Moore's law LED style

Is there a rule of thumb for the improvement progression on LED output over time similar to that used for computing power (which I believe is a doubling every 18 months)? And how reliable is this law expected to be as far as looking into the future?

I am guessing the history is much more limited for LED output than for computing power, and therefore a bit more of a stretch to apply history to future projections. Also I do not know that the fundamentals are the same as for computing. Seems that step-changes - i.e. fundamental "inventions" - are required for LED's, as contrasted to chips where technology/manufacturing improvements continue to drive the trend to cram more into less space. Just guessing - I am sure this is oversimplified.

Thoughts?

I believe so... expecially when you compare it with the Incand's... it is following the Moore's law to some extent.... (read my latest posts on this subject for McGizmo's lights and also for HDS Lights).... The lights have great electronics... and have stabilized... the lights will also last for years and years to come... the ONLY thing that currently i view is the LED that is improving .... and this needs to be taken care of by making lights easily moddable. (not that HDS/McGizmo's lights are difficult to open/modify)

lightrod said:

Is there a rule of thumb for the improvement progression on LED output over time similar to that used for computing power (which I believe is a doubling every 18 months)? And how reliable is this law expected to be as far as looking into the future?

I am guessing the history is much more limited for LED output than for computing power, and therefore a bit more of a stretch to apply history to future projections.

Click to expand...

Moore's law isn't about power, it's about the number of transistors on a mainstream CPU doubling every 24 months.

Big difference. Apples to oranges.

Moore's law is one of the most mis-quoted phrases of modern times.

amanichen said:

Moore's law isn't about power, it's about the number of transistors on a mainstream CPU doubling every 24 months.

Big difference. Apples to oranges.

Moore's law is one of the most mis-quoted phrases of modern times.

Click to expand...

Yes, but the increased density of transistors per inch DOES decrease the processing time of a processor for a given architecture. This increases the number of computations possible per second, increasing the perceived "processing power".

Moore's law is correlated to processing power because that's what people care about. Nobody generally gives a poop how many transistors we can cram onto a chip. If it didn't make processors faster we wouldn't do it. Now quit being pedantic.

Rando said:

Yes, but the increased density of transistors per inch DOES decrease the processing time of a processor for a given architecture. This increases the number of computations possible per second, increasing the perceived "processing power".

Moore's law is correlated to processing power because that's what people care about. Nobody generally gives a poop how many transistors we can cram onto a chip. If it didn't make processors faster we wouldn't do it. Now quit being pedantic.

Click to expand...

Again, Moore's Law is one of the most frequently mis-quoted and misunderstood phrases in modern times.

If you want to call me pedantic, I'll call you ignorant, but I'd rather keep on topic and away from stupid insults.

Actually Intel did it. (Put more transitiors and lowered performance at the same time) They called it the Pentium 4.

Moore's law only cares about transistor count he didn't give a damn about performance because performance can be measured in so many ways.

One chip might be able to do FPU calculations and be terrible at Integers and one that's viceversa.

I dunno, I think moore's law is relevant although maybe not a direct analogy. CPU technology evolves due to improvements in manufacturing on the nano scale (i.e. more transistors per unit area). This is made possible by the production of more refined semiconductor materials with better thermal properties.

LED technology has little to do with nanoscale engineering (it's not about cramming a lot on to one chip) but the refinement of semiconductor materials and thermal properties are very relevant.

I think some of you might be being a bit too literal, I think what Lightrod meant is can we expect LED technology to advance as quickly as computers.

While LED as a light source will continue to develop, I don't think there is any chance that it will progress in accordance with Moore's law. First of all, it doesn't have the market motivation that microprocessors have, and secondly it's now owned by a European company that isn't really known for taking the kinds of risks necessary for those kinds of gains, and isn't really known for innovation.

lightrod said:

Is there a rule of thumb for the improvement progression on LED output over time similar to that used for computing power (which I believe is a doubling every 18 months)?

Click to expand...

All the evidence I've seen points to a aggresive 'arms-race' for more brighter/efficient/cheaper LEDs. (and sometimes all three at once!)

Moores Law It is not. It IS certain that a 2006 LED is better than a 1997 LED, yes?

So would not a 2016 LED Utterly Eclipse a 2006 LED...only time will tell.

Also more transistors = more heat and less electrical efficiency as they get smaller.

Thus no modern CPU's have gone past 4 Ghz. (we should have 7,8,9+ Ghz systems now)

However there are 2 core, and 4+ core processors out now (or soon).

CPU's have a geometric rate of increase, LEDs are more linear.

That said, they will be the top of the food chain in small scale lighting in ten years.

Thanks for the excellent points folks. Still not really clear on what history has shown so far. As far as the future - guess we'll just have to see (makes it all a bit more fun I suppose!).

One might argue that during CPF's short existance, the combination of brightness, efficiency, cost/value, size, manufacturers, power sources and plethora new features has, roughly, followed Moore's law.

tangentially speaking...

Ray Kurzweil posits that "Moore's Law" (or more precisely, the corrolary about computing power) predates Moore, and actually goes all the way back to the abacus. Sometimes I wonder if his projections are overly optimistic, but I have to admit that the graphs are spooky:

http://www.kurzweilai.net/articles/art0134.html?printable=1

Melchior said:

Also more transistors = more heat and less electrical efficiency as they get smaller.

Thus no modern CPU's have gone past 4 Ghz. (we should have 7,8,9+ Ghz systems now)

Click to expand...

Just FYI, IBM SiGe chips can operate up to 350GHz at room temp. Without over the top nitrogen cooling or stuff like that.

Re: tangentially speaking...

NFW said:

Ray Kurzweil posits that "Moore's Law" (or more precisely, the corrolary about computing power) predates Moore, and actually goes all the way back to the abacus. Sometimes I wonder if his projections are overly optimistic, but I have to admit that the graphs are spooky:

http://www.kurzweilai.net/articles/art0134.html?printable=1

Click to expand...

Technology singularity believers are the modern-day equivalent of the centuries old practice of using religious and numerological "evidence" to predict doomsday.