hyPACK-2013 Proceedings

hyPACK-2013 four-days technology workshop CD soft-copy Proceedings contains different suite of programs for Scientific and Engineering Computations on Multi-Core processor Systems, Computing Systems with Intel Xeon Phi Co-processors & GPGPU Accelerator Devices, and High-Performance GPU Cluster. By understanding the presentation material covered and the programs in the CD soft-copy of Hands-on as building blocks, scientists and engineers can piece together more complicated software tools that are tailored specifically for their needs, on Multi-Core processor Systems and GPUs. The hypack-2013 hands-on session programs provide foundation for application user to design good parallel algorithms and write parallel programs on Multi-Core processors, GPGPUs, HPC GPU Cluster, Distributed Shared Memory Systems in-order to extract performance of large-scale applications.

The hyPACK-2013 technology workshop CD soft-copy proceedings does offer the application users a great opportunity to learn about the fundamentals of writing parallel programs using different programming paradigms (MPI, OpenMP, Intel TBB, MKL on Xeon Phi, POSIX Threads, Cilk Plus, OpeMP 4.0, Compiler Offload pragmas & Vectorization, CUDA/OpenCL enabled NVIDIA GPUs-CUDA, NVIDIA - OpenACC, NVIDIA GPUs-OpenCL, AMD-APP-OpenCL, emphasizing on optimization techniques to extract the performance on emerging parallel processing platforms with GPU Accelerators.

The workshop will cover an overview of Intel Xeon Co-processosrs, GPU Computing - CUDA Programming Software toolkit with Hands-on Session, NVIDIA-CUDA with OpenACC, an overview of Open Computing Language (OpenCL), an Overview of Heterogeneous computing - OpenCL, GPGPUs - AMD-APP (SDK) with Hands-on Session, Programming on HPC GPU Cluster based on CUDA / OpenCL and as well as demonstration of software on GPUs by Industry experts. A special session is arranged on Power efficiency issues of emerging parallel processing platforms on last day of the workshop. Participants will use Multi-Core Processors, and Systems with GPGPUs / GPU accelerator devices and Hybrid Heterogeneous HPC GPU Cluster.

In hyPACK-2013 Hands-on Sessions, participants will get an opportunity to write codes for Numerical Linear algebra (NLA) based on GPU programming - CUDA & OpenCL. Also, programs based on Hybrid Computing (Mixed programming) will be discussed. On second half of each day, participants will get an opportunity to walk-through and execute some of the programs specifically designed for this workshop. Example programs on GPGPUs / GPU computing (CUDA enabled NVIDIA GPUs; OpenACC-NVIDIA CUDA-PGI Compilers, AMD-APP SDK- OpenCL ) based on Numerical computations for selective class of scientific & engineering applications are provided. An overview of software tools, tuning & performance issues on Systems with GPUs will be discussed. Suite of Optimised programs on GPGPU systems and Heterogeneous HPC GPU Cluster will be made available. This session includes class-room lectures, key-note talks on GPGPUs / GPU computing and demonstration of Software from application point of view. Demonstration of Application kernels such as Solution of Partial differential Equations, Image Processing - Edge Detection, & String Search Algorithms will be demonstrated.