• Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Moore's law (always get more)

Status
Not open for further replies.

rjvh

New Member
just wondering and i gues there are some chip developers on this forum that can give an answer

according moore's law every 2 years the capacity/capability of a procesor is twice as it whas 2 years before

we seen all the rise in speed of the procesors we see now the multy cores poping up higher than 8 cores comercialy available

where does it end??
like the materials where we work with have also their maximums what they can doo

for example is it still posible to go higher in clock frequency or are the matterial properties used completly?

how many layers we can build in a procesor ?? and is there a maximum ??

can we still put more transistors on a square mm than that we do now ???

can sombody give me some insight in this

Robert-Jan
 

dknguyen

Well-Known Member
Most Helpful Member
rjvh said:
just wondering and i gues there are some chip developers on this forum that can give an answer

according moore's law every 2 years the capacity/capability of a procesor is twice as it whas 2 years before

we seen all the rise in speed of the procesors we see now the multy cores poping up higher than 8 cores comercialy available

where does it end??
like the materials where we work with have also their maximums what they can doo

for example is it still posible to go higher in clock frequency or are the matterial properties used completly?

how many layers we can build in a procesor ?? and is there a maximum ??

can we still put more transistors on a square mm than that we do now ???

can sombody give me some insight in this

Robert-Jan
Of course, there's a maximum on everything. But everytime we reach current max that is technologically possible, R&D finds more ways to add more layers, make smaller transistors, etc.

One of the biggest things is a while back they found ways to etch silicon features smaller than the wavelength of deep UV light used photolithigraphy (deep UV photolithograpy being the most developed and dominant mass production technology) allowing them to continue use basically the same mature technology- so then they didn't have to resort to more difficult and more expensive "extreme UV".

The issue? Materials that can work with extreme UV is much harder to get (I don't think they've found them yet, let alone feasible ones) than those that work with deep UV. THe same problem goes for x-rays and other things like that- the materials that pass those wavelengths at efficiencies good enough to be viable don't exist yet.

And then there is the electron beam- as far as I know it can make the finest features of all and is very accurate. The drawback? It's not a mass-production method (imagine drawing in things bit-by-bit like a pencil, that is electron-beam. Now imagine colouring in a stencil, that is photolithography). Electron beam stuff takes much longer to make than photo-lithography stuff, but it produces much better quality things with low turnaround time, so it is used to make the masks for photolithographic processes. It is also used in research where quantities are small and turnaround time needs to be high.

As far as materials used in the product themselves as opposed to producing the product, I believe people are working on electronics that use ballistic electron, optics, and germanium-typish diodes, even organic molecules, and quantum that are both faster, lower power, and more efficient than silicon. BUt reliability and mass production are the biggest caveats of any new technology and is often the biggest hurdle that must be overcome. Technologies that are extending silicon further are things like straining or compressing the silicon so it alters the crystal lattice and improves the carrier mobility of the silicon (Intel is using this right now).

THere is an unbelievable amount of research going into keeping up with Moore's Law. I think I heard that the semiconductor industry reinvests 30% of it's profits back into R&D which is a ridiculous amount. For comparison I think the auto industry is somewhere around 5-7%, maybe 10%.

You might have also noticed that it's a bit hard to find processors faster than 4Ghz, and you find more "higher performance processors" that are multi-core and run at lower clock speeds. Guess why that is? THe speed limit has gotten more expensive and difficult to overcome than making multiple parallel cores.
 
Last edited:

rjvh

New Member
Thanks for your explanation dknguyen

so moore's law will be slowing dow considarebly if i see your explanation

Robert-Jan
 

dknguyen

Well-Known Member
Most Helpful Member
rjvh said:
Thanks for your explanation dknguyen

so moore's law will be slowing dow considarebly if i see your explanation

Robert-Jan
Only if the researchers' work doesn't go anywhere. THere is the question about whether the pace of development is due to Moore's Law, or whether the pace of development is due to industry working hard to keep up with Moore's Law just because it exists.
 
Last edited:

quixotron

New Member
Yeah I have a buddy that works at Intel, he stated the fastest clock speeds poop out at 3.8 Ghz. Intel wasn't able to reach/cool down at 4 Ghz. Thats where the multi-core processors come into play. One way to speed up a program is to have a faster processor, however if you took computer architecture or microcomputer course, you'll know that you can get the same performance by subdividing the task and using RISC type processors.


Pentium used finite state machines in order to optimize cpu rocesses and rogram running times, they have been doing this for several years.

So yeah, chip manufacturers know the ups and downs of the cpu industry. I frget what law it is, but I recall that even if you have an infinite number of processors, you'll only improve performance upto 4 X max. Research is definitely there.

even so, programs like electro-magnetic device simulators require tons of processing power and memory.. The cpu I use has 16GB of memory and the complex functions still take hours to run. One of my programs took seven days to complete-a department record I think!
 
Status
Not open for further replies.

Latest threads

EE World Online Articles

Loading
Top