Everspin announced today that CogniMem's CogniBlox product line uses their 4MB MRAM chips. The CogniBlox is a reconfigurable and stackable pattern recognition module that enables a cognitive computing platform. The CogniBlox uses Everspin's MRAM to store and restore the content of the neurons. Each CogniBlox module features four CM1K neuron chips or a total of 4,096 silicon neurons.
Everspin says that MRAM is suitable for massively parallel pattern recognition tasks because it's the only memory technology that offers SRAM-like speeds, non-volatility and unlimited endurance. Having a non-volatile memory solution means that the knowledge learned is saved in the case of a power interruption - without the need for external batteries or other solutions.
CogniBlox can be used to create high-performance cognitive computing, sensor fusion, video analytics and other solutions. Applications span machine vision and speech recognition to data mining. CogniBlox enables system configurations of up to one million silicon neurons using 250 boards that deliver unprecedented performance of 0.13 petaops with a typical power consumption of 500 watts, allowing for 256 million connections every 10 microseconds.