SCHEDULE: NOV 16-22, 2013
When viewing the Technical Program schedule, on the far righthand side is a column labeled "PLANNER." Use this planner to build your own schedule. Once you select an event and want to add it to your personal schedule, just click on the calendar icon of your choice (outlook calendar, ical calendar or google calendar) and that event will be stored there. As you select events in this manner, you will have your own schedule to guide you through the week.
Enabling Efficient Data Movement in MPI and PGAS Models on Heterogeneous Clusters with High Performance Interconnects
SESSION: Doctoral Showcase - Dissertation Research
EVENT TYPE: Doctoral Showcase - Dissertation Research Showcase
TIME: 11:30AM - 11:45AM
SESSION CHAIR: Alistair Rendell
Presenter(s):Sreeram Potluri
ROOM:601/603
ABSTRACT:
Modern clusters with accelerators (GPUs) and coprocessors (Xeon Phis), while providing high compute density and high performance per watt, introduce heterogeneity and require developers to use a hierarchy of programming models and paradigms to design applications. This has increased the complexity and cost of data movement on these clusters. We propose MVAPICH2-GPU, a model to use MPI for communication from CPU and GPU memories, in a unified manner. We also extend the OpenSHMEM PGAS model to support such unified communication. These models considerably simplify data movement in MPI and OpenSHMEM applications running on GPU clusters. We also propose designs in MPI and OpenSHMEM runtimes to transparently optimize data movement on GPU clusters, using state-of-the-art GPU technologies such as CUDA IPC and GPUDirect RDMA. Further, we present an optimized MPI runtime for communication on clusters with Intel Xeon Phi coprocessors. We evaluate our designs using micro-benchmarks, application kernels and end-applications.
Chair/Presenter Details:
Alistair Rendell (Chair) - Australian National University
Sreeram Potluri - Ohio State University
Click here to download .ics calendar file
