 4chan has a /pol/? If you hate the black kids down your street go tell them and fuck off.

>>59293203
Зaшивaйcя, пopвaткa.

>>59293179
>wait for EUV
I don't know exactly what you're wishing there, I don't know details on manufacturing, but something tells me whatever you are wishing there doesn't involve Intel sitting idling by doing nothing in the meantime.

>>59293142
You're the faggot citing marketshare in a discussion about semiconductor processes and processor architecture.

File: file.png (80KB, 349x642px) Image search: [Google]

80KB, 349x642px

>>59293251

>>59293236
Intel won't have their 7nm node online until 2021~, the facility which will run the process isn't even fully built yet. Until that time they have only their apparently problematic 10nm bulk FinFET node to rely on.
Both TSMC and Samsung are aiming for volume production of 7nm bulk parts in 2019 utilizing EUV light sources.
In this same time frame Global Foundries will be bringing IBM's 7nm SOI FinFET process online. The ramp up starts later this year as a matter of fact. They will have risk production in later 2018, and will have products in channel in 2019. Reduced complexity from fewer critical mask layers, dramatically higher channel insulation, proportionately larger Weff, lower gate delay, higher performance SRAM cells. It is a true high performance node at low power. Intel has nothing to address that in any sort of relevant time frame.

>>59293236
Well EUV is pretty critical to making 7nm viable. But IBM already has a process for 7nm FinFET.

Of course, this doesn't preclude Intel from working on its own 7nm process.

>>59293283
>anon: The fuck? Intel created abomination named NetBurst.
>you: yes yes, they are so useless, that's why even in /g/ a board full of AMD shills 99% has Intel cpus.

You should consider suicide, but I'll let drug-resistant tuberculosis do the job for me, Rajesh.

File: ticktock.png (71KB, 470x1133px) Image search: [Google]

71KB, 470x1133px

>>59291571
Yeah bro, they're busy "optimizing". I'm sure there's no risk at all that they're losing their competitive advantage.

>>59292704
lmao what racism? i never said anything about race

>>59293400
I guess they mainly address it by being alone on the desktop, together with AMD hogging the x86 patents, and AMD without its own foundries to compete on equal grounds without paying fees to Global Foundries.
I guess Samsung would destroy them if they had full access to the x86 IP.

>>59293663
Thank AMD and their amazing contributions in competing with them. But oh wait, they don't even have to compete. They are a cartel with Intel together hogging the x86 IP/patents.

>>59291211
>10nm
>2020

AHAHAHAHAHAHAHAHAHAHAHAH

PREPARE FOR INFINITE SKYLAKE REFRESHES

>>59293794
Holy fucking shit. Learn to speak English or do not post here.

>>59293945
There is literally no grammatical error on that sentence you illiterate child.

>>59293919
Do you retards even understand how small 10nm is at this point? Fucking microscopes can't see it. You act like Intel are a bunch of bumbling idiots.

>>59294048
I see you fixed your typo, but you failed to make your post in any way relevant to this thread.

>>59293794
>together with AMD hogging the x86 patents
>together hogging the x86 IP/patents

>What is via

Also the majority of the basic x86-64, sse, and sse2 patents should be expiring soon

So a company will be able to build a cpu without sse3, sse4, axv, axv2, axv512, fma3, bmi, f16, aes-ni, rdrand, or clmul

>>59294048
>You act like Intel are a bunch of bubbling idiots

Because they are, they have the money and they're still incompetent retards.

>>59294075
you fail in life to be honest.

>>59294076
>a cpu without sse3, sse4, axv, axv2, axv512, fma3, bmi, f16, aes-ni, rdrand, or clmul
Don't they hold them by the balls via other patents like cpu placement on motherboards, and the like?

>>59294007
This is just a laughably childish fallacy.
Intel is falling behind the rest of the semiconductor industry. You can't write that off by saying "b-but its still pretty small!" They're having problems with their 10nm process, a process which is still a bulk FinFET one. They won't have something to replace it for about 4 years baring any major delays, which given their recent track record would be likely.
The competition is already moving away away from bulk to advanced SOI. A move that actually brings ground breaking new technologies to market.

You don't even have a point to make.

>>59294027
It is rambling gibberish, street shitter.
Intel's foundry is not free. The largest chunk of their revenue goes to covering operating costs along with R&D expenditures. You insinuating that intel holds some intrinsic advantage because AMD is a customer of Global Foundries makes no sense. It is something only a clueless street shitting Indian would claim. Intel's X86 business is a customer of their foundry business. Intel writes purchase orders to themselves, and they have to pay themselves. Having the same company name is nothing more than legal semantics. Operating under one umbrella does not stop them from being different entities.

>>59294149
>It is rambling gibberish, street shitter.
You can die in a fire racist child. There was no grammatical error there. How do you feel that I can speak better English than you when I never speak English in real life?

You are pathetic.

Go find a black child in dah hood to tell him how you think he is a nigger if you're so brave, coward.

>>59294146
Can you show me this patent?

>>59294202
Calm down there, Deepak. If you get too riled up you might cause an elephant stampede that destroys your village.

>>59291238
MOAR GIGAHURTZ

File: file.png (217KB, 480x480px) Image search: [Google]

217KB, 480x480px

>>59294291
whatever coward, keep us posted for when you call that black guy a nigger in front of his face.

>>59294232
It was a question. I guess you say no. I was wondering what would be the point if the x86_64 part is inaccessible.