基于堆排序的片上Nor Flash损耗均衡硬件快速实现

发布时间：2018-01-25 07:32

本文关键词： Nor Flash 损耗均衡硬件实现堆排序 Verilog　出处：《计算机工程与科学》2017年11期 　论文类型：期刊论文

【摘要】：传统的Flash损耗均衡的实现大部分都是针对Nand Flash,且基本都是通过文件系统进行的。针对Nor Flash损耗均衡的研究较少,而且对于无法植入操作系统或者植入代价过大的情况,文件系统的方法将不再适用。在硬件层面实现了Flash损耗均衡算法,以满足上述应用条件,同时大大减少了芯片软件层面的开销。通过Verilog语言实现了损耗均衡、地址映射、垃圾回收、Flash接口等四个模块,每次写操作到来时通过堆排序寻找擦写次数最小的块,将其物理地址与虚拟地址对应并更新到地址映射表,并在垃圾块达到一定数量时进行回收操作,从而实现Flash的损耗均衡。实验结果表明,硬件实现的损耗均衡算法在初始化、堆删除、读操作的时间消耗上分别比软件算法最多快了14、16.4、17.8倍,获得了较大的速度提升。
[Abstract]:The traditional realization of Flash loss equalization is mostly aimed at Nand Flash, and it is basically carried out through file system. There are few researches on Nor Flash loss equalization. In addition, the method of file system is no longer suitable for the cases where the operating system cannot be implanted or the cost is too high. The Flash loss equalization algorithm is implemented on the hardware level to meet the above application conditions. At the same time, the cost of chip software level is greatly reduced. Through Verilog language, four modules are realized: loss balance, address mapping, garbage collection and flash interface. When the write operation arrives, the smallest block is searched by heap sorting, and the physical address is corresponding to the virtual address and updated to the address mapping table, and the garbage block is collected when a certain number of garbage blocks reach a certain number. Experimental results show that the loss equalization algorithm implemented by hardware is 14.4 16.4 faster than the software algorithm in initialization, heap deletion and read operation. 17.8 times, obtained the bigger speed rise.
【作者单位】：浙江大学信息与电子工程学院;杭州朔天科技有限公司;
【基金】：国家863计划(2012AA041701)
【分类号】：TP333
【正文快照】： 1引言得益于所具有的大容量、高速度、低功耗、非易失性等优点,闪存(Flash)正在人们的日常生活中占据着越来越多、越来越重要的位置。闪存通常可分为Nand Flash和Nor Flash两类,其中NandFlash一般用作大容量高密度的数据存储,而NorFlash一般用作程序存储。由于Flash在更新数据

【相似文献】