Originally posted by biz-engineer As far as cost / image quality, I think Falc is right. But he does not comment about the size and weight of the glass to put in front of a larger sensor.
A statement about that was implied in my post.
Because size and weight of the glass to put in front of a sensor is (as a first order approximation) independend of the sensor size (assuming a given image quality), I focussed on the cost of said glass. And that's decreasing with increasing sensor size.
Originally posted by Nicolas06 Outside of a few areas (graphics, scientific computations) general purpose program don't gain much with this approach.
...
So I stand my point. We stagnate.
Even though this doesn't effect my argument, I disagree.
I agree that Moore's law does no longer apply. Point taken.
But Moore's law never was a law, it was a stupid extrapolation from observation. Unlike the laws of physics.
As long as no law of physics needs be broken though (which has never been possible and never will be), there is no direct limit for technological progress and no reason either why it should happen in a linear (or exponential) fashion.
And there fortunately is no law of physics which sets a lower limit for the space-time volume or energy required to do a computation. Even quantum physics sets no such barrier.
Which btw is unlike photography because physics does indeed set a hard barrier how the best possible camera (in a given space-time volume) can perform.
So, there will be progress in computational hardware. But nobody said it will be easy. Maybe, some new materials need be researched, maybe, it is sufficient to further miniaturize volume and power consumption of the current technology.
IMHO, drastically reducing power consumption alone will do the trick. Because wrt all other parameters, we already beat the humain brain made from organic materials. But we are still way below its processing power (*) resulting from its 0.15 quadrillion synapses [
http://postcog.ucd.ie/files/Pakkenberg%202003.pdf] (~300 Petaflops). CMOS has the same processing density but it would simply melt down if scaled to the capacity of our brain (which is no lightweight construction either, with its separate circuitry for cooling and fuel) ...
Once that problem is solved though, we can leave all other problems to them (i.e., those artificial brains). No way the progress could stop there
___
(*) The current #1 supercomputer (33 Petaflops or 1/10 the speed of a single human brain) consumes 18 MW or what 1 million brains consume, leaving a gap of 1:10,000,000 wrt energy efficiency ...
Last edited by falconeye; 03-06-2016 at 03:56 PM.