Yet no one in the industry is panicking. Why? Does Intel or IBM have an ace up there sleeves that every industry insider knows will continue to fuel exponential growth?
Or is ITRS no longer considered the leading authority on setting roadmaps for the semiconductor industry?
Our whole economic system depends on computer processors getting exponentially better, smaller, faster and cheaper. If no one knows what'll happen in 2020s, shouldnt we at least be pushing governments to expand support to riskier projects that may extend moores law? Am I missing something?
spectrum.ieee.org/semiconductors/devices/transistors-could-stop-shrinking-in-2021
>>57788966
"Law"
It already is dead. Nobody in the industry is panicking because they're all working with the same law of physics.
>>57788966
>Does Intel or IBM have an ace up there sleeves that every industry insider knows will continue to fuel exponential growth?
Nope.
Everybody relies on ASML now.
>>57788966
http://semiaccurate.com/forums/showpost.php?p=270664&postcount=10
This thread isn't worth any more response than this.
>>57788966
nobodies panicking because they are locking down systems so software can slowly drop performance making old chips look bad over time. Then release a new "more powerful" chip that work well with the newest rehash of the software. Look at the iPhone and other systems, only a few % point better on paper, yet usability is hugely different because of software tricks. Those who don't update are given lower services or just dropped, forcing new hardware that is not really needed. Planned obsolesces at its finest.
>>57790430
You're such a fucking moron
My processor design professor gets triggered so hard when you mention Moore's law to him. He goes off on a rant about how glorious the days of guaranteed yearly single core performance increase was and how multicore processors are a comedic stopgap that keep us from facing the truth that we can't shove more transistors onto a CPU without having more and more dark silicon. It's actually kind of impressive how valiantly he fights against Intel's jewery.
>>57790430
Anyone can make a faster cellphone than Apple when they don't need to worry about battery life. This is beside the point. Everyone faces the same issues when attempting to shrink die sizes even further. Exploits like row hammer take advantage of quantum tunneling which only became a recent issue. Silicon is on its way out because of known limitations and people have been experimenting with different substrates for over a decade. Moores law was never even a "law". The only gains to be made until a better substrate is economically feasible are in parallel computing.
>>57788966
>>57789064
it's dead since 2012
>>57788966
They are figuring AMD will go bankrupt and everyone will pay double for 1% IPC gains.
>Thus you played yourself.
>>57788966
Die stacking when?
>>57791380
>noise reducting card for video
the fact that it is sold out on newegg is terrifying.
>>57791504
wait what is this I'm a dum