Abstract

This paper proposes a two-level exclusive cache model which does not shuffle blocks between the levels. The two- level cache system is visualized to be either in logical level 1 or level 2. An algorithm that classifies at each reference the physical cache levels into one of logical level 1 or logical level 2 cache is developed. Expressions for the average memory access time of the proposed model are derived and conditions for it to outperform exclusive cache proposed earlier in literature are derived. The model has scalability. Simulations on SPEC2000 show an improvement in average memory access time for two configurations of L1-L2 namely direct-direct with 67%, set associative-set associative with 64%. A degradation of 1% in average memory access time in direct-set associative configuration in L1-L2 is observed.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call