When Intel/Amd is going to advance <=10nm processors?

When Intel/Amd is going to advance

Other urls found in this thread:

semiwiki.com/forum/content/1342-industry-standard-finfet-versus-intel-tri-gate.html
twitter.com/SFWRedditVideos

Probably around the same time that GlobalFoundries and TSMC does, seeing how Intel fucked around with everything but processors for a while.

AMD is tied to GF for the most part, but still uses TSMC, so they're dependent on those fabs.

Samsung will do what Samsung wants to do, which might be another iteration of FinFet.

tl:dr : Tech moves at the speed of fabs.

Yeah but when 10nm?

amd is gonna skip 10nm and move straight to 7nm in 2019

Zen Two Is On Schedule For 2018
AMD Will Be First For 10-7-And-Most Possible Five Nanometer Intel Have Absolute Nothing For Until 2020

unfunny

10nm laptop Cannonlake was scheduled for "by the end of 2017", but by all indications Intel fucked up hard with their 10nm process. Desktop and Server processors both cancelled until Icelake, due in late 2018.

Zen 2 will also be out in late 2018 on GloFlo's "7nm" process, which uses a 14nm backend so the die scaling won't be so good, transistors will be smaller but i haven't seen any pitch measurements yet. Too early for much to have leaked on it.

Who knows
Now stop whining you fag

...

Seems like Intel is fucked.

>Intel is not Qui-Gon Jinn
you fucked up, op

Not my picture it was only one I found with amd or intel word in my computer.

Makes sense character-wise but they matched up the colors of the logos with the lightsabers.

See Sup Forumsentlemen this is the problem with AMD's future.

I love AMD and I'm AMD fanboy myself BUT the future looks very uncertain to everyone else except for Intel.

Why?

When Global Foundation, TSMC and Samsung (these three are the ones who are going to make the AMD processors for AMD) decided to go with FinFET transistot technology while Intel went with Tri-Gate technology.

Tri-Gate technology is much MUCH more scalable and efficient compared to FinFET. And even tho Intel was the only one going this route this wasn't problem for them as Intel as a company is so wealthy compared to these other companies. Intel had the money to research and develope the Tri-Gate process alone and in the future it will pay off.

And there has been rumours that Intel has already investigated the even more superior transistor technology. Some circular/closed transistor technology.

I hate to be the bearer of bad news but it might be that Intel became a "Google" of processor world during last ten years when they were basically the only company succeeding.

ARM can't do anything either. Why? Well, even tho ARM looks very efficient on the paper it's actually very unefficient.

The ARM architecture consumes very little power but it's also very unefficient. x86_64 is about 10 times more efficient than ARM and on top of that Intel owns all the x86_64 instruction set pantens and has promised to sue anyone who even thinks to emulate those instructions on some other architecture.

The only real contender which can be a real threat to Intel in the future is RISC-V / OpenRISC.

Patience, goy.

>spouting meme buzzwords
>not knowing what's actually behind them
>Reddit spacing
Fuck off, will you?

Hint: "Tri-Gate" is a marketing buzzword for a transistor that is basically FinFET.

> Hint: "Tri-Gate" is a marketing buzzword for a transistor that is basically FinFET

Wrong. FinFET and Tri-Gate are completely different transistor technologies.

Tri-Gate is easier to scale down past 7nm with right materials. It's not even probably possible to get past 7nm with FinFET.

And of course this isn't a problem now for AMD and others but it will be in the future. While Intel just keeps scaling down others have to search for new transistor technologies and use money to develope and research what ever they pick.

Like it or not but Intel has done it's homework while others have been slacking off.

You can't jew the Jews I guess...

Even if the stupid shit you're spouting was true, Intel is stuck just trying to get to 10nm while IBM is already on 5nm.

>As I previously wrote, other than the unique profile chosen by Intel for their Tri-Gate implementation, there really is no key technical distinction between Tri-Gate and the industry standard term FinFET.
semiwiki.com/forum/content/1342-industry-standard-finfet-versus-intel-tri-gate.html

>14nm BEOL
Stop lying around.
It's denser than Intel's 10abortion due to SRAM alone.

>Vmin = 0.7
What did they mean by this?

Tri-Gate is Intel's meme buzzword name for FinFETs.
Neck yourself Brian.

That they are woefully late with their 10nm node to the market. It's fucking shit FEOL-wise compared to other nodes on the list.
Fuck it was supposed to come out a fucking year ago.

Picture goddamn related here.

Tri-Gate is a type of FinFET you asshat. FinFET refers to any number of 3D gates, while Tri-Gate is Intel's specific implementation for 3 gate fins.

Intel's implementation may be a bit better, but honestly if you thing FinFET/tri-gate is going to be the "killer" thing, I'm going to blow your goddamn mind.

The next big thing for 7nm and below is not Silicon, but rather EUV lithography or Gallium-based chips. The fundamental move off of Silicone is going to be the thing that makes or breaks fortunes. Maybe FinFET techniques will work there, maybe they won't, but transistor architecture pales compared to a completely new base.

