Plasmon Computing: A Summary of the IEEE Spectrum Interview with De Los Santos
This text details a new approach to computing called plasmon computing, developed by De Los Santos, aiming to overcome the limitations of conventional CMOS technology. Here’s a breakdown of the key concepts:
The Problem with CMOS:
* Downscaling Limits: CMOS relies on shrinking transistors, but this leads to quantum effects, leakage, and increased power dissipation (“short channel effects”).
* Power Consumption: Current technology is struggling to address the high power demands of applications like AI.
the Solution: plasmon Computing
* Inspiration: Inspired by “fluidic logic” (using fluids like air jets to perform logic), De Los Santos proposes using “plasmons” as the fluidic element.
* What are Plasmons? They are disturbances in electron density – like waves on the surface of water, but in a “sea of electrons.” They are created by interacting electrons with electromagnetic waves.
* Why Plasmons are Promising:
* Low Energy: exciting plasmons requires very little energy (attoJoules or less).
* High Speed: Plasmons propagate at the speed of light in the medium.
* Low Power Dissipation: No power is used unless a plasmon wave is actively propagating. CMOS dissipates power even when idle.
How Logic Operations Work:
* Y-Shaped Wires: Logic is implemented using long, thin wires shaped like the letter “Y.”
* Bias Plasmon: A plasmon is launched at the base of the Y, representing a bit of facts.
* Splitting: Without any other input,the plasmon splits into two equal signals at the output of the Y.
* (The text ends mid-description, but implies further manipulation of the plasmon at the Y junction will implement logic functions.)
In essence, plasmon computing aims to create a faster, more energy-efficient computing paradigm by leveraging the properties of electron density waves instead of relying on the limitations of shrinking transistors. The timing is particularly relevant given the increasing power demands of AI.