Graphene Transistor One Step Closer

MSimon · Post by **MSimon** » Wed Feb 03, 2010 1:33 pm

graphene research

Currently you can get SiC transistors that are rated at 10 W and 2 GHz

http://search.digikey.com/scripts/DkSea ... Transistor

Cost is roughly $10 a watt.

The goal for graphene is 100X - 1,000X the speed - 200 to 2,000 GHz. In actuality the speed improvement is vs silicon. So 20 to 30 GHz power is probably more likely. With the possibility of getting up to 300 GHz. With higher speeds at lower powers.

hanelyp · Post by **hanelyp** » Thu Feb 04, 2010 4:53 am

Any guesses when we may have graphine CPUs? And how fast?

kunkmiester · Post by **kunkmiester** » Thu Feb 04, 2010 5:03 am

I'm not sure how switching speed goes into processing speed, but the impression I get is if they can make a graphene chip, it'll mean that switching speed isn't such an issue anymore. I'll not say that 30Ghz switching means a 30Ghz processor--someone's likely to correct me.

MSimon · Post by **MSimon** » Thu Feb 04, 2010 10:42 am

kunkmiester wrote:I'm not sure how switching speed goes into processing speed, but the impression I get is if they can make a graphene chip, it'll mean that switching speed isn't such an issue anymore. I'll not say that 30Ghz switching means a 30Ghz processor--someone's likely to correct me.

It is not just switching speed. It is wire delay. These days.

It is also about logic levels (decisions - AND - OR - NOT). Generally it is 6 levels of logic between clocks.

So divide transistor speed by 100 to get clock speed. But even that is not straight forward. Power goes up directly with clock speed and with the square of operating voltage. But lower voltage slows transistor operation and makes them more prone to error.

The CPU on your desk top runs at from about .9V to about 1.6V. Depending on its temperature. When it gets hot the voltage and clock speed are dropped. And the fans get turned on. Definitely a kluge.