May 13, 2025
Gadget

Engineers have created a thermodynamic computer that needs to be heated for effective calculations

  • January 24, 2024
  • 0

The founders of Normal Computing, former leading developers of AI and quantum platforms at Google Brain and Alphabet X, aim to bridge the gap between classical and quantum

Engineers have created a thermodynamic computer that needs to be heated for effective calculations

The founders of Normal Computing, former leading developers of AI and quantum platforms at Google Brain and Alphabet X, aim to bridge the gap between classical and quantum computing with this migration option.

How does everything work?

A thermodynamic computer works by heating a matrix of data and then allowing it to cool while the coefficients are naturally calculated and read from records. According to the company, this approach provides higher speed compared to traditional computers and is promising for use in the field of artificial intelligence.

According to documentation on the company’s website, the use of natural processes, particularly thermodynamics, for calculations is consistent with the energy-efficient and fast nature of this innovative computational approach. The founders emphasize its suitability for calculations involving randomness in probabilistic and generative AI.

Normal Computing has successfully developed a prototype thermodynamic computer specifically designed for matrix inverse computing and demonstrated its feasibility. The company claims that this technique can be applied to a variety of linear algebra calculations.

prototype work

The prototype features a matrix of repeated groups of elements known as stochastic processors (SPUs). These units, defined as oscillators with an oscillating capacitive circuit, charge the capacitor to the desired value. Over time, the card reaches thermodynamic equilibrium with the environment, usually the heatsink.

The resulting values ​​in the matrix elements are then read as they cool and the desired calculation result is obtained.

Patrick Coles, Chief Scientist of the Division of Conventional Computing and former Head of the Quantum Computing Division at Los Alamos National Laboratory. expressed doubts about the future of quantum technologiesStating that quantum computers are still in the academic stage and will not bring commercial value yet, Dr.

Coles emphasized that advanced computing, rather than talent, may be the limiting factor in the ubiquity of AI’s benefits. The company’s main goal is to greatly expand the capabilities and efficiency of AI with the new thermodynamic stack.

Source: 24 Tv

Leave a Reply

Your email address will not be published. Required fields are marked *