Power/Performance Bits: Jan. 22


Efficient neural net training Researchers from the University of California San Diego and Adesto Technologies teamed up to improve neural network training efficiency with new hardware and algorithms that allow computation to be performed in memory. The team used an energy-efficient spiking neural network for implementing unsupervised learning in hardware. Spiking neural networks more closel... » read more

Can Graphene Be Mass Manufactured?


Since the isolation of graphene in 2004, the high mobility and unique transport properties of 2-dimensional semiconductors have tantalized physicists and materials scientists. Their in-plane carrier transport and lack of dangling bonds potentially can minimize line/edge scattering and other effects of extreme scaling. While 2-D materials cannot compete with silicon at current device dime... » read more

Power/Performance Bits: May 22


Sensing without battery power Engineers at the National University of Singapore developed an IoT-focused sensor chip that can continue operating when its battery runs out of energy. The chip, BATLESS, uses a power management technique that allows it to self-start and continue to function under dim light without any battery assistance. The chip can operate in two different modes: minimum-ene... » read more

Power/Performance Bits: May 24


Reducing MRAM chip area Researchers from Tohoku University developed a technology to stack magnetic tunnel junctions (MTJ) directly on the via without causing deterioration to its electric/magnetic characteristics. The team focused on reducing the memory cell area of spin-transfer torque magnetic random access memory (STT-MRAM) in order to lower manufacturing costs, making them more compe... » read more