Remember when Intel’s manufacturing was so revered that people took reports like this seriously? We used to believe Intel could defy the laws of physics.
Now TSMC’s upcoming 5nm EUV process is looking to be the defining node of the next few years, and Intel can’t even reach parity with TSMC’s first-gen 7nm process.
What went wrong?
Other urls found in this thread:
en.wikipedia.org
twitter.com
TSMC's 7nm is the equivalent of Intel's 10nm
Also, it's not always good to be the first to go. Intel knows this ever since they laid the first finfet back in the ivy bridge days.
Cope.
I thought this was a topic about lithography, not some teenager reading headlines.
It's actually Samsung who has the first EUV, laying the Exynos 9825, with yield issues
TSMC is laying a 5G version of HiSilicon's Kirin 990 in EUV, with likely yield issues
EUV has been in R&D for more than a decade, you aren't fucking kidding me child. Lithography is perhaps my favorite subject, even though I understand little.
You on otherhand... you don't know anything.
Ryzen 4GHz is the equivalent of Intel 5GHz
Is that because 10nm also doesn't clock past 4ghz either?
Why yes, I do support TSMC and Samsung samurai overlords annihilating Intel jews.
TSMC 7nm will go up to 4.7 with 3950X
What does this have to do with lithography? And even that is a blatant misnomer. An ever-more wide design is not intended to operate at even higher frequency.
>Our second generation 7nm (N7+) technology entered risk production in August 2018, and will be the industry's first commercially available EUV process technology.
Based TSMC
Cope.
The goal is better IPC, not simply increasing cycles.
Im calling bullshit.
They will have better architecture thanks to Jim shitwrecker Keller and they will be better than AMD but there will not be a "leap" that will change the computing industry.
>it's a gaming benchmark that shows absolutely nothing but a GPU bottleneck
And seethe.
>wait for sekrit sauce superpower uarch bro
What, sunny cope in 2022?
By the time Intel gets to 7nm sunny cope in 2022 AMD will be on 3nm with Zen6 and 32 mainstream cores.
it's still a GPU bottleneck, it isn't showing anything
and besides this, there exist only 2 chips in EUV, and both of them are sub 100m^2 SoC, not desktop chips, and nobody has any reliability charts on that yet
By the time Intel gets to 5nm AMD will be on 1nm quandum CPUs and sell them even to advanced alien races.
Intel is skipping to 7nm.
you're literally looking at the difference a little cache made, which was zilch in the real world.
Let's just sell more cores to oblivious hobbyists
Whatever helps you cope at night.
keep saying the same thing, it really makes you look more intelligent
en.wikipedia.org
>2019 Ice Lake 10nm
>2019 / 2020 Lakefield 10nm
>2020 Tiger Lake 10m
>2021 Alder Lake 10nm
OH NONONNONONO
>(2022) Meteor Lake 7nm
FUCKING 2022 FOR 7NM AHAHAHAAHAH
What is a list of processors all but confirmed to have no desktop versions going to do?
You're getting comet lake and rocket lake and you'll enjoy your skylake based housefires.
Have you ever heard of trickledown?
The fabrication starts with simpler designs and becomes more elaborate as the technology stabilizes
But keep it up, no I haven't watched this side of the industry for 20 years. Nope.
Fuck shitlake. Let it die. Ringbus is dead.
AMD will be on refined 5nm+ by 2022 dude.
Intel doesn't make mobile SoC. The only EUV so far is a mobile SoC. You can't just massively die shrink a full-fledged CPU like that.
TSMC's 7nm for AMD's desktop chips is still FinFet, it is not EUV.
>DUDE QUANTUM TRANSISTORS 10NM LMAO
Intel is truly finished.
TL'DR a company saving valuable resources by not blowing it on a low-yield design that isn't ready yet.
TSMC isn't a fabless either. A fabless summons TSMC to fabricate their litho.
10nm won't be on desktop consumer mass market you dumbass. It'll be 7nm between 2021-2023.
>TSMC's 5nm technology is the second available EUV process technology. It showed promising imaging capability with expected good wafer yield.
>Our 5nm technology entered risk production in March 2019 and target for volume production in 2020.
With clocks that are under 4Ghz on mutli core while going in housefire mode in signle clock. top kek
they all are shit
>mutli core
We won't call 4 cores "multicore" by that time. That will be the equivalent of a pentium right now.
>muh bottleneck
You really don't know how limiting factors work, do you?
Nice 0.1 stutters, now install the mitigations and disable HT :^)
Literally slower than 14+++++++
Just wait for 10++
What about this one?
Remember: CS:GO was specifically tailored to the C2D era and at MAXIMUM could efficiently use 2 cores. i9 has 8 cores and 3900X has 12 cores.
4 cores already is the equivalent of low end, anything else is poverty tier.
But Ice Lake can't go over 4 cores.
So? Intel’s 10nm is practically vaporware and TSMC is on 7nm gen 2.
>something that exists is the equivalent of something that doesn't exist
Maybe.
But do we have any confidence in intel’s 7nm process? I think analyst’s have a high degree of confidence that TSMC will reach 5nm EUV next year. But Intel has only announced one 7nm product—a GPU—for 2021. CPUs could very easily slip to 2022 or beyond.
I wonder how many pluses 10nm is going to have by the time that is out.
>1080p high settings
>GPU bottleneck
Jim is working on car chips.
This article speculation has more holes than Intel's Hyper threading technology.
Are you retarded? He left Tesla a long time ago. Hes in charge of the entire Intel computing group
Wasnt Jim working on Intel's "GPU" tho?
Both Raja and him working on it would be overkill.
The days of superstar architects are long gone.
>EUV has been in R&D for more than a decade
It's been R&D's since the 1990's.
>iPhone poster
Saged, reported, hidden, ignored, dropped, cringe and bluepilled.
Shit in street Rakesh
>10nm
So, like they are going to be releasing older hipster tech for nostalgia or something?
>TSMC's 7nm is the equivalent of Intel's 10nm
It is not. TSMC's 7nm doesn't fry itself when goes beyond 4ghz.
I want to see what Intel delivers, not what the predictions are. I suppose articles like this are actually a form of advertising/shilling?
ok, this is epic denial
Actually I was wrong. Intel's CPU transistor density is about twice that of TSMC's.
100M transistors per square mm
also, 14nm++ led to a decrease in density, from roughly 45M to 37M per sq.mm, hence some fluctuations as the node and lithography matured
Intel's 7nm further down the road takes the density close to a target 250M transistors per sq.mm.
This more than a 100x than that of Penryn or 45nm