Search

With the process node further scaling down, the bit cell of magnetic random access memory (MRAM) device suffers from severe failure and reliability problems due to its process variation. In this article, we explored the variability of a promising toggle spin torques MRAM (TST-MRAM) in detail, which achieves ultrafast switching speed and high endurance and is expected to be applied as upper-level caches. Our research is conducted with a 14 nm FinFET design kit, the basic performance of TST-MRAM and different pulse conditions are explored, and then the variation effect of device parameters is discussed and compared in detail as well as the write error rate (WER) under different voltages. The results demonstrate that the variation of access transistors dominates the variability, where the transistor in the spin–orbit torque (SOT) branch has a greater negative impact than the transistor in the spin transfer torque (STT) branch. To further reduce WER, a better structure with diodes of bit cell is proposed. [ABSTRACT FROM AUTHOR]

As the speed gap of the modern processor and the off-chip main memory enlarges, on-chip cache capacity increases to sustain the performance scaling. As a result, the cache power occupies a large portion of the total power budget. Spin transfer torque magnetic memory (STT-MRAM) is proposed as a promising solution for the low power cache design due to its high integration density and ultralow leakage power. Nevertheless, the high write power and latency of STT-MRAM become new barriers for the commercialization of this emerging technology. In this paper, we investigate the thermal effect on the access performance of STT-MRAM, and observe that the temperature can affect the write delay and energy significantly. Then, we explore the nonuniform cache access (NUCA) design of the chip-multiprocessors with STT-MRAM-based last level cache (LLC). A thermal aware data migration policy, called “Thermosiphon,” which takes advantage of the thermal property of STT-MRAM, is proposed to reduce the LLC write energy. This policy splits the LLC into different regions dynamically based on the thermal distribution monitored by thermal sensors available on-chip, and adaptively migrates write intensive data among different thermal regions considering the thermal gradient. Compared to the conventional NUCA design, our proposed design can save 41.2% write energy at most and 13.01% on average with negligible hardware overhead. [ABSTRACT FROM AUTHOR]