Register to CSCS-USI Summer School on parallel programming MPI, OpenMP, CUDA and OpenACC
CSCS and USI organize a 10-day Summer School focused on parallel programming MPI, OpenMP, CUDA and
OpenACC aimed at students who are new to the world of high performance computing and hybrid systems and
who wish to learn the basic skills required to write, develop and maintain parallel applications in scientific computing.
The purpose of the summer school is to teach programming skills and therefore a large proportion of the course will
be dedicated to practical exercises.
CSCS-USI Summer School 2014 – 30 June – 10 July, 2014
Potential participants are advised to apply by filling the application form and send it in pdf format to themis.athanassiadou(at)cscs.ch.
Deadline for sending your application is April 30th, 2014
Acceptance notification: May 9th, 2014
Registration will be opened immediately on May 10th, 2014
Registration fees (include conference fee and accommodation):
- CHF 1000.00 for graduate students
- CHF 2000.00 for participants from industry
Tentative Agenda
| Week 1 (5 days) | Week 2 (4 days) |
|---|---|
| General:– Parallel Architectures and Programming Models- HPC Computing Keynote
CSCS |
General:– Overview of GPU architecture- Overview of GPU programming languages
Keynote |
| MPI – 1:– MPI Overview- MPI Process Model
– Point-to-Point Communication – Non-Blocking Communication – Derived Datatypes – Virtual Topologies Collective Communication |
CUDA:– program structure / kernels- memory and data transfer
– threads and warps – tiling – Performance Considerations MPI CUDA
|
| OpenMP:– Overview and execution model- Variables + Scope
– Work sharing directives – Reductions – Race conditions – Pitfalls /caveats |
OpenACC:– Overview and execution model- Directives
– Reveal tutorial – Clauses – Gangs/workers – CUDA interoperability Performance Considerations |
| MPI – 2, -3:– One sided- communicationMPI I/O | OpenCL– Execution modelComparison with CUDA/OpenACC |