Intermediate-par-cluster

Week long workshop covering intermediate topics in parallel programming and cluster computing.

Target Audience

Teach HPC, build HPC, and/or do HPC or both? This workshop more mechanics (teach HPC, build HPC), HPC Applications workshop more drivers (do HPC).
NCSA ACCESS Center audience is mostly undergraduate, August OU audience is mostly faculty

Pre-reqs

C/C++/FORTRAN experience - submit a copy of the biggest/complex parallel program you have written/maintained with your application to the workshop.
Linux experience
Viewing SIPE in some form
MPI send and receive (at least point-to-point)

Notes

Focus on lab exercises as a mechanism for learning, well tested, scaffolded and polished. More of them than at the typical intro workshop.
Core topics and parallel sessions ala intro workshops.
Project oriented ala intro workshops?

Memory Hierarchy (review and more depth)
- Shared memory architectures
- Distributed memory architectures
Collective operations (review and more depth)
Debugging (review and more depth)
Overlapping communication and computation
- Using I MPI routines
- Memory access vs disk access
Benchmarking, Profiling and Tuning your Code
- MPI routines
- gprof
- PAPI
GPGPU
- First look, place in context
- Architecture, memory and compute
- NAMD lab
Petascale computing
- Multi-core/Many-core
- Hybrid models
  - MPI and OpenMP
  - MPI and CUDA