P-ray: A suite of micro-benchmarks for multi-core architectures

Abstract. The increasing complexity of computer architectures has made the approach of automatically generating code that is optimized for the target machine a growing area of interest. Examples of such systems are library generators, such as ATLAS, SPIRAL, or FFTW. To generate op-timized code without manual intervention, these systems need to know the values of certain hardware parameters, such as the cache size or the number of registers. Current benchmark suites such as X-Ray or LM-bench can automatically determine some of these parameters for single processor super-scalar machines but cannot determine multi-core specific characteristics. In this paper, we present P-Ray, a suite of micro-benchmarks that fo-cus on hardware characteristics specific to multi-core architectures. Such characteristics include the number of cores that share the L2 cache, the different processors ’ interconnection topologies, and the bandwidth-to-memory for multi-cores. Our experiments show that, for several different architectures tested (desktop and server), P-Ray generates accurate re-sults.