Sampling-Based Phase Classification and Prediction for Multi-threaded Program Execution on Multi-core Architectures
Performance characteristics of application programs on large-scale systems are often significantly different from those on smaller systems. SPEC OMP2001 is a benchmark suite intended for measuring performance of modern shared memory parallel systems. The first component of the suite, SPEC OMPM2001, is developed for medium-scale (4to 16-way) systems. We present our experiences on benchmark development in achieving good scalability using the OpenMP API. This paper then analyzes the published results of SPEC OMPM2001 on large systems (32-way and larger), based on application program behavior and systems’ architectural features. The ongoing development of the SPEC OMP2001 benchmark suites is also discussed. Its main feature is the increased data set for large-scale systems. We refer to this suite as SPEC OMPL2001, in contrast to the current SPEC OMPM2001 (medium data set) suite.