A team of Japanese researchers has proposed a model for the near future of computing based on identifying a 1 from a 0 according to the direction in which an electron spins, rather than where the chip had previously stashed an electrical charge.
The information-processing ability of nearly all modern computing gear is based on the quick-and-effective shuffling of electrical charges around circuits whose telling feature is the vast number of transistors that are present in pairs so that one can be charged to indicate a 1 and the other can be charged to represent a 0.
Spintronics (a portmanteau word meaning "spin transport electronics") is a promising non-volatile memory technology that stores 1s and 0s according to which direction a captive electron spins rather than the electrical charge it can deliver. Chips designed to provide non-volatile magnetoresistive random access memory (MRAM) would use less power, would theoretically run much faster, would be able to store the same data for years even when completely deprived of power, and could store and retrieve far more data far more quickly than conventional DRAM, SRAM, and NAND memory circuits.
Layered structure of computer systems.
(Source: Journal of Applied Physics)
On April 10, 2014, Stanford University professor Stuart Parkin was awarded the million-Euro Millenium Technology Prize for research that led to his proposal of MRAM chips in 1995 and his 1988 development of the giant magnetoresistance (GMR) spin-valve read head that allowed a thousand-fold increase in the storage capacity of magnetic disk drives.
Because they don't have to continually send data away from the main processing cores to be stored in a non-volatile medium, MRAM and other non-volatile processors should allow logic functions and memory to become more integrated, making both memory and processing functions far more efficient. This is according to a description by Pedram Kahlili, who teaches electrical engineering at the University of California, Los Angeles, and who helped develop a variant of the technology called magnetoelectric RAM (MeRAM) that is 1,000 times more efficient than existing memory. Kahlili explained the state of the art of non-volatile memory technology in a February 2013 piece for EETimes.
Several forms of non-volatile memory, including MRAM and spin-transfer torque magnetoresistive random-access memory (STT-MRAM), have been on the market since 2000, primarily as a way to shorten long boot-up times for computers. But they have not been developed to the point that they could take enough of the memory processing work away from volatile memory chips to allow for the development of computers that run effectively using minimal power or none at all. So says a research paper published April 6 in the Journal of Applied Physics by a research team representing the science-in-the-public-interest agency called Japan National Projects.