Hybrid memory data management method, device and storage medium

Abstract

The invention discloses a method for managing mixed memory data, which includes: according to the data page access type, allocating the access data to a corresponding memory; wherein, the memory includes a first memory and a second memory; when the memory is When a data page is selected, the access record information of the data page is updated to maintain the access time locality information and read-write information of all data pages; according to the memory space of the first memory and the read-write information, and historical access read and write information of the second memory, and perform data page migration between the first memory and the second memory. The invention can effectively improve the migration efficiency of mixed memory data pages and improve the performance of the computer system.

Description

technical field [0001] The invention relates to the technical field of computer memory management, in particular to a method, device and storage medium for managing mixed memory data. Background technique [0002] Dynamic random access memory (Dynamic Random Access Memory, DRAM) has been widely used in computer systems due to its good read and write performance in the past few decades. With the development of computer technology and the advent of the era of big data, problems such as "storage wall" and "energy consumption wall" have emerged due to the low storage density of DRAM and high refresh energy consumption, which has led to computer systems' capacity, performance, and functionality of traditional storage systems. Power consumption and scalability pose serious challenges. Non-volatile memory (Non-VolatileMemory, NVM) has the advantages of non-volatility, byte addressability, zero leakage power consumption, high scalability, etc., and is considered as a potential obje...

AI Technical Summary

Problems solved by technology

[0005] The existing statistical method of data popularity is accompanied by a large amount of data calculation and storage overhead. Each data allocation and migration must be based on the new statistical popularity, which greatly increases computing energy consumption and access delay; and, the existing migration The mechanism will generate a large number of invalid page migrations, so that the migration operation does not really bring about performance improvement, that is, the performance benefit of some migrated pages of NVM is less than the performance loss when not migrating, but because the migration operation requires the system to interrupt the copy, etc. operation, reducing system performance

Images