IT'S FUCKING OVER

>en.wikipedia.org/wiki/List_of_Intel_CPU_microarchitectures
(2019) Cascade Lake ? MHz
(2019) Cooper Lake
(2020) Ice Lake 10 nm
(2020) Tiger Lake
(2021) Alder Lake
(2022) Meteor Lake 7 nm

Attached: 1451005843763.jpg (250x251, 31K)

Other urls found in this thread:

semiengineering.com/big-trouble-at-3nm/
hardocp.com/news/2018/12/07/tsmc_building_new_8inch_fab_capacity_updates_5nm_node_progress
twitter.com/SFWRedditImages

I believe

this will be...

Not to mention process slowing down, temps getting higher

why did they go from 10nm back to 14nm then back to 10nm?

>wikipedia
you amdrones are getting desperate

Attached: 11387.png (1200x1200, 670K)

Cannon Lake was a single Core i3 2C/4T with no iGPU
It's like the only thing to even come out of Intel's abortion of a 10nm process and it's not even profitable to make
I'm not even sure it exists to be honest

though they've been on 14mn for 5 years, whats taking them so long

Why should they hurry? They have no competition.

>7nm
>2022
BRB shorting intel stocks

Attached: intel.png (1610x734, 275K)

It's true 7nm, not AMD's "7nm" (actual 12nm)

cope

>check out the source
>some cpu wiki
>Preliminary Data! Information presented in this article deal with future products, data, features, and specifications that have yet to be finalized, announced, or released. Information may be incomplete and can change by final release.
it's fucking nothing

Denial is the first step :)

this
wikipedia is a lie
7nm is a lie
ROME IS A FUCKING LIE
stop being mean to intel

Attached: 1515411243361.png (552x661, 288K)

It's more like not even Wikipedia can be bothered to keep Intel's constantly shifting delays in order anymore. Might as well just put "Not any fucking time soon".

Why are there so many fucking lakes and canyons yet no actual progress?

So Ocean Cove isn't coming untill 5 years from now, huh. As a waitfag is it worth to wait a bit more for DDR 5 and PCIE 5? I lost hope for Intel releasing anything meaningful. If the rumours about Navi are true at least we'll have that to look forward to.

They keep drowning and falling in them, respectively.
The only thing worth waiting for at the moment is Zen 2.

>(2020) Ice Lake 10 nm

Attached: 1473646069630.png (576x507, 59K)

What are the chances of ddr5 coming out in 2020?

I would count on 2021 for reasonable DDR5 prices and ZEN 3. Wouldn't want to be one of those early adopter losers.

All but confirmed, Zen 3 is highly implied to be ddr5 which will come out in 2020

What about 5nm and 3nm?

Intel 14nm+++++++++++ superforce by 2020

semiengineering.com/big-trouble-at-3nm/

3nm is rather inneficient and probably won't be available to the masses

>AMD's 7nm is fake
what does this mean? I'm asking at least third time about this and yet no one gave me explanation or any source on this.
Can someone elaborate why AMD's 7nm is not 7nm and why AMD calls it 7nm?

TSMC 7nm is approximately the same as Intel 10nm. TSMC 7nm EUV is approximately the same as Intel 7nm. It's all guesses and extrapolations though.

>7nm is approximately same as 10nm
what does that mean?

He's being generous. In reality Intel's 10nm is more advanced than TSMC's 3nm (which is really a rebrand of their 14nm)

Intel's 10nm is equivalent of TSMC's 3nm

It's so revolutionary that's why Intel is having trouble with it
Scaling back to 12nm (TSMC's ~5nm) is a smart move to refine the technology while still maintaining a huge lead over the competition

I pity the fools who bought into the AMD marketing machine without doing proper research
No wonder AMD staff is getting disgusted and are leaving the sinking ship en masse

>more advanced
how?

yeah sure... but they had to ditch it and are developing 10nm-- which is probably only as good as Samsung/TSMC 10nm

Attached: process node 2.jpg (1322x301, 68K)

>7nm in 2022

AMD will be on 3nm by then

this here, basically

Attached: 1534300275168.png (1400x1274, 206K)

>1000nm
did they forget that we used μm before nm?

They measure differently.

He doesn't know, he's just parroting things he's read online. My guess is they're measuring the size of the transistors differently.

THIS GOY GETS IT

Attached: 1487297768743.jpg (960x878, 125K)

On the desktop? Middling. It could arrive towards the back end of 2020 if a suitable platform is launching at that time, but early 2021 is more likely, with 2022 being the year that it's predicted to gain widespread adoption and overtake DDR4 in terms of sales.

how differently? they use something else instead of meters? or does nano has different definition?
are there AMD's nm defined in unit system and intel's nm having different meanings?
doesn't nm means channel length of transistor's fin? what do AMD measure with 7nm?

:^)

Attached: IMG_0573.jpg (355x331, 65K)

The truth is that Intel's 10nm used to be better than TSMC's 7nm, but they've had so many problems trying to scale conventional FinFETs down and get good yields that they've backed off to the point where it's more equivalent to TSMC's or GloFlo's 12nm.

