Accelerator Architecture For In-Memory Computation of CNN Inferences Using Racetrack Memory


A new technical paper titled "Hardware-software co-exploration with racetrack memory based in-memory computing for CNN inference in embedded systems" was published by researchers at National University of Singapore, A*STAR, Chinese Academy of Sciences, and Hong Kong University of Science and Technology. Abstract "Deep neural networks generate and process large volumes of data, posing challe... » read more

Power/Performance Bits: Feb. 5


Photonic-magnetic memory Researchers at the Eindhoven University of Technology (TU/e) have developed a hybrid photonic-magnetic memory device that takes advantage of the speed of optical writing and stability of magnetic drives. "All-optical switching for data storage has been known for about a decade. When all-optical switching was first observed in ferromagnetic materials - amongst the mo... » read more

Power/Performance Bits: Nov. 10


Singing to your storage Existing research on 'racetrack memory', which uses tiny magnetic wires, each one hundreds of times thinner than a human hair, down which magnetic bits of data run like racing cars around a track, has focused on using either magnetic fields or electric currents to move the data bits down the wires. However, both these options create heat and reduce power efficiency. ... » read more