spin-transfer torque magnetic random access memory (STT-MRAM)

A Practical MRAM Logic-in-Memory Platform With Reference-Bitline Structure

A Practical MRAM Logic-in-Memory Platform With Reference-Bitline Structure 150 150

Abstract:

The memory wall due to data movement is a major challenge that limits performance and energy efficiency in bitwise-operation-intensive applications. To address this issue, this article proposes a global merged reference-based magnetic random access memory (MRAM) logic-in-memory (LIM) structure. The proposed structure generates operation-specific reference resistances by controlling the effective …

View on IEEE Xplore

Sensing- and Periphery-Aware Modeling of a Hybrid MTJ–CMOS Nonvolatile 1-bit Sign-Weight Memory Subsystem for Energy-Constrained Edge Inference

Sensing- and Periphery-Aware Modeling of a Hybrid MTJ–CMOS Nonvolatile 1-bit Sign-Weight Memory Subsystem for Energy-Constrained Edge Inference 150 150

Abstract:

The spin-transfer-torque magnetic random access memory (STT-MRAM) is attractive for nonvolatile 1-bit sign-weight storage, but a favorable cell or local-read metric does not establish subsystem efficiency. We analyze a hybrid magnetic tunnel junction (MTJ)–complementary metal–oxide–semiconductor (CMOS) memory subsystem, in which MTJs store static signs while sensing, reference …

View on IEEE Xplore

Merged-Reference-Compatible Offset-Canceling Discharge-Based Sensing Circuit for STT-MRAM in 28-nm CMOS

Merged-Reference-Compatible Offset-Canceling Discharge-Based Sensing Circuit for STT-MRAM in 28-nm CMOS 150 150

Abstract:

This article proposes an offset-canceling (OC) discharge-based sensing circuit (OCDC) for spin-transfer-torque magnetic random access memory (STT-MRAM). The proposed circuit simultaneously implements a merged reference line (MRL) architecture and threshold voltage ( $\mathbf {V}_{\mathbf {TH}}$ ) mismatch cancellation, resolving the structural incompatibility between the conventional OC schemes and MRL. A diode-connection-based, …

View on IEEE Xplore

A Nonvolatile AI-Edge Processor With Lossless-Compressed-Computing STT-MRAM Near-Memory-Compute Macro Using Dynamic Floating-/Fixed-Point Accumulation

A Nonvolatile AI-Edge Processor With Lossless-Compressed-Computing STT-MRAM Near-Memory-Compute Macro Using Dynamic Floating-/Fixed-Point Accumulation 150 150

Abstract:

Nonvolatile AI-edge processors based on near-memory-compute (nvNMC) enable energy-efficient multiply-and-accumulate (MAC) operations with short wakeup latency for edge inference operations. Lossless compression is required for floating-point (FP) neural network (NN) models under on-chip memory capacity constraints; however, this imposes several challenges: 1) data decompression overhead due to lossless FP weight encoding; 2) …

View on IEEE Xplore