Moores' law

[IAR117]

I need to know if it's still relevant, for skl

[TheDane]

27 minutes ago, IAR117 said:

I need to know if it's still relevant, for skl

SKL???

[Dash Lambda]

2 hours ago, TheDane said:

SKL???

Only meaning I can find is "school," which makes me wonder what school he goes to where he would decide to use that shortening.

 

@IAR117 Not really. Doubling your component density on the same basic technology only works until you start dealing with quantum mechanics, and microarchitecture is (I think) the only widespread technology where we're at that point.

[IAR117]

6 hours ago, TheDane said:

SKL???

Text slang for school lol

[IAR117]

3 hours ago, Dash Lambda said:

Only meaning I can find is "school," which makes me wonder what school he goes to where he would decide to use that shortening.

 

@IAR117 Not really. Doubling your component density on the same basic technology only works until you start dealing with quantum mechanics, and microarchitecture is (I think) the only widespread technology where we're at that point.

UK?!?! So you're saying that Moore's law is no longer relevant?

[TheDane]

5 hours ago, IAR117 said:

UK?!?! So you're saying that Moore's law is no longer relevant?

It is relevant since it is the best assumption we can make about the evolution of transistors over time. I will first become irrelevant when someone defines a law that is more precise than Moores Law.

 

And please for the love of god stop using slang on LTT when it isn't common knowledge, it just wastes our time and it will take longer before you get your answer

[Dash Lambda]

1 hour ago, TheDane said:

It is relevant since it is the best assumption we can make about the evolution of transistors over time. It will first become irrelevant when someone defines a law that is more precise than Moore's Law.

I'd say that it's irrelevant at this point because it made assumptions about component density and computing power over time that just haven't held up. It's not like standardized units or anything where even if it's not the best definition we need something, it's just a conjecture that doesn't work anymore.

 

EDIT: Reading about it, while it's fallen apart in the last decade, many people (a couple notable ones in the semiconductor industry) say that Moore's Law is still valid just as it was every other time it was challenged. That's a bit short-sited because current semiconductor technology can only support down to a 5nm process node with current projections, and it's certainly getting exponentially more difficult to make those leaps. Essentially, traditional microprocessors follow a logarithmic growth curve, where Moore's Law supposes an exponential one. We're currently at what looks like an inflection point, and will need some other technology for the same advances we've seen.
One possibility is heavily multi-layered architectures, but it'll be a long time before we've resolved the heat, power, and manufacturing issues to the point where we can make fully 3D high-performance processors, not to mention the fact that just having more circuitry can't make single-threaded performance substantially better than it is at this point. The rest is either theoretical or very, very experimental.

[IAR117]

1 hour ago, TheDane said:

It is relevant since it is the best assumption we can make about the evolution of transistors over time. I will first become irrelevant when someone defines a law that is more precise than Moores Law.

 

And please for the love of god stop using slang on LTT when it isn't common knowledge, it just waists our time and i will take longer before you get your answer

Ok, sorry mum