Linus Torvalds Sees Lots of Hardware Headaches Ahead
1–6 of 6 posts
Re: Linus Torvalds Sees Lots of Hardware Headaches Ahead
#2Wirth's law is relevant for how software is constantly outrunning hardware.
I'm quite pessimistic but I'm sure a lot of stuff involved with os and ui and browsers will have to be thrown away, and built again.
The fact that js doesn't really run well on smartphone proves something is wrong.
Re: Linus Torvalds Sees Lots of Hardware Headaches Ahead
#3>Moore’s Law has guaranteed a doubling of hardware performance every 18 months for decades. But as processor vendors approach the limits of Moore’s Law, many developers will need to reoptimize their code to continue achieving increased performance.
But Moore's law is dead for long time already. I'm using mid-range CPU from 2009 and performance is still reasonable. (while 1999 CPU in 2009 would be rather useless)
Re: Linus Torvalds Sees Lots of Hardware Headaches Ahead
#4>Moore’s Law has guaranteed a doubling of hardware performance every 18 months for decades. But as processor vendors approach the limits of Moore’s Law, many developers will need to reoptimize their code to continue achieving increased performance. But Moore's law is dead for long time already. I'm using mid-range CPU from 2009 and performance is still reasonable. (while 1999 CPU in 2009 would be rather useless)
I think a part of this is that we hit a bottleneck on human data absorption. There's lots of real and hypothetical media formats (like VR) that won't run on a 2009 computer. But so long as you've got your email and websites and video, your machine is still perfectly useful. New computers are doing all those same things. But I remember when the concept of a video running on a computer was just wild.
Re: Linus Torvalds Sees Lots of Hardware Headaches Ahead
#5>Moore’s Law has guaranteed a doubling of hardware performance every 18 months for decades. But as processor vendors approach the limits of Moore’s Law, many developers will need to reoptimize their code to continue achieving increased performance. But Moore's law is dead for long time already. I'm using mid-range CPU from 2009 and performance is still reasonable. (while 1999 CPU in 2009 would be rather useless)
>I'm using mid-range CPU from 2009 and performance is still reasonable.
Performance was reasonable ... until the whole speculative execution happened. How reasonable is no Hyper Threading?
Re: Linus Torvalds Sees Lots of Hardware Headaches Ahead
#6>Moore’s Law has guaranteed a doubling of hardware performance every 18 months for decades. But as processor vendors approach the limits of Moore’s Law, many developers will need to reoptimize their code to continue achieving increased performance. But Moore's law is dead for long time already. I'm using mid-range CPU from 2009 and performance is still reasonable. (while 1999 CPU in 2009 would be rather useless)
Moore's law was about transistor counts which have continued upwards, though not at the same place recently. Saying that Moore's law has been long dead because serial performance has not increased by many orders of magnitude is a severe misunderstanding.