An Energy-Efficient 3D Stacked STT-RAM Cache Architecture for CMPs

Abstract

In this chapter, we introduce how to adopt spin-transfer torque random access memory (STT-RAM) as on-chip L2 caches to achieve better performance and lower energy consumption, compared to traditional L2 cache designs. STT-RAM is a promising memory technology for on-chip cache design because of its fast read access, high density, and non-volatility. Using 3D heterogeneous integrations, it becomes feasible and cost-efficient to stack STT-RAM atop conventional chip multiprocessors (CMPs). However, one disadvantage of STT-RAM is its long write latency and its high write energy. In this chapter, we first stack STT-RAM-based L2 caches directly atop CMPs and compare it against SRAM counterparts in terms of performance and energy. We observe that the direct STT-RAM stacking might harm the chip performance due to the aforementioned long write latency and high write energy. To solve this problem, we then propose two architectural techniques: read-preemptive write buffer and SRAM–STT-RAM hybrid L2 cache. The simulation result shows that our optimized STT-RAM L2 cache improves performance by 4.91 % and reduces power by 73.5 % compared to the conventional SRAM L2 cache with the similar area.