Spin-transfer torque magnetic memory (STT-MRAM) has gained significant research interest due to its nonvolatility and zero standby leakage, near unlimited endurance, excellent integration density, acceptable read and write performance, and compatibility with CMOS process technology. However, several obstacles need to be overcome for STT-MRAM to become the universal memory technology. This paper first reviews the fundamentals of STT-MRAM and discusses key experimental breakthroughs. The state of the art in STT-MRAM is then discussed, beginning with the device design concepts and challenges. The corresponding bit-cell design solutions are also presented, followed by the STT-MRAM cache architectures suitable for on-chip applications.

The recent advances in thin-film thermoelectric (TE) materials have created opportunities for on-chip cooling and energy-harvesting with heat-fluxes >100W/cm2. However, it remains unclear how effective these materials are in the context of realistic microprocessor floorplan and workloads. Moreover, these TE materials suffer from contact parasitics that can significantly impact their performance. To evaluate the workload dependent performance of on-chip TE devices, we developed a hierarchical simulation methodology that connects an architectural simulator and a power estimation tool with a thermal simulator capable of simulating TE devices. The well-known HotSpot thermal simulator is modified to incorporate TE equations along with contact parasitics in the TE module. SimpleScalar and McPAT were used to generate the runtime power of different functional units in an Out-of-Order processor across the SPEC2000 workloads. The power-map generated by McPAT is used by our TE enhanced HotSpot simulator to evaluate the cooling and harvesting capabilities of on-chip TE modules. Our results indicate that it is possible to obtain 11°C peak cooling at the hot-spots, or harvest upto 85mW of power from the hot-spots. We also show that on-chip TE devices can aid in boosting the clock frequency of the processor from 1200MHz to 1600MHz under iso-temperature comparison with the no-TE case. This framework also allows for the rapid design space exploration of TE module's material/physical parameters and the optimum placement options for the TE module on the chip floorplan.

The emergence of spin-transfer torque magnetic RAM (STT-MRAM) as a leading candidate for future high-performance nonvolatile memory has led to increased research interest. Current STT-MRAM technology faces several major obstacles in attaining its potential. One of the major issues is in the design of 1T-1MTJ STT-MRAM bit-cells under process variations: the bit-cells need to be significantly upsized to improve bit-cell failure, resulting in increased bit-cell area and power dissipation. In this paper, we analyze four circuit-level solutions that enable smaller 1T-1MTJ STT-MRAM bit-cells with improved yield, namely, bit-line voltage boosting, word-line voltage boosting, access transistor body biasing, and an applied external magnetic field. Results from simulation using 45-nm bulk CMOS access transistor and 40-nm magnetic tunneling junction technology show that word-line voltage boosting can be the best failure mitigation technique. Bit-cells designed with word-line boosting for write has a bit-cell area reduced by > 75% at iso-failure probability, compared to bit-cells without any failure mitigation technique. When bit-cell failure probability is optimized instead, 5 Oe of applied external magnetic field assisted write reduces power consumption by 15% , compared to bit-cells designed without failure mitigation techniques.

Spin-transfer torque magnetic RAM (STT-MRAM) is a promising memory technology for lower level caches because of its high density and nonvolatile nature. However, the high write latency is a bottleneck to its widespread adoption as the future on-chip memory. In this paper, we propose a new cache architecture-asymmetric write architecture with redundant blocks (AWARE)-that can improve the write latency by taking advantage of the asymmetric write characteristics of 1T-1MTJ STT-MRAM bit-cells. Due to the nature of the storage element in STT-MRAM, the time required for the two-state transitions ( 1→ 0 and 0→ 1) is not identical. In other words, one of the state transitions is slower than the other direction. In conventional cache architecture, the overall write latency is limited by the slower transition. However, the AWARE cache design introduces redundant blocks in each row, and they are preset to the initial state that enables the faster transition. Hence the write operations performed in these redundant blocks are much faster than the conventional write scheme. The write latency in AWARE is improved by 30% over conventional cache architecture with no area penalty in the data array. Moreover, the additional tag bits introduced in this technique result in penalty on the total cache area. In addition, the write energy increases modestly by 7% in the proposed cache design. However, this write-energy increase can be mitigated by sacrificing the cache capacity.

Electron-spin based data storage for on-chip memories has the potential for ultra-high density, low power consumption, very high endurance, and reasonably low read/write latency. In this article, we discuss the design challenges associated with spin-transfer torque (STT) MRAM in its state-of-the-art configuration. We propose an alternative bit cell configuration and three new genres of magnetic tunnel junction (MTJ) structures to improve STT-MRAM bit cell stabilities, write endurance, and reduce write energy consumption. The proposed multi-port, multi-pillar MTJ structures offer the unique possibility of electrical and spatial isolation of memory read and write. In order to realize ultralow power under process variations, we propose device, bit-cell and architecture level design techniques. Such design alternatives at multiple levels of design abstraction has been found to achieve substantially enhanced robustness, density, reliability and low power as compared to their charge-based counterparts for future embedded applications.