Implementing and Optimizing OpenACC on GPUs

Prof. Barbara Chapman and her team have considerable experience with parallel programming interfaces, compiler and runtime technology. On March 30th, three senior PhD students from Prof. Barbara Chapman's team will visit IACS at Stony Brook University to present their research results. This will include an introduction to GPU architecture, and presentations regarding (1) how OpenACC is implemented in OpenUH compiler, (2) how a new auto-tuning technique is able to better explore the optimization space, and (3) a use case describing effective solutions to port sparse FFT to three parallel architectures including multicore CPUs, GPGPUs and heterogeneous multicore embedded systems.