Neuromorphic Processor with radiation tolerant MRAM

Space missions require high-performance, reliable computing platforms and can function in challenging environments. The von Neumann bottleneck constrains performance due to the time and energy consumed during the required data exchange between main memory chip sets and the processor. Neuromorphic computing could emerge as a game changer for space applications where mission success relies on fast and autonomous analysis of a vast array of incoming information from multiple sources. The future space applications will drive the need for Reduced size, weight, and power constrains Data retention On-board adaptive learning capability Autonomous, onboard, and fast data analysis Neuromorphic processors aligns with the above capabilities. Neuromorphic architectures are inherently fault tolerant, and several hardware implementations have high-radiation tolerance. In addition, neuromorphic algorithms are well-suited to classes of problems of interest to the space community. Present Neuromorphic solutions for Space applications require FLASH memory for boot and weight storage in case of power loss or intermittent power failures. The FLASH memory has limitations on speed and life-time is limited by about 1M cycles of memory operations due to its endurance. For Deep Space Missions where continuous learning is required with updates on the non-volatile memory, a robust radiation tolerant memory with SRAM like performance but still with non volatility and high endurance is required. MRAM which offers 2.5X to 3.5X density advantage over SRAM, 1000X better endurance over FLASH, high radiation tolerance above 100Krad to 1Mrad and ultra-low power standby leakage which is critical for long battery life between solar recharge is a big advantage for these critical SPACE missions. Numem proposes in Phase-I to create a interface system with MRAM which can connect with AKIDA Neuromorphic Processor from Brainchip to either limit or replace FLASH operations with MRAM. More »