DRBAC: Dynamic Row Buffer Access Control for Power and Performance of DRAM Systems

Performance of dynamic random access memory (DRAM) has been steadily improved to overcome the concern that the DRAM access time may become the performance bottleneck of a system. Besides, DRAM power consumption has become a critical issue in mobile and server systems. The open page policy is widely used to minimize the memory access latency and the power consumption of the activate and the precharge commands. In this paper, we analyze DRAM power and performance according to memory request characteristics of applications. Especially, we observe that the row buffer access control influences the overall performance and power consumption. Further, the power-delay product (PDP) is sensitive to the row buffer hit ratio and the memory request frequency. Thus, we propose a method called dynamic row buffer access control (DRBAC) that changes the row buffer access limit dynamically based on the memory request characteristics. From simulation results, it is verified that DRBAC reduces the PDP value by up to 17.8% compared to the conventional method for various benchmarks. Therefore, we conclude that the proposed DRBAC is very effective for low power and high performance DRAM systems.

[1]  William J. Dally,et al.  Memory access scheduling , 2000, Proceedings of 27th International Symposium on Computer Architecture (IEEE Cat. No.RS00201).

[2]  Dhiraj K. Pradhan,et al.  Formal Model for the Reduction of the Dynamic Energy Consumption in Multi-Layer Memory Subsystems ∗ , 2008 .

[3]  Masaki Kondo,et al.  Reducing Memory System Energy by Software-Controlled On-Chip Memory , 2003 .

[4]  Bruce Jacob,et al.  DRAMSim2: A Cycle Accurate Memory System Simulator , 2011, IEEE Computer Architecture Letters.

[5]  Karthick Rajamani,et al.  Energy Management for Commercial Servers , 2003, Computer.

[6]  Xu Cheng,et al.  Page policy control with memory partitioning for DRAM performance and power efficiency , 2013, International Symposium on Low Power Electronics and Design (ISLPED).

[7]  Ying Xu,et al.  Prediction in Dynamic SDRAM Controller Policies , 2009, SAMOS.

[8]  In-Cheol Park,et al.  History-based memory mode prediction for improving memory performance , 2002, Proceedings of the 2003 International Symposium on Circuits and Systems, 2003. ISCAS '03..

[9]  Todd M. Austin,et al.  SimpleScalar: An Infrastructure for Computer System Modeling , 2002, Computer.

[10]  Gernot Heiser,et al.  An Analysis of Power Consumption in a Smartphone , 2010, USENIX Annual Technical Conference.

[11]  David W. Nellans,et al.  Prediction Based DRAM Row-Buffer Management in the Many-Core Era , 2011, 2011 International Conference on Parallel Architectures and Compilation Techniques.

[12]  V.V. Stankovic,et al.  DRAM Controller with a Close-Page Predictor , 2005, EUROCON 2005 - The International Conference on "Computer as a Tool".