Networked Power-Gated MRAMs for Memory-Based Computing
Networked Power-Gated MRAMs for Memory-Based Computing
复制标题
DOI：
10.1109/tvlsi.2018.2856458
复制
发表时间：
2018-08
影响因子：
2.8
通讯作者：
J. Diguet;N. Onizawa;Mostafa Rizk;J. Sepúlveda;A. Baghdadi;T. Hanyu
中科院分区：
文献类型：
--
作者：
J. Diguet;N. Onizawa;Mostafa Rizk;J. Sepúlveda;A. Baghdadi;T. Hanyu
新兴的非易失性存储器技术为原始计算架构打开了新的前景。在本文中，我们提出了一种新型的灵活和节能的架构，依赖于电源门控分布式磁阻随机存取存储器（MRAM）。所提出的架构使用片上网络（NoC）互连基于MRAM的集群，处理元件和管理器。NoC通过分组的方式将应用特定的命令分发到MRAM设备。可配置的网络接口允许将MRAM设备转换为能够响应传入命令的智能单元。在这种情况下，提出了三种类型的MRAM设计，具有不同的功率门控策略和粒度。一个相关的数据库搜索引擎的案例研究被认为是说明这个建议的架构的好处。它是用稀疏神经网络方法实现的，并在SystemC中模拟了不同的场景，包括数百个数据库查询。硬件设计和准确的功率估计已经进行。所得到的结果表明，重要的数据库命中率约94%的功耗降低。与基于静态随机存取存储器的实现相比，针对65纳米技术的节能达到87%。此外，一种新的非对称读/写MRAM类型提供了从39%到50%的能量减少相对于其他固定粒度的模型。这导致了基于存储器的计算的低功耗、高度可扩展和可配置的实现。
Emerging nonvolatile memory technologies open new perspectives for original computing architectures. In this paper, we propose a new type of flexible and energy-efficient architecture that relies on power-gated distributed magnetoresistive random access memory (MRAM). The proposed architecture uses a network-on-chip (NoC) to interconnect MRAM-based clusters, processing elements, and managers. The NoC distributes application-specific commands to MRAM devices by means of packets. Configurable network interfaces allow to transform MRAM devices into smart units able to respond to incoming commands. In this context, three types of MRAM designs are proposed with different power-gating policies and granularities. A relevant database search engine case study is considered to illustrate the benefits of this proposed architecture. It is implemented with a sparse-neural-network approach and simulated in SystemC with different scenarios including hundreds of database queries. Hardware designs and accurate power estimations have been conducted. The obtained results demonstrate important power reduction with database hit rates of about 94%. Targeting 65-nm technology, energy savings reach 87% when compared with an static random access memory-based implementation. Moreover, a new asymmetric read/write MRAM type provides from 39% to 50% energy reduction with respect to the other fixed-granularity models. This results in a low-power, highly scalable, and configurable implementation of memory-based computing.