Centre For Development of Advanced Computing
&
Centre for Modelling Simulation and Design (CMSD)University of Hyderabad (UoH)
jointly conducted
five days technology workshop
On
Heterogeneous Computing
CPU/GPU HPC Cluster - Algorithms & Performance of Application Kernels
Heterogeneous Computing
CPU/GPU HPC Cluster - Algorithms & Performance of Application Kernels
(Initiatives on Power Efficiency - Green Computing)
(HeGaPa – 2012)” which is scheduled from
July 16-20, 2012 at CMSD, UoH
The workshop was aimed at understanding performance issues of various application kernels
on HPC (High Performance Computing) GPU Cluster based on Multi-Core Processors & GPU Accelerators and
AMD APUs.
Important links
|
HeGaPa-2012 Tech. Prog.:
|
Day 1 :
|
HPC Custer with GPU Accelerators
AMD Prog. on Host-CPUs (MPI, Pthreads, OpenMP);
AMD APU APP
HP AMD A8-4500K (Trinity) - OpenCL;
Heterogeneous Computing - OpenCL - Tuning &
Performance Issues.
|
Day 2 :
|
CUDA - NVIDIA GPUs; CUDA SDK/APIs;
CUDA –Tuning & Performance;
CUDA Toolkit; NVIDIA -
PGI Accelerator Compilers -
OpenACC
Directives;
|
Day 3 :
|
HPC GPU Cluster CUDA - GPUs; CUDA 5.0 &
PGI
Accelerator Compilers with
OpenACC
Directives - Matrix Comptuations - Applications
Streams Concurrent Asynchronization Execution
|
Day 4 :
|
GPU Cluster - AMD APP SDK
AMD APU
HP AMD A8-4500K (Trinity)
OpenCL Prog.; Perf. of
Numerical Linear Algebra Kernels on AMD APUs /GPU Cluster.
|
Day 5 :
|
Energy Efficiency (Day 5:) Computing : Performance & Power Efficiency;
Power Issues in HPC Systems; List of Supercomputers : Graph-500 Benchmarks; Measurements - flop per watt
|
Apps :
(Day 2-5)
|
Applications Kernels : Selective Application kernels on HPC GPU Cluster
include the following : Image Processing; Dense & Sparse Matrix Comps;
Solution of Partial differential Eqs. – FDM-FEM; FFT Lib.; Computational
Physics – Monte Carlo Methods ; Bio-Informatics Apps.;
Invited lectures on Information Sciences
|
|
|
|
For more information, Contact Email: hpcfte@cdac.in
Workshop Co-ordinator :
vcvrao@cdac.in