Cache replacement method and device thereof, and heterogeneous multi-core system

Abstract

The invention provides a cache replacement method, which is applied to a heterogeneous multi-core system. The method comprises the following steps: when it is determined that a first application program currently running is an application program running on a GPU; determining a running type of the first application according to a performance parameter of the first application; determining a cachepriority of first data accessed by the first application according to a running type of the first application, the cache priority of the first data including a priority other than a highest priority;determining a cache replacement algorithm according to the algorithm type identification of the record; caching the first data into a cache queue of the shared cache according to the determined cachereplacement algorithm and a cache priority of the first data. The cache replacement method can improve the utilization rate of cache while ensuring the performance of GPU, which is helpful to improvethe computing power and computing speed of heterogeneous multi-core system.

Description

technical field [0001] The invention relates to the technical field of data processing, in particular to a cache replacement method and device thereof, and a heterogeneous multi-core system. Background technique [0002] A heterogeneous multi-core system refers to the integration of multiple single-core or multi-core processors of different types on a chip, which has the characteristics of small size and high computing performance. [0003] In a heterogeneous multi-core system, a multi-level cache is usually used, each processor has its own independent private level-1 or multi-level cache Cache, and all processors have a shared cache (called a shared cache), and the shared cache is multiple Commonly used by processors, usually the last level of cache; under multi-level cache, when the processor processes data, it can first read data from the level-1 cache closest to the processor. When there is more data, it is read from the caches of the following levels in turn. When ther...

AI Technical Summary

Problems solved by technology

[0004] Under the multi-level cache architecture, although the capacity of the last-level shared cache is large, its capacity is also limited. Sharing the last-level cache by multiple different types of processors will cause competition for cache resources. Therefore, the utilization of cache Rate is an important factor affecting the computing power and computing speed of heterogeneous multi-core systems

Images