Method and apparatus for access control for quality of service optimization with aid of machine learning

Abstract

A method for performing access control regarding quality of service (QoS) optimization of a memory device with aid of machine learning an associated apparatus (e.g. the memory device and a controllerthereof) are provided. The method may include: performing background scan on the NV memory to collect valley information of voltage distribution of memory cells within the NV memory, and performing machine learning based on a reinforcement learning model according to the valley information, in order to prepare a plurality of tables through the machine learning based on the reinforcement learning model in advance, for use of reading data from the NV memory; during a first time interval, writing first data and read the first data using a first table within the plurality of tables; and during a second time interval, reading the first data using a second table within the plurality of tables. According to the method and the related equipment, the proper operation of the whole system is ensured,and the problems such as bit error rate increase, random reading delay increase and related problems are avoided.

Description

technical field [0001] The present invention relates to the access of flash memory, and in particular to a method for performing access control and related equipment (for example: all said memory device and its controller such as a memory controller in said memory device). Background technique [0002] Due to the continuous development of memory technology in recent years, various portable or non-portable memory devices (for example: memory cards that meet SD / MMC, CF, MS, XD and UFS standards respectively; and for example: solid state drives (solid state drives) state drive, SSD); another example: embedded (embedded) memory devices conforming to UFS and EMMC specifications respectively) are widely implemented in many applications. Therefore, access control of memories in these memory devices has become a very hot topic. [0003] For commonly used NAND flash memory, it mainly includes two types of flash memory: single level cell (SLC) and multiple level cell (MLC). Each tr...

AI Technical Summary

Problems solved by technology

In addition, a flash memory solution implemented by a three-dimensional (3D) NAND flash architecture has been proposed to achieve higher storage capacity, but some additional problems such as random write latency and random read latency have been derived.

Random read latency issues can lead to poor Quality of Service (QoS) in most cases since most user behavior can correspond to random reads

In addition, in the case of increased memory cells in the stack of 3D NAND flash architecture, a floating gate (FG) solution can be replaced by a charge trapping (charge -trap, a new solution for CT) materials, which may correspondingly cause data-retention problems, making the delay time of reading data increase, so the quality of service may become lower

In particular, the data retention issue may trigger a moving read mechanism in the management mechanism to perform one or more additional read operations, resulting in overall performance degradation

Images