Basically, Intel is completely fucked.

Attached: giphy_1.gif (350x396, 541K)

yes, exactly, so what does AMD measure and what does intel measure with nm's? isn't this the same thing?

As long as TSMC gets cash injections from fucking Apple Nvidia and AMD Intel won't ever be able to compete. Their fabs are done for.

its a combination of multiple different things, 'nm' is mostly just used as marketing so there is no logic to it

What's the point of shrinking it if it's still a fucking Swiss cheese architecture with Jewish cum filled IHS? Not buying Intel until they get rid of Core.

so you're saying 22/14/10/7 nm's doesn't matter? it shows nothing?

Im saying you cant compare 1 fab to the other, you can compare 22nm intel to 14nm intel and say 14 is better, but comparing 14nm Intel to TSMC 12nm and saying it's worse doesnt have to be the case.

>Not buying Intel until they get rid of Core
Wait for new uarch in uhh 2023 maybe

shut up goy

2023? You do realize they haven't had a successful uArch since 1995, right? Intel tried to make IA64, which ended up as a market failure because optimizing compilers/JITs were in their infancy, so programmers couldn't keep the icache full. NetBurst had a head-on collision with the laws of physics. So what did Intel do in the face of two monumental flops? They pivoted back to the tried & true P6, took SMT from NetBurst (arguably its only good feature, though recent exploits have shown that to be a lie.) Then, using all their newly gained transistors, they implemented AMD64 in uCode, copied AMD by bringing the memory controller on-die, started scaling out to more cores, and later they'd copy-paste the vector units from Larabee. (Which was also a commerical failure in its own domain.) Core is nothing more than iterations of P6 which has cannibalized IP from other product lines.

The ultimately irony is that IA64 was a much better architecture, and it would probably succeed now that we've spent hundreds of person years on optimizing compilers which exploit UB in an attempt to piss off C programers. Not to mention Lars Bak has created not one but *two* world-class JITs in the interim, all in a futile attempt to make Java (HotSpot) and JavaScript (V8) perform reasonably well; despite the mountains of crap that enterprise / web developers throw at them.

Intel hasn't seriously innovated in decades, the people on Jow Forums shitposting from their i7-960 and 2600k CPUs are the only ones who saw through the ruse.

You've by lied to by a shill this entire thread. GG.
AMD doesn't have a foundry, they're using TSMC. By objective measures like connected poly pitch, minimum metal pitch, SRAM cell size, total logic cell area, TSMC's 7nm non EUV process is higher density(smaller feature size) than intel's 10nm process they originally publicly detailed.

Process names are not just marketed PR. The number one producer of lithography tooling in the world, ASML, establishes loose guidelines on node definitions. They do this in line with what their tooling will be capable of. Its almost a self fulfilling prophecy in a way.
When you hear 14nm/10nm/7nm/etc what is being referred to is not any one specific feature of a chip. It is denoting that this chip was made on a node with tooling of certain characteristics. M2 half pitch is the metric that node definitions rely on for consistency.
A 14nm transitional FinFET process(20nm BEOL) has an M2 half pitch of 32nm. A 7nm FinFET process with corrected scaling has an M2 half pitch between 11 and 15nm.

Processes are not named for physical gate length, this hasn't worked since the late 90s. They are not named for any characteristic of their Fins either, as one retard ITT tried to imply. M2 half pitch and the tooling used are what defines a process node. Nothing else.
TSMC's 7nm node isn't "fake." Its higher density than intel's 10nm, and TSMC's 7nm EUV is pushing density even higher.

Don't buy into retard shilling.

ni

>5nm node to risk production in the second quarter of 2019, followed by volume production a year later

hardocp.com/news/2018/12/07/tsmc_building_new_8inch_fab_capacity_updates_5nm_node_progress

INTLEL GARBAGE FABS BTFOREVER

Attached: DAC2018 screenshot.jpg (620x350, 30K)

Before you choke on your foreskins, Schlomo, I'd like some source on that.

Attached: npc_rabbi_jew.png (1013x1189, 524K)

>TSMC's 7nm node isn't "fake." Its higher density than intel's 10nm
I'd just like to interject for a moment, what you are referring to as "Intel's 10nm", in fact, actually doesn't exist or as I've recently taken to calling it, Another Shoah.

they had released some 10nm CPUs for laptops, didn't they?
yeah, it was shit, but it was 10nm

That's what they'd like you to believe.

Yes, there are in fact some 10nm chips in the wild now. Its one die however. A dual core i3, low clocks, IGP is present but binned off. They're selling them in some NUCs now that have a discrete AMD GPU ironically.

Intel will have relaxed pitches for their parts that actually go into production.

Why do they like lakes so much?

2019 - bloodlake

>You do realize they haven't had a successful uArch since 1995
YIKES

You are not allowed to trademark geographic names. Using them for code names guarantees you will not accidentally infringe on someone else's trademark and protects you from another company trademarking your code name. Also geography provides a nearly endless supply of code names. They used to use river names (e.g. Willamette, Tualatin), now it's lakes.