Research by an EPFL team, published in the journal Nature Communications, is a leap forward towards the development of a more energy-efficient computer architecture. Their discovery, in the field of magnonics, makes it possible to change nano-magnets from a state 0 to a state 1.
A team from EPFL has made a discovery that it says promises the development of more sustainable computing. Featured in the journal Nature Communications, their advance takes advantage of magnonics. It is a subfield of engineering that aims to advance information technology in terms of speed, device architecture and energy consumption, explains EPFL in its press release.
A magnon is the precise amount of energy required to cause a collective excitation known as a spin wave to change the magnetization of a material. According to Dirk Grundler, director of the Laboratory of Magnetic and Magnetic Materials in the Nanoscale (LMGN) at the Faculty of Engineering, energy losses are a significant obstacle for electronics, because the speed of data and the requirements in storage material increase. You should know that with the traditional computer architecture, the processors and the memory are separated. This configuration has the disadvantage of slowing down calculations and wasting energy.
Better meet the needs of AI and Big Data
EPFL researchers have therefore sought new computer architectures capable of better meeting the needs of AI and Big Data. Korbinian Baumgaertl, a PhD student at LMGN, came up with the idea of fabricating yttrium iron garnet (YIG) nanomagnetic devices. The researcher succeeded in using radio frequency signals to move the spin waves in the YIG. This procedure made it possible to modify the way the surface nano-magnets were magnetized. “The two possible orientations of these nano-magnets represent the magnetic states 0 and 1, which makes it possible to encode and store digital information”, specifies Dirk Grundler.
This technology has the potential to change the current architectural paradigm of computing systems, ending the power-hungry separation of processors and memory storage. A technology that is part of what is called “memory computing”. The LMGN team is now striving to perfect its method. “Now that we’ve shown that spin waves can write data by moving nanomagnets from state 0 to state 1, we need to work out a process to get them back from state 1 to state 0. This is called switching,” says Dirk Grundler.