I saw something that said the controversial Moore’s law was now down to 13 months and has been for more than a decade. Expect waves of disruption. Sorry no link but it should be out there.
You do know what Moore’s law referred to? Many use it to say what it never intended to and doesn’t actually work for.
For instance some say that computing power follows Moore’s law, and they falsely base that on #transistors on a die directly relates to computing power. Or that disk storage follows Moore’s law.
Moore’s law was a theory and has over the decades seen differences to reality on the positive/negative side of the theory.
Also we now can see the limits to transistor sizing and at some stage Moore’s law ceases to exists when that limit is approached/reached.
After that we will have different technology to create logic systems and of necessity a new “law” will be proposed to describe the growth of that new technology.
I know, but this person was saying roughly cost availability of flops in application etc. Graphics cards get held up a lot. But of course you’re right. Still on these innacurate terms people used to cite 24 months.
Yea, lately the trend has been to increase the die size, add more processors, memory etc. And each of the types of logic have different density abilities. It seems to be that the increase in density is more to do with adding more dense structures than actually reducing the overall density.
In other words you can fit more kittens than humans into a phone box, and so they optimise their cpus to gain their extra abilities (speed etc) not by making the ALU logic more dense, but by adding denser components that also have the effect of speeding up the overall CPU. EG adding cache speeds up overall processing
Thus it appears on gross figures that chip densities have increased by ‘Y’ but in reality for particular logic units the density has increased by only ‘X’. Where X is less than Y. So I can make my pretty charts and proclaim the density increase rate is now higher but in fact that is over inflating the reality. Makes a good lecture but is a tad misleading.
And this is in face of the quickly looming limitations of transistor size. Even quantium transistors have density limits although their reported compute speed is higher than traditional electronics.
For a number of years all the big chip designers/developers have been researching how to make 3D chip designs where logic units can be more multistory structures and improve the apparent density.
Of course the factors that slows down the density then the rate increases when that is solved is heat. Each time density is increased the speed&density is limited by heat. Heat has been the major factor in the rate being so low. After all we already have the (non-quantium) 1 atom transistor storage(memory) cell demonstrated by IBM research labs I believe. But heat defeats its ability to be made into large scale memory systems.
Some try to include DNA storage into the rate which is not what the theory covers. But DNA storage is apparently the highest density storage we have to date, just not commercially usable yet.
Some Flash memory systems are now using minor 3D layering to increase memory per chip.
Disk storage follows another law 10x every 5 years. But when flash (or other semiconductor) memory takes over from disk storage then the increase rate for storage will more follow Moore’s law. At that time expect Flash to drop further in price because of increased volume.
It is an interesting topic the speed at which the various aspect of computing are increasing.
The biggest looming hurdle is the density limit of semiconductor transistor technology, and always heat