hyPACK-2013 HPC Cluster (Hybrid Computing - Heterogeneous Prog.)

Stand-alone Multi-core Processor systems are interconnected with appropriate high-speed network and their combined computational power can be applied to solve a variety of computationally intensive applications. Low-latency, high-speed networks (Infiniband, Myrinet, Gigabit, C-DAC PARAMNet) can be integrated with massively parallel processors to build sclable HPC Clusters. InfiniBand is emerging as a high-performance interconnects delivering low latency and high bandwidth. Several experiments have been carried out to use InfiniBand as an open standard to deliver performance and scalability to MPI applications. To-day, the there is a growing demand to use HPC Clusters with different accelerators such as RC-FPGAs (C-DAC), Intel many-core CPUs (x86 based Many-Core & Intel MIC), GPUs (CUDA enabled NVIDIA GPUs, AMD-ATI GPUs) in an HPC GPU Cluster environment to solve complex applications. The links for HPC Accelerators are given below.

These hybrid HPC clusters with different accelerators can address some of the heterogeneous computing workloads. These hybrid systems can be made "adaptive" to the application it is running, assigning the most effective resources in real-time as per application demands, without requiring modifications to the application. The hybrid computing system supports Parallel Programming models, which include Shared memory programming (POSIX Threads, OpenMP, Intel TBB), and MPI 2.0 standard on Multi Core Processors. The Linux programming environment is provided on Cluster and the operating environment can be designed to run large complex application that can make use of RC-FPGAs, Intel MIC, GPU accelerators attached to Multi-Core Processor systems in an efficient way. The inter-node communication and inter-GPU communication in typical HPC GPU cluster takes place via host node. GPU and the controlling CPU thread on host-cpu communicate via memcopies, while CPU threads exchange data using the same methods as applications not accelerated with GPUs.

• Matrix Computations : Matrix - Vector Multiplication, Matrix-Matrix Multiplication based on MPI and OpenCL/CUDA Implementation on HPC GPU Cluster

• Application Kernels demonstration on HPC GPU Clusters (CUDA Prog & Intel TBB)

• Performance of Matrix Computations on HPC GPU Cluster using vendor supplied tuned mathematical libraries (CUBLAS, MAGMA on NVIDIA GPUs) & BLAS on AMD-ATI Fire Stream GPU Accelerators or AMD Trinity APUs )

• Selective Numerical Computational kernels on Parallel Processing Systems with GPU Accelerator devices using MPI & CUDA enabled NVIDIA GPUs & OpenCL AMD-ATI GPUs or AMD Trinity APUs on HPC GPU Cluster

• Special Class of Application Kernels, and Numerical Linear algebra on Multi-Core Processors using Mixed Mode of Programming ( TBB-CUDA, MPI-CUDA, Pthreads-CUDA) on HPC GPU Cluster.

• Special Class of Application Kernels, and Numerical Linear algebra on Multi-Core Processors using Heterogeneous Programming ( OpenMP-OpenCL, MPI-OpenCL, Pthreads-OpenCL) on HPC GPU Cluster.

• HPC-GPU Cluster (MPI on host-CPU & GPU - OpenCL - Solution of Partial differential Equations by Finite Difference Methods & Finite Element Methods

• HPC GPU Cluster (MPI on host-CPU & GPU - OpenCL - Image Processing (Edge Detection , Face Detection & Image Inpainting) algorithms

• Heterogeneous Programming (MPI on host-CPU & GPU - OpenCL - String Search algorithms & Sequence Analysis Applications

• HPC GPU Cluster (MPI on host-CPU & GPU-OpenCL - Open source software Benchmarks - Solution of Matrix system Ax=b of Linear Equations (MAGMA on CUDA enabled GPUs & LINPACK solvers)