Maker Pro
Maker Pro

TeraWave Technology Web-Research Results, tables and supercomputers


Mar 25, 2014
Mar 25, 2014
Some web reading related to the topic of highest frequency generation, which users may find interesting -

Arranged in (roughly) alpabetical order of subject title:

For those interested in the 'extreme' out-there science of frequencies, a Physics-based question/answer forum has posts re Planck wavelengths, string theory, etc.

Hundreds more of Google hits, too numerous to list all here.

Many sites above have additional links to related source material and articles.

A 2012 article mentions a tunneling diode oscillator capable of 1.111 THz electrical frequency.
Six years later, this has no doubt improved.

Some interesting points I remember reading were:

Circuit capacitance and resistance limit the practical applications of Terawave devices.
The highest obtainable frequency of a junction is limited by the structure of its material (how quickly it can change from one energy state to another).
The heat generated vs. cooling capability is another limiting factor.

The highest theoretical frequency possible is at the planck wavelength, about 1E-44m.
As photon energy (GeV) increases with frequency, a point is reached where a sub-microscopic black hole could be formed.

As Gallium Nitride (Gan) and other technology becomes more commonplace, communication, computers, radar, astronomy, weather prediction, medical scans etc will benefit.

My input:

Taking the 1.111 THz tunnel-diode oscillator as an example, and comparing with my current Pentium CPU:
Intel Core 2 Duo E7500 @ 2.93 GHz; Wolfdale 45nm Technology (480 times faster clock-speed).

Fast processing:

Fastest (2018) supercomputer is in USA, the IBM Summit.
Raw clock speed per GPU appears to be 300 Gbps, (I think that calulates to 2x300 = 600 GHz?) but from what I understand, a FLOP is a floating point operation per second, or computation, by moving decimal points in very large and very small numbers.
So a Flop would take multiple clock cycles to compute.
At 200 Petaflops, or 3.3 Exaops, it's fast!!

FYI, Peta, symbol P, is 10^15 or 1E15 units.
Exa, symbol E, is 10^18 or 1E18 units.
The trick that supercomputers use is massive parallel processing.
With a huge network of supercooling methods to prevent meltdown.

Although the combined computational speed of the Summit is fast, in the mid - Petaflop range, the individual Nvidia GPU's are rated at a relatively slower 300 Gbps, still fast for a processor.

Maybe members would like to contribute other url's & snippets, comment, correct, curse.