Cache-Fair Thread Scheduling for Multicore Processors

Abstract

We present a new operating system scheduling algorithm for multicore processors. Our algorithm reduces the effects of unequal CPU cache sharing that occur on these processors and cause unfair CPU sharing, priority inversion, and inadequate CPU accounting. We describe the implementation of our algorithm in the Solaris operating system and demonstrate that it produces fairer schedules enabling better priority enforcement and improved performance stability for applications. With conventional scheduling algorithms, application performance on multicore processors varies by up to 36% depending on the runtime characteristics of concurrent processes. We reduce this variability by up to a factor of seven.