TPU Inference Servers for Efficient Edge Data Centers - DOWNLOAD

Researchers open up ‘green IoT’ devices based on improvements to magnetic RAM

Researchers open up ‘green IoT’ devices based on improvements to magnetic RAM

A team of researchers at Tokyo University of Science say they have discovered a method to correct a disturbance source in next-generation magnetic random access memory (RAM). The discovery could enable the mass commercialization of the magnetic RAM for ultra-low power IoT electronics, the team says.

A press release from the university notes the growing attention on a new type of magnetic memory known as ‘spin orbit torque RAM’ (SOT-RAM). By using spin-orbit interactions, SOT-RAM can significantly reduce the write current and lower power consumption, which opens the possibility for more energy-efficient electronics and IoT devices. Researchers anticipated that since the memory readout and write current paths are different, potential disturbances on stored values would be minor.

But a team of researchers led by professor Takayuki Kawahara of the Tokyo University of Science found that it was not the case with SOT-RAM. They discovered that disturbance is created because the readout current is actually shared with the write current, leading to an unbalanced spin current that can flip the stored bit. This effect would hinder the reliability of SOT-RAM.

To solve the dissonance, the team developed a new reading method for SOT-RAM published in another study to nullify the readout disturbance. They altered the original SOT-RAM structure to create a bi-directional read path that cancels out disturbances that are conjured by the spin currents when reading a value. To confirm their theory, the team conducted simulations to test its effectiveness. The effectiveness was said to be very favorable.

“We confirmed that the proposed method reduces the readout disturbance by at least 10 times for all material parameters and device geometries compared with the conventional read path in SOT-RAM,” Kawahara comments.

As an additional confirmation, the team tested their method against an array structure that would be used in an actual SOT-RAM. The results here also showed promise, with a sufficient reduction in readout disturbance being viable even with about 1,000 memory cells combined. With their research, the team says they are continuing to refine their method to reach a higher number of integrated cells.

The research team says the discovery helps to build low-power electronics that could be applied in personal computers, mobile devices, and large-scale servers.

“We expect next-generation SOT-RAMs to employ write currents an order of magnitude lower than current STT-RAMs, resulting in significant power savings,” says Kawahara. “The results of our work will help solve one of the inherent problems of SOT-RAMs, which will be essential for their commercialization.”

A greener IoT world might not be too far away. Other recent developments for more energy-efficient RAM include compute-in-memory, with the potential to cut energy consumption by ten times while continuing to boost high-performance processing.

Article Topics

 |   |   |   | 

Comments

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Featured Edge Computing Company

Edge Ecosystem Videos

Latest News