Droop Mitigating Last Level Cache Architecture for STTRAM
Radha Krishna Aluru1 and Swaroop Ghosh2
1Department of Computer Science and Engineering, University of South Florida, Tampa, FL, USA.
aluru@mail.usf.edu
2School of Electrical Engineering and Computer Science, Pennsylvania State University, University Park, PA-16801, USA.
szg212@psu.edu
ABSTRACT
Spin-Transfer Torque Random Access Memory (STTRAM) is one of the emerging Non-Volatile Memory (NVM) technologies especially preferred for the Last Level Cache (LLC). The amount of current needed to switch the magnetization is high (\sima100mA per bit). For a full cache line (512-bit) write, this extremely high current results in a voltage droop in the conventional cache architecture. Due to this droop, the write operation fails especially, when the farthest bank of the cache is accessed. In this paper, we propose a new cache architecture to mitigate this problem of droop and make the write operation successful. Instead of continuously writing the entire cache line (512-bit) in a single bank, the proposed architecture writes 64-bits in multiple physically separated locations across the cache. The simulation results obtained (both circuit and micro-architectural) comparing our proposed architecture against the conventional are found to be 1.96% (IPC) and 5.21% (energy).
Keywords: Droop, LLC, STTRAM, Bank, Latency, Energy, SPLASH benchmarks.
