7nm is already here Jow Forums

b-but quantum tunneling

Are you retarded? There are already Intel 10nm CPUs on the market.

>TSMC 7nm:exists
>Intel 10nm: "exists"

Yeah, a one-off china-only batch of dual cores with their IGPs disabled that exists solely so intel can tell their shareholders that yes, they "have" a 10nm product.

That's one more product than AMD's 7nm, which has zero product

but if you consider the fact that 10nm was promised by Intel and yet its delayed by...2 years now with made-up-on-the-spot compromises that everyday further tarnishes Intel's image while AMD seems to show tangible proof that they are going to actually deliver something on the year they promised.

And AMD is on track to have 7nm chips out before Intel can release their 10nm chips in things other than shitty Chinese laptops.

AMD's 7nm is DOA already

>2 years
I see you've been reading intel marketing material where they re-wrote history. 10nm is 5 years late.

Um no sweaty they actually just unveiled it at the recent New Horizon event

Attached: rome.png (678x541, 525K)

Correction - 3 years late. Say what you want, but if things went smoothly as Intel had planned as far as the tick-tock method is concerned, Kaby Lake of 2016 was supposedly to be the 10nm tick, not a 14nm update.

>Hell, AMD was on the verge or failure
There you go again with your utterly stupid "but AMD" when we're talking about process nodes. Here's a clue: Intel and AMD are not competitors, they haven't been since GF was branched off to a separate company. Intel's competitors are TSMC, GF and Samsung. It's even what OP is about so it's kind of odd that you're this utterly clueless.

>There are already Intel 10nm CPUs on the market
That's still very limited production of very small chips which is basically just a stunt to keep the stock from crashing.

>one more product than AMD's 7nm
gee, a company that doesn't have a 7nm node or any other node for that matters and designs, not fabricates, chips doesn't make chips. such insights.

Just face it, Intel's fallen 2-3 years behind. It may not sound like much but it's quite significant. Intel's stuck on 14nm while Qualcomm, Apple and AMD get to use TSMC's 7nm - which is volume production right now.

AMD's one if not two generations behind. Intel's 10nm refresh (which can arrive in as early as 2020) will be better than AMD's 5nm which is at least 10 years away.

So when are we actually supposed to be seeing proper 10nm stuff from intel? How many months of curb-stomping intel with 7nm Zen 2 does AMD have to take advantage of?

Attached: 1471921133185.jpg (407x576, 68K)

It's interesting that you're able to reply with retarded fagtalk when you're clearly unable to read what you're replying to.

AMD does not have any fabrication facilities. They don't and won't have any 5nm node.

trustedreviews.com/news/intel-cannon-lake-specs-performance-release-date-2948967
Assuming it isn't delayed again (lol) then probably 6ish months.

>AMDrones actually believe that Intel will sit idly by until 2020-2021 releasing nothing in between
Intel isn't AMD who can release nothing for 10 years.

If Semiaccurate is accurate that 10nm is dead they're obviously moving their desktop parts to TSMC seeing how their low yield parts are already being built by them.

6 months worth of hoarding for one batch of products?

Imagine 8180 die on 7nm. A whole 1 fully working die from an entire wafer.

Anyone have the estimated yields intel got for that 1-off batch of chips they sent to china? Wasn't it some horrifically low value like 2%?

Maybe.
I'm imagining whole wafers coming back dead.

1.9% and Woodscrews might look good in comparison.

Well, as I asked before, if anyone could get me the estimated yields of the dual cores intel shat out and sent to china, i can use that in combination with this:
caly-technologies.com/die-yield-calculator/
to calculate how bad it would be for intel to try and fab their massive fucking dies on the 10nm node.

Scratch that, I went digging around a bit and found a plebbit post that does what I'm trying to do: reddit.com/r/intel/comments/96cd2g/10nm_yields_is_the_problem_but_what_about/
To sum: 10% yields with a tiny 8.2x8.6mm die gives a defect density of about 4 per square cm. Attempting to fab the monster XCC die with that kind of defect density gives a yield of 0.13%.

Intel would essentially be forced to sell each XCC class 10nm chip for well over $50k USD.

The problem with intel is that they're in denial about 10nm being dead. The corporate structure is wretching pretty bad.

I think it's obvious Intel's only hope is to move their entire production over to TSMC. Just write their own foundry off as a loss at this point.

In awe at the size of this CPU, what an absolute unit.

Knowing Intel the MSRP would be like $200k.

>I think it's obvious Intel's only hope is to move their entire production over to TSMC.
Not happening.

Their only choice at the moment is to move their desktop parts to TSMC until they can figure out 10nm++++ and their own version of 7nm/5nm or pay off billions to TSMC or Samsung for their 7nm/5nm designs.

Intel's power are their fabs, they are not a CPU-only company.

