Check the date – this doesn’t seem to be the April Fools’ joke. Feng Zhao and Brandon Suoka, research engineers at WSU’s School of Engineering and Computer Science, have developed a “honey memorizer” that they hope to use extensively as a key component of neuromorphic hardware. Their report was published in the Journal of Physics.
The process is deceptively simple. First, the researchers hardened and processed a sample of honey. They then compress it into two metallic electrodes. The design is known as the Portmanteau of ‘Memorister’, ‘Memory’ and ‘Resistor’.
What this invention is a memristor, instead of two adhesive pieces of metal, honey itself is a physical property. In particular, Memristor materials have a voltage-dependent resistance. Their structural changes, by contrast, create a kind of physical history of electrical activity throughout the memory. Honey fits that description, since it varies in reverse between a viscous liquid and a crystalline solid. For that reason, honey can serve as a physical history. This is what makes honey honey candidates for use in neuromorphic systems.
Humming sound
Researchers have high hopes for their honey memory that, as neuromorphic hardware, it will be able to surpass a level competitor or even “traditional” Von Neumann architecture. But they also identify their inventions as renewable, biodegradable, non-toxic, low-energy and even antibacterial. “Honey is not wasted,” says author Feng Zhao. “It has a very low moisture concentration, so bacteria can’t survive in it. That means these computer chips will be very stable and reliable for a very long time.”
“When we want to dispose of devices using computer chips made of honey, we can easily dissolve them in water,” he added. “Because of these special properties, honey is very useful for building renewable and biodegradable neuromorphic systems.”
There are some obstacles that they must overcome. The electrical and thermal limitations of the device may be the most important. Zhao and Suoka offer honey memorizers as an ingredient for the neuromorphic system because they work at low energy. A major selling point of neuromorphic systems is that they run in dozens of watts, as opposed to hundreds of watts that a consumer or server CPU can draw. Honey memorizers can handle a maximum of twenty watts and have corresponding thermal boundaries.
Neuromorphic computing is not suitable for all tasks. It is unlikely that we will see neuromorphic processors in desktop hardware anytime soon. Instead, neuromorphic architectures such as Intel’s Lohi lend themselves to big-data research.
In addition to his team work on honey, Zhao hopes to continue exploring other renewable and biological solutions for the use of neuromorphic computing. One possible target: the proteins and sugars found in aloe vera leaves.
John Sullivan, Wikipedia feature image