Researchers from TSMC and the University of Texas show that SOT-MRAM is uniquely effective for AI applications

Researchers from the University of Texas at Austin, together with TSMC, have produced 300 mm-compatible SOT-MRAM devices and have shown SOT-MRAM is uniquely effective candidate for implementation of crossbar accelerators in memory- and energy-limited applications such as AI.

The SOT-MRAM devices offer 150% tunnel magnetoresistance (TMR) ratio, fast (2 ns) and low voltage (<1 V) operation, low energy dissipation (2 pJ), low write noise (0.1%), and low device-to-device variation of 10%. 

The researchers have shown that SOT-MRAM characteristics are effective for inference on calibrated AI models. They have leveraged the bi-stable anisotropy and stochastic switching of SOT-MRAM to train binary neural networks.