The reflection looks like some old uncle with a massive bush stache and huge chin

>That's one more product than AMD's 7nm, which has zero product
Are you an actual shill, dense, or just obscenely narrow minded? Cooking food one time ever doesn't make you a chef. Intel releasing a tiny batch of gimped 10nm CPUs doesn't constitute a working node.

Cooking food one time makes you a better chef than the guy who cooked zero time

You'll take a known shitty chef/product over an unknown one?
I guess that's your choice

>shitty
Intel's 14nm chip can hit 5.3GHz. AMD's competing product barely hits 4.3GHz

Of course 7nm is already here.

Apple's A12 beats Intel Xeons in single-threaded performance.
Apple's A12X beats almost every laptop out there in CPU performance.

Of course Jow Forums forgot to notice the biggest shift in CPU progress of the last 20 years because Jow Forums is tech illiterate and thinks 'Fruit Company Bad™'.

Attached: Apple A12X.jpg (429x219, 20K)

How about the Instinct MI60? 7nm, out Q4 2018, which is now. Wang even showed it on stage in the new horizon event.

Intel mobile socs are kill
Intel hpc accelerators are kill
Intel 3d xyz point is kill
Intel nand business is going to kill
Intel other μarches are kill, they solely relly upon 2 μarches, desktop and server
Intel fabs are kill
Intel fpgas business are kill ( they bought altera when altera was equal to xilinx, and now xilinx is way ahead)
Intel modems are shit. They sell to apple, only because they sell them cheap.
Every fucking product intel makes, is either killed by them or crushed by the competition.

You fucking idiot, stop comparing clocks on different cous.
How many stages does the pipeline of the intel cpu have, and how many does the amd cpu have?

I'd rather graphene was here or anything other than multicore-jerking.

Another stupid comparison.
....
Try opening the desktop version of youtube on any mobile or even worse, watch a youtube video on the desktop version on a smartphone and check how slow things are.

>multicore-jerking.
For/g/ brainlets, multicore is "jerking".
Multicore is the only solution when you want to greatly increase IPC

amdrones thinking that a billion dollar company intel would sink...all because they said so

when amd was irrelevant for the past decade and they still made it into 2018

aaand here we go again.

Geekbench is THE WORST possible benchmark to compare CPUs of different architecture.

The same shitty "x86 is dead" articles were happening in 2015 when the iPad Pro beat x86 laptops in Geekbench. Only once it was tested in non-synthetic benchmarks it had worse performance even when adjusting for the different power usage (which will usually favour the low-powered CPU).

The same is happening again right now, Geekbench and a comparison of the iPad Pros HARDWARE TRANSCODING compared to SOFTWARE TRANSCODING on Intel CPUs. The first proper tests already show that while the A12x is impressive, its nowhere near x86 CPUs.

Meanwhile the A12 and A12X are beating Intel x86 CPUs in industry-standard SPECfp2006 CPU benchmarks.

Intel is dying, they are the CPU company that can no longer shrink transistors. That means death.

Attached: 1294436354922.png (468x494, 31K)

>optimisim
nice meme
amd will flop so hard its going to be bulldozer 2.0

>tfw you realize Intel's future is making 5G modem chips for Apple devices

Attached: 1366154265779.jpg (365x450, 42K)

While Apple can't make even a single transistor. Are you insane (yes).

Delete this antisemitic thread

Apple has the best fabless semiconductor team in the world, they hit the jackpot when they bought Intrinsity and PA Semi.

And yet they can't make a single fucking transistor.
Did you see what I replied to?

Apple can make the same amount of 7nm transistors as Intel.

Attached: 1330279302257.png (222x211, 4K)

I've made more 7nm transistors than intel too

2.5 years ago

>How many months of curb-stomping intel with 7nm Zen 2 does AMD have to take advantage of?
how many years *
and the answer is 2. then intel gives up and goes on the same 7nm process with a new architecture and wins again

Except they dont?

Why are you lying about something so easily disproven?

For pseudo-intellectual brainlets like yourself, multicore is anything but a destabilizing pile of meme shit which wastes everyone's time from the coders of the software to the user.

You mean besides the fact that 7nm Rome processors are already sampling to customers? Or that AMD had a working 64-core Rome for the presentation where it live stomped dual 8180s?

Attached: fuckinhell.png (1009x953, 285K)

Wrong. TSMC 7nm is smaller in one axis then Intel's proposed 10nm.

Fuck off.

Attached: 1536763698454.png (731x918, 65K)

Apple uses TSMC.

Zen4 is rumored to be 5nm useibg TSMC.

>axis
This image isn't a direct representation of how the BEOL is formed, and are of little value if you don't actually know what the connected poly pitch and minimum metal pitch are.
TSMC's total cell area for logic is way smaller than intel's 10nm. TSMC also has significantly denser SRAM than intel now, intel had previously always led the industry there.
Intel has fallen behind in more than CPP.

>their own foundry
fully depreciated already, not much loss to write off

I LOVE CHINA