Read more.Quote:
A 2nm production facility will be ready in Hsinchu, Taiwan, by 2024.
Printable View
Read more.Quote:
A 2nm production facility will be ready in Hsinchu, Taiwan, by 2024.
Finally, Ryzen @ 5Ghz.
2nm! Amazing.
Just WOW!
Even if it takes 5 years, that's mind-boggling! :D
So where do we go after the ever decreasing node? When the day comes that big boys in the CPU market have exhausted a 0.1 or some other ridiculous nm measurement, what are the ideas floating about for consumer processors for the future? Realistically, or theoretically realistic, if such a term exists.
2nm, that's 10 silicon atoms across. It amazes me we got this far tbh.
Moore's law has stopped for Intel, but not for TSMC.
The most likely step is GAAFETs (Gate all-around) where instead of the fin sticking up in the air and being surrounded on three sides by the gate (it's actually the top and two sides but i hope you catch my drift) a bit like a doorway it would be surrounded on four sides a bit like a window, it should go without saying though that like a lot of theories it could end up in the dustbin as trying to fabricate what's essentially a tiny little wire sheathed in another wire at that sort of scale could throw up some unexpected problems.Quote:
Originally Posted by blokeinkent
Isn't 2nm more like 400 silicon atoms? Although it's not really that small though is it? Going on the diagram isn't it 20nm, that being the gate pitch, so it's more like 4k silicon atoms, granted that still really small, but still.Quote:
Originally Posted by DanceswithUnix
EDIT: Having had a quick Google i think i maybe miles off with my recollection of how many silicon atoms are in a nanometer so maybe it's best to disregard ^^that^^. :)
I think these numbers went from gate size to minimum feature size to whatever the marketing guys think sounds nice size.Quote:
Originally Posted by Corky34
But however you cut it, the things are dead tiny :)
Well it certainly beats the process my first Pentium was on (800nm)
lol yeah and I bet it cost twice as much as well :)Quote:
Originally Posted by Iota
Pfft, first computer was an 8 bit Z80 at 1.77MHz, made on a 5um process.Quote:
Originally Posted by Iota
For the youngsters, the Z80 was an enhanced rip off by Zilog of the Intel 8080, where the 8080 was a rip-off by Intel of the Datapoint 2200. The technology gets refined, but I don't think the shenanigans have changed much :)
Not to be a niggle but "scalping" is the decidedly sub-dermal removal of hair from the top of the head. Did you mean to say?...Quote:
TSMC, which recently scalped the Qualcomm Snapdragon 865 SoC and Nvidia Ampere GPU contracts from under Samsung's nose
If so, it's a bit gruesome, don't you think? Just sayin'. ;O)Quote:
TSMC, which recently more than shaved the moustache of the Qualcomm Snapdragon 865 SoC and Nvidia Ampere GPU contracts from under Samsung's nose and removed its upper lip...
I am announcing R&D for 45picometers process. Eat that TSMC.
My first CPU was a 386SX based on 1500nm tech. Amazingly, they only stopped production in 2007!Quote:
Originally Posted by Iota
Crikey, if this stuff gets any smaller they're going to have to employ me to manipulate the atoms with my penis.
No one else has the experience in handling something so small.
LOL! I bow to a master of self-deprecating humour. :-DQuote:
Originally Posted by philehidiot
Mind you, to be comfortable posting such an "admission", it has to be suspected that you have an, er, "atom-manipulating tool" of ginormous proportions, something that your average adult "chip-grinding" star can only dream of. :o)
This is pretty impressive, but "2nm" here would have been equal to 5nm if the same metrics used a decade ago were used to determine feature sizes used here. Compared to a decade ago, assume a process step is equal to a half-node. So a 10nm to 7nm is really a half node jump.
It's taking more and more to achieve less than before.
It's both awesome and scary.
Things are becoming so small that if you fart, You may get a different calculation off the cpu!