Coarse-grained last-level caching bypass method

Abstract

The invention relates to a coarse-grained last-level caching bypass method, which is named Global Bypass, and views the bypass as a policy choice equivalent to replacement instead of supplementation. The method is switched between a bypass mode and a replacement mode; in the bypass mode, all failures are bypast, and the access characteristic of each individual is not evaluated; and in the replacement mode, the Global Bypass uses the basic replacement algorithm of the last-level caching to replace, the basic replacement algorithm can be any existing replacement algorithm, and because the access characteristic of each individual is not needed to be evaluated, the realizing cost of the Global Bypass is low.

Description

technical field [0001] The invention relates to the management strategy of the last-level high-speed cache in the processor, and mainly uses a coarse-grained bypass strategy Global Bypass to improve the performance and energy efficiency of the system. Background technique [0002] Power consumption has now become a key limiting factor for scaling single-core and multi-core processors, and the last-level cache has been proven to be energy efficient. So there will be a larger last-level cache in future processors. The last-level cache management strategy is critical to performance. In the last-level cache, because the reuse distance of a large number of cache blocks is greater than the capacity of the cache, the performance of the commonly used LRU and its approximation algorithm is not good. There have been many previous works trying to capture the access of these long-distance reused blocks, which can be mainly divided into four types: replacement algorithm, dead block pre...

AI Technical Summary

Problems solved by technology

[0007] These methods either have limited performance improvements, or require a large amount of storage overhead and hardware changes

Images