And that 5nm transistor technology is called GAA (gate-all-around) which is the exact one that Intel has been researching for Tri-Gate successor (at least rumors say this). FinFET can't go under 7nm with current materials.

Don't be mad at me. Be mad at AMD for slacking with their home work.

Intel's 5nm is III-V based FinFETs you cockgobbling retard.

what are CPP and MMP?

This is the worst pajeet Intel hired to date.
Barely fucking different.
Let's not forget GloFo is going to insert EUV relatively soon.

are you retarded? downs?

the views towards amd's gpu's and upcoming cpu's were different back when I made the image. if I were to go back and edit, I would probs switch the colors and update the logo's.

Hey Brian why the fuck 10nm is so fucking late? You promised it ages ago.

>Not working for a tech company and knowing when 7nm will be announced

AMD is skipping 10nm altogether and going straight to 7nm

Not AMD, GloFo.

hush, they don't know that.

Really can local Intel Rajeets explain me why the fuck Intel's 10nm is so fucking late?
Also why the fuck it's so shit FEOL-wise?
According to Intel themselves first batch of 10nm shit will clock lower than their 14nm thingies.

Still wrong. Intel 5nm process is going to be GAA (gate-all-around) from silicon nanosheets. Just like IBM.

What part of " You can't go under 7nm with traditional FinFET technology" you don't understand?

I'm still waiting for 10nm or 7nm processor to upgrade my old 3570k.

question how abundant is gallium comapared to silicon?

silicon is the second most abundant element in the Earth's crust after oxygen, because silicon bonds to many oxygens per molecule of silicate

You do know AMD owns the 64bit extensions to x86 and intel licenses them from AMD.

You actually have no idea what you're talking about.

why is the AMD dude crying?
AMD was always the one who knew Nvidia and Intel are jews and tried it's best not to fall to their level

AMD should be a smug face

Going by the movie, AMD kills Nvidia, then Intel kills AMD?

man I love Personal computer tech we are moving so fast its like we're in the bullet train traveling at over 900KPH and trying to keep up with time whist trying to get your mind around it.

to think 5~7 years ago it was impossiable and now look at us we are breaking though the nanometer scale like it it was fucking nothing.

Blows my mind never have we moved so fast up the scale in my 30+ years of computing.

Mind blowing tech advances and we are not even at the top yet.
I feel like
SHEEEEEEEEEEEEEEIT slow down lah.

AMD is skipping 10nm and going right to 7nm.

Intel's 10nm node is up and running but is a complete mess right now, clocks are complete shit, and yields are 1.7%

I think that 10nm is going to be DOA unless yeilds get better.
other wise costs will go out the window just for a chip.
Unless intel plans on making it sever only node or what ever.

crying because he knows he's going to lose, like he always has.

By 2020 AMD will be onto 7nm LP+, provided courtesy of IBM and their black magic wizards.
All Intel can do by then, is "improve" their 10nm node enough to make it usable, they won't get a new node up and running in 2 years.
Not to mention it'll be 2020-21 before Intel get a new arch going, and then there's the question of if it will beat out Zen2.

Even if they dedicate their 10nm node to Servers only, it won't be enough due to their insistence on going for monolithic dies which turn out a perfect die less often than you see a unicorn shit. And due to that, their pricing is fucked up, look at their "Metal" range of Xeons, top end costs $13k, per processor, for 24 cores. Meanwhile an EPYC at 32 cores will run $4k and have 2 socket support.

Everything about Intel is fucked right now. They want to maintain their 60% margins, and are willing to sacrifice what is possibly upwards of 50% of their largest revenue source to AMD.

In a way, would make more sense if INTEL was studying the CPU glue and trying to get ahead of everyone and doing a megazord CPU made out of 6 glued dies with heatpipes etc..

The main reason behind Intel's slander of Infinity Fabric, is that it's actually fucking amazing, and they can't just say "Yeh, the company we neglected to pay attention to while we pumped all our money into useless projects and diversity hiring turned out to be kind of dangerous"

The main problem with AMD was always fucking incompetent management.
That's not the case anymore.
Dark times are ahead of Intel.

>Intel owns all the x86_64 instruction set pantens
You are literally a retard.
Go buy your DoA Itanium while you still can.

Minimal voltage for the gates to function.
Intel is 0.2 volts in disadvantage meaning higher voltages and thus higher temperatures.

2019.
Intel in 2020.

Intel is STILL struggling with 10nm m8

7nm will finish them.

>You must be 18 or older to post on this website

your feeble words can't wipe away a decade of ineptitude.

Loose what? That would imply there's something to win.
AMD is pretty happy with doing it's own thing and doing it well.

>decade

>He doesn't remember Phenom
>He thinks Piledriver has been AMD's only arch for 10 years

>You must be 18 or older to post on this website

It's sad when you lose so much that you stop thinking that winning is even possible.

Barcelona was meh and Phenom2 was pretty fucking late.

>t. I'm 16
It's sad that people actually don't know shit and think that way.