Towards a Better Lifetime for Non-volatile Caches in Chip Multiprocessors

To address the limitations imposed by the conventional SRAM cache, the emerging Non-Volatile Memory (NVM) technologies are widely used nowadays as they provide among other things high density and low leakage. However, the weak endurance of these memory technologies limits their use as a replacement of SRAM. Further, the large intra-set write variation introduced by the application and the existing cache management policies significantly reduces the lifetime of the NVM caches. This paper proposes a technique to reduce intra-set write variation to increase the effective lifetime of the NVM caches. This is achieved by partitioning the cache into equal size windows and using different windows during the execution in order to distribute the writes evenly across the set. Full system simulation results show that our policy reduces the write variation and improves the cache lifetime by 2.1 times and 1.8 times for a single and dual core system, respectively compared to an existing policy.

[1]  Dong Li,et al.  WriteSmoothing: improving lifetime of non-volatile caches using intra-set wear-leveling , 2014, GLSVLSI '14.

[2]  Somayeh Sardashti,et al.  The gem5 simulator , 2011, CARN.

[3]  Kinam Kim,et al.  Bi-layered RRAM with unlimited endurance and extremely uniform switching , 2011, 2011 Symposium on VLSI Technology - Digest of Technical Papers.

[4]  Yiming Huai,et al.  Spin-Transfer Torque MRAM (STT-MRAM): Challenges and Prospects , 2008 .

[5]  Dong Li,et al.  LastingNVCache: A Technique for Improving the Lifetime of Non-volatile Caches , 2014, 2014 IEEE Computer Society Annual Symposium on VLSI.

[6]  Xueti Tang,et al.  Spin-transfer torque magnetic random access memory (STT-MRAM) , 2013, JETC.

[7]  Yiran Chen,et al.  On-chip caches built on multilevel spin-transfer torque RAM cells and its optimizations , 2013, JETC.

[8]  Kai Li,et al.  The PARSEC benchmark suite: Characterization and architectural implications , 2008, 2008 International Conference on Parallel Architectures and Compilation Techniques (PACT).

[9]  Cong Xu,et al.  NVSim: A Circuit-Level Performance, Energy, and Area Model for Emerging Nonvolatile Memory , 2012, IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems.

[10]  Yuan Xie,et al.  i2WAP: Improving non-volatile cache lifetime by reducing inter- and intra-set write variations , 2013, 2013 IEEE 19th International Symposium on High Performance Computer Architecture (HPCA).

[11]  Hemangee K. Kapoor,et al.  Towards a dynamic associativity enabled write prediction based hybrid cache , 2016, 2016 20th International Symposium on VLSI Design and Test (VDAT).

[12]  Sudhanva Gurumurthi,et al.  Phase Change Memory: From Devices to Systems , 2011, Phase Change Memory.

[13]  Hemangee K. Kapoor,et al.  Restricting writes for energy-efficient hybrid cache in multi-core architectures , 2016, 2016 IFIP/IEEE International Conference on Very Large Scale Integration (VLSI-SoC).