News Water-Based Chips Could Be Breakthrough For Neural Networking, AI

[Admin]

Researchers at Harvard University in partnership with biotech startup DNA Script have designed a processor that uses ions instead of electrons to process information. The research has gotten the first integrated circuit of ionic transistors to perform a calculation.

Water-Based Chips Could Be Breakthrough For Neural Networking, AI : Read more

 

[Rob1C]

The link to the Advanced Materials site in the article here requires an account to read the paper, this link does not: https://www.researchgate.net/publication/362891571_Aqueous_analog_MAC_machine

 

[bit_user]

@Francisco Alexandre Pires , thanks for the writeup!

But...

Ionic processor design aims to bring learnings lessons learned from biology's processing systems

FTFY. Please don't perpetuate this abuse of the English language.

The researchers further tuned their microprocessor design, placing these analogue transistors (capable of representing a 0 or a 1) into a 16 x 16 matrix grid array. This allowed the ionic processor to perform matrix multiplication tasks,

I don't see how individual transistors would enable digital matrix multiplication. I suspect they're actually multiplying them in the analog domain. Analog computing is also much more power-efficient, though it tends to be inherently noisy.

Ah, and the subsequent quote confirms this:

“Our ionic circuit performs the matrix multiplication in water in an analog manner that is based fully on electrochemical machinery.”

Lastly,

“It will be very interesting to employ more diverse ionic species and to see how we can exploit them to make rich the contents of information to be processed.”

Implies they're either contemplating non-base2 arithmetic, or perhaps there's some way they can use different ions to enable vector processing?