Performance analysis of massively parallel programs for graphics processing units

Any modern Graphics Processing Unit (graphics card) is a good platform to run massively parallel programs. Still, we lack tools observe and measure performance characteristics of GPU-based software. We state that due complex memory hierarchy thou- sands execution threads the all issues are about efficient use graphics card hierarchy. propose GPGPUSim simulator, previously used mostly for architecture validation, validation CUDA-based program. provide examples which show how simulation analysis