We can certainly build 4 state logic, Flash already does this to squash 2 bits into a single flash cell.
However if you can't build a 4 state gate with fewer than twice the transistors of the binary gate, you will use more die area. If you fiddle with an inverter you can get it to produce vcc/3 when vcc*2/3 is put on it's gate, but the inverter has a lot of gain between the supplies, so any deviation on the input will be magnified. A circuit with a stairstep-like transfer function is needed to fix this, and that would require lots of parts.
Also in a 2 supply system forcing a gate to output vcc*2/3 would allow DC current to flow, which adds to the power problem. If you use 4 supplies then you'll be wasting area.
Multi-level logic is fine for RAM and storage applications since you only use sqrt(n) decoders, where n is the number of bits, so they can be big and complex.
DeBiers: "A diamond is forever!" SynthDiamondCoalition: "She'll never know!"
Ka-Ching!
Re:It's not about transistor count
on
The Diamond Age
·
· Score: 1
A 600'C operating range would make for some challenging design corners. Or maybe you have to 'prime' it with a blowtorch before it'll work right.
Re:Gemstones as investments.
on
The Diamond Age
·
· Score: 2, Funny
I was buying an engagement ring 6 months ago, and one of the guys there was demanding to know where the stones came from. He appearantly didn't want a 'blood diamond'. I was like: "Dude, for what we're paying, a dozen people BETTER have died smuggling these rocks."
GaAs wafers are only a few inches in diameter, and some are semicircular as well, yet devices made on those substrates end up in our cell phones and PCs. Probably diamond chips will start as a boutique thing, competing with high temperature SOI, rad-hard and discrete power devices. It sounds like it lacks the mobility of III-V semiconductors so probably not RF. Let's not forget MEMS either, we should be able to cut some tough little machines out of diamond.
Still, there are some big problems: #1 No native oxide - you can't grow a super quality oxide from diamond, so one will have to be depositied. Maybe more diamond could be grown for gate insulator, but uniformity will probably be very tricky.
#2 Interconnect - sure your substrate can take 400'C, but those Al/Cu wire are gonna burn! You'll never dope the diamond enough to compete with metal, and nanotubes are way on the horizon. Tungsten interconnect/bond wires?
#3 Lithography on a transparent/prismatic substrate?
Just as III-V has never replaced silicon, diamond never will either. 90% of applications don't worry about heat or power much and they'll stay on silicon. This will prevent economies of scale from making them cheap.
That said, having recently gone through the pain of buying a diamond, I'm never buying a natural one again! (Don't tell my wife.;^)
>if you could pop a pill that would make you never die from something biological, the *average* age you would live to be is about 600, after you calculate in train wrecks, falling down stairs, car crashes, and well, anywhere you can kill yourself mechanically or chemically.
An we thought safety standards now were crazy! We'll all probably just curl up in matrix-esque pods and go about our buisness in a virtual world where we bounce instead of splat.
Actually we've already begun to slow down and will settle at something like 11 billion sometime around 2050. As people get more civilized, they have fewer children. Europe's population is actually shrinking right now, and the US isn't going up or down. As the lesser developed countries mature, their fertility will drop as well.
Moore's law for medicine suggests that in 2020 we'll be adding a year to the average lifespan every year. Moore's law for technology suggests we'll probably have unlimited virtual space to live in.
Um, more like 51 flights, *wham*, 62 more flights *wham*. Seems more like a fairly constant ~2% failure rate.
They'll inspect the TPS in orbit after this though, that should improve rentry safety considerably.
Slashdot Mirror
We can certainly build 4 state logic, Flash already does this to squash 2 bits into a single flash cell.
However if you can't build a 4 state gate with fewer than twice the transistors of the binary gate, you will use more die area. If you fiddle with an inverter you can get it to produce vcc/3 when vcc*2/3 is put on it's gate, but the inverter has a lot of gain between the supplies, so any deviation on the input will be magnified. A circuit with a stairstep-like transfer function is needed to fix this, and that would require lots of parts.
Also in a 2 supply system forcing a gate to output vcc*2/3 would allow DC current to flow, which adds to the power problem. If you use 4 supplies then you'll be wasting area.
Multi-level logic is fine for RAM and storage applications since you only use sqrt(n) decoders, where n is the number of bits, so they can be big and complex.
All your base are belong to us!
DeBiers: "A diamond is forever!"
SynthDiamondCoalition: "She'll never know!"
Ka-Ching!
A 600'C operating range would make for some challenging design corners. Or maybe you have to 'prime' it with a blowtorch before it'll work right.
I was buying an engagement ring 6 months ago, and one of the guys there was demanding to know where the stones came from. He appearantly didn't want a 'blood diamond'. I was like: "Dude, for what we're paying, a dozen people BETTER have died smuggling these rocks."
GaAs wafers are only a few inches in diameter, and some are semicircular as well, yet devices made on those substrates end up in our cell phones and PCs. Probably diamond chips will start as a boutique thing, competing with high temperature SOI, rad-hard and discrete power devices. It sounds like it lacks the mobility of III-V semiconductors so probably not RF. Let's not forget MEMS either, we should be able to cut some tough little machines out of diamond.
;^)
Still, there are some big problems:
#1 No native oxide - you can't grow a super quality oxide from diamond, so one will have to be depositied. Maybe more diamond could be grown for gate insulator, but uniformity will probably be very tricky.
#2 Interconnect - sure your substrate can take 400'C, but those Al/Cu wire are gonna burn! You'll never dope the diamond enough to compete with metal, and nanotubes are way on the horizon. Tungsten interconnect/bond wires?
#3 Lithography on a transparent/prismatic substrate?
Just as III-V has never replaced silicon, diamond never will either. 90% of applications don't worry about heat or power much and they'll stay on silicon. This will prevent economies of scale from making them cheap.
That said, having recently gone through the pain of buying a diamond, I'm never buying a natural one again! (Don't tell my wife.
>if you could pop a pill that would make you never die from something biological, the *average* age you would live to be is about 600, after you calculate in train wrecks, falling down stairs, car crashes, and well, anywhere you can kill yourself mechanically or chemically. An we thought safety standards now were crazy! We'll all probably just curl up in matrix-esque pods and go about our buisness in a virtual world where we bounce instead of splat.
Actually we've already begun to slow down and will settle at something like 11 billion sometime around 2050. As people get more civilized, they have fewer children. Europe's population is actually shrinking right now, and the US isn't going up or down. As the lesser developed countries mature, their fertility will drop as well. Moore's law for medicine suggests that in 2020 we'll be adding a year to the average lifespan every year. Moore's law for technology suggests we'll probably have unlimited virtual space to live in.
Um, more like 51 flights, *wham*, 62 more flights *wham*. Seems more like a fairly constant ~2% failure rate. They'll inspect the TPS in orbit after this though, that should improve rentry safety considerably.
That'll be fine until best buy sues everyone who ever purchased one of those for theft.