Logic-in-Memory (LIM) architectures, particularly those based on Spin-Transfer Torque Magnetic Random-Access Memory (STT-MRAM), offer a promising solution to over-come the von-Neumann bottleneck of traditional computing platforms. The smart material implication (SIMPLY) LIM scheme has been proposed to execute computations efficiently within memory units. This paper presents reliability enhanced STT-MRAM-based SIMPLY and sFALSE operations, assessing improvements in read margin (RM) and bit error rate (BER). Furthermore, the energy efficiency of the SIMPLY scheme is evaluated for the implementation of some logic operators such as NAND, XOR, and XNOR. Simulation results show that SIMPLY + scheme results to be energy efficient, while assuring a 3.7× larger RM as compared to the conventional SIMPLY.
Improving Reliability of STT-MRAM-Based Smart Material Implication
Lanuzza M.;
2024-01-01
Abstract
Logic-in-Memory (LIM) architectures, particularly those based on Spin-Transfer Torque Magnetic Random-Access Memory (STT-MRAM), offer a promising solution to over-come the von-Neumann bottleneck of traditional computing platforms. The smart material implication (SIMPLY) LIM scheme has been proposed to execute computations efficiently within memory units. This paper presents reliability enhanced STT-MRAM-based SIMPLY and sFALSE operations, assessing improvements in read margin (RM) and bit error rate (BER). Furthermore, the energy efficiency of the SIMPLY scheme is evaluated for the implementation of some logic operators such as NAND, XOR, and XNOR. Simulation results show that SIMPLY + scheme results to be energy efficient, while assuring a 3.7× larger RM as compared to the conventional SIMPLY.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
