The threadblock size and shape choice is one of the most important user decisions when a parallel problem is coded to run in GPU architectures. In fact, threadblock configuration has a significant impact on the global performance of the program. Unfortunately , the programmer has not enough information about the subtle interactions between this choice of parameters and the underlying hardware. This paper presents uBench, a suite of micro-benchmarks, in order to explore the impact on performance derived from the combination of (1) the threadblock size and shape choice criteria, and (2) the GPU hardware resources and configurations. Each micro-benchmark has been designed as simple as possible to focus on a single effect derived from the hardware or threadblock parameter choice. As an example of the capabilities of this benchmark suite, this paper shows an experimental evaluation of the Fermi architecture, in terms of configuration parameters. This study confirms some previous experimental results and gives new insights on the influence of these parameters on the performance delivered by this GPU architecture.
