TeraHertz Transistor

Yeah, it would have patched up the problems for quite a while... but you can only upgrade your PC for so long before you have to rebuild it totally or fall behind.

Dont quote me on this but I believe Intel faced some trouble with this technology and production was halted? I might be confusing technologies though.

Hopefully their transition to 65 will go better than their transition to 90nm, or we're going to have computers which use nuclear ovens as their CPU.

For THz speeds they also should upgrade light speed (doing one 1THz cycle light passes 0.3mm.... they'll need really seriously asynchronous processor) and maybe downgrade Boltzmann constant...

Quote:

Original post by Dmytry For THz speeds they also should upgrade light speed (doing one 1THz cycle light passes 0.3mm.... they'll need really seriously asynchronous processor) and maybe downgrade Boltzmann constant...

Roffle++

Quote:

Original post by Dmytry For THz speeds they also should upgrade light speed (doing one 1THz cycle light passes 0.3mm.... they'll need really seriously asynchronous processor) and maybe downgrade Boltzmann constant...

How so? I did not read the PDFs but I seem to recall that Intel developed a new material superior to SiO₂ as an insulating layer which could reduce leakage while remaining considerably thinner. The transistors could be shrunk to 32nm with a 15nm CL. The 3D transistors could operate at 1THZ and very low voltages. Anyway I believe the technologies required for the THZ transistors that above and the paradigm shift required in design proved to be unwieldy, difficult to implement.

Better linky

Very interesting. To say the least.

Very, very interesting.

Also, would light not be a faster medium?

(from a sciam a while back), Wouldn't you have small silicon lazers, which shine through controlling crystals. (i fogot what they are. They basically change there opacity via voltage... LCD?)

You then have capaciters, which gets hit with the lazer light and looses its charge. Now, you put them both together, and you've got yourslef a very fast lazer repeater (or transister!).

From,
Nice coder

Quote:

Original post by Daerax Quote: Original post by Dmytry For THz speeds they also should upgrade light speed (doing one 1THz cycle light passes 0.3mm.... they'll need really seriously asynchronous processor) and maybe downgrade Boltzmann constant... How so? I did not read the PDFs but I seem to recall that Intel developed a new material superior to SiO2 as an insulating layer which could reduce leakage while remaining considerably thinner. The transistors could be shrunk to 32nm with a 15nm CL. The 3D transistors could operate at 1THZ and very low voltages. Anyway I believe the technologies required for the THZ transistors that above and the paradigm shift required in design proved to be unwieldy, difficult to implement.

how so what?
Speed of light = 3E8 m/s
1THz cycle = 1E-12 s
Distance light passes during 1THz cycle = 0.3 mm. No information can travel faster.

So as i have said, their processor will have to be very asynchronous in sense that they can not keep parts of chip working in synch and have to take light speed into account.

Also, 1THz it's far infrared.

Quote:

Original post by Daerax Quote: Original post by Dmytry For THz speeds they also should upgrade light speed (doing one 1THz cycle light passes 0.3mm.... they'll need really seriously asynchronous processor) and maybe downgrade Boltzmann constant... How so? I did not read the PDFs but I seem to recall that Intel developed a new material superior to SiO2 as an insulating layer which could reduce leakage while remaining considerably thinner. The transistors could be shrunk to 32nm with a 15nm CL. The 3D transistors could operate at 1THZ and very low voltages. Anyway I believe the technologies required for the THZ transistors that above and the paradigm shift required in design proved to be unwieldy, difficult to implement.

how so what?
Speed of light = 3E8 m/s
1THz cycle = 1E-12 s
Distance light passes during 1THz cycle = 0.3 mm. No information can travel faster.

So as i have said, their processor will have to be very asynchronous in sense that they can not keep even blocks on chip working in synch and have to take speed of light into account. That itself is a quite big problem. Also, their leakage is inversely proportinal to speed of light too, and can not be decreased by order of magnitude no matter how they change materials.
In summary, to get from 1GHz to 1THz lot bigger redesign is needed than to get from 1MHz to 1GHz .

Also, 1THz it's actually not even a radio but far infrared.

Not only that, the P4 already has several completely empty pipeline cycles because signals can't get from one end of the chip to the other (party due to signal strength as well; many signals need to be repeated).

Quote:

Original post by Dmytry Quote: Original post by Daerax Quote: Original post by Dmytry For THz speeds they also should upgrade light speed (doing one 1THz cycle light passes 0.3mm.... they'll need really seriously asynchronous processor) and maybe downgrade Boltzmann constant... How so? I did not read the PDFs but I seem to recall that Intel developed a new material superior to SiO2 as an insulating layer which could reduce leakage while remaining considerably thinner. The transistors could be shrunk to 32nm with a 15nm CL. The 3D transistors could operate at 1THZ and very low voltages. Anyway I believe the technologies required for the THZ transistors that above and the paradigm shift required in design proved to be unwieldy, difficult to implement. how so what? Speed of light = 3E8 m/s 1THz cycle = 1E-12 s Distance light passes during 1THz cycle = 0.3 mm. No information can travel faster. So as i have said, their processor will have to be very asynchronous in sense that they can not keep even blocks on chip working in synch and have to take speed of light into account. That itself is a quite big problem. Also, their leakage is inversely proportinal to speed of light too, and can not be decreased by order of magnitude no matter how they change materials. In summary, to get from 1GHz to 1THz lot bigger redesign is needed than to get from 1MHz to 1GHz . Also, 1THz it's actually not even a radio but far infrared.

Oho!! I see! We are not discussing the same thing here. I believe what Intel is refering to when they say Terahertz is the switch charateristic of the transistor and not the fequency of the oscillating crystal which acts as the operational clock for chips (meaning not the speed of the processor). Nonetheless, I think that number might be overly optimistic, an operational frequency obtainable attainable only under the most favorable of circumstances (I will win my fifth Nobel Prize two days after I get struck by lightning 3 times over, seperated by hour intervals, when going to collect my lottery winnings in different states before the condition required for such are ever realized).

I dont know what you mean by the inverse porportionality with regards to current leakage but this redesign of the transistor allowed for 10⁴ more effeciency at leakage stoppage (Intel and their fantastic numbers again).

Anyway as I alluded to above (with the pc upgrading anology) I too think a complete overhaul in current design methods is required to allow speeds of greater than 100~200GHZ (is that number right?).