To consolidate the lecture material, each day's approximately 4 hours of lecture are complemented by 3 hours of hands-on sessions.

Course participants should have basic UNIX/Linux knowledge (login with secure shell, shell commands, basic programming, vi or emacs editors).

LRZ Building, University campus Garching, near Munich.

Day 1

• Introduction to High Performance Computing (Weinberg)
• Secure shell (Brietzke)
• Source code versioning with SVN (Guillen)
• Handling console and GUI based interfaces (Weinberg)
• Building programs with GNU MAKE (Guillen)

Day 2

• Basic parallel programming models: elements of MPI and OpenMP (Weinberg)
• Processor architectures (Hager)

Day 3

• Principles of code optimization: unrolling, blocking, dependencies, C++ issues, bandwidth issues, performance projections (Hager)
• Basics of software engineering (Guillen)
• Advanced MPI programming (Wittmann)

Day 4

• Advanced OpenMP programming (Weinberg)
• Performance Libraries (Weinberg)
• Parallel architectures: multi-core, multi-socket, ccNUMA, cache coherence and affinity, tools for handling memory affinity (Treibig)
• Parallel algorithms: data parallelism, domain decomposition, task parallelism, master-worker, granularity, load balancing, scalability models (Treibig)

Day 5

• Processor-specific optimization strategies: compiler switches, avoiding cache thrashing, exploiting SIMD capabilities (Treibig)

Each day is comprised of approximately 5 hours of lectures and 2 hours of hands-on sessions.

 Day 1

• Intel Tracing Tools: MPI tracing and correctness checking
• Intel Threading tools for OpenMP correctness checking and profiling
• Profiling on SGI Altix systems: histx and lipfpm
• Introduction to Scalasca

 Day 2

• Parallel application performance analysis with Scalasca
• Parallel input/output with MPI-IO

 Day 3

• I/O tuning on high performance file systems
• Portability of binary files, big/little endian issues
• Using I/O libraries (pNetCDF, HDF5)

Day 4

• Introduction to the PGAS languages Coarray Fortran and UPC: Partitioned global address space languages have emerged as an alternative to other parallel programming models, promising a shorter development cycle due to improved programmability while keeping the performance level on par with MPI. This course introduces the parallel  facilities integrated into the Fortran language (coarrays) and the C language (unified parallel C), respectively. A hands-on session allows to experiment with the new concepts, using prototype implementations on LRZ's HPC systems.

• Introduction to Eclipse IDE
• Introduction to CDT
• Hands-on with CDT
• Short introduction and demo of Photran

Please register via the LRZ registration form (registration is not yet open)

This four-day workshop gives an introduction to the usage of the new Petaflop class Supercomputer at LRZ, SuperMUC. The first three days of this are dedicated to presentations by Intel on their software development stack (compilers, tools and libraries); the remaining day will be comprised of talks and exercises delivered by IBM and LRZ on usage of the IBM-specific aspects of the new system (IBM MPI, LoadLeveler, HPC Toolkit) and recommendations on tuning and optimizing for the new system.

The following schedule is preliminary and may undergo changes.

• Introduction to the x86 Westmere and Sandy Bridge Processor Architecture
• Usage of Intel Compilers: switches, optimization, profiling, OpenMP, advanced features
• Usage of the Intel debugger
• Explicit and compiler-directed vectorization with SSE and AVX

 Tuesday schedule:

• Programming for NUMA
• Intel Cilk Plus
• Fortran standard support and extensions in Intel Fortran
• Intel Inspector XE (correctness checking for memory and threading)
• Threading Building Blocks, OpenCL and other parallel programming models
• Static Security Analysis
• Performance Libraries (MKL, IPP)
• VTune Amplifier XE (performance analysis)

 Wednesday schedule:

• Intel Performance Monitoring Unit (PMU) and other performance analysis tools
• Intel Cluster Tools Overview and Usage
• Intel Many Integrated Core (MIC) Architecture
• A case study illustrating the use of the Intel toolchain

Thursday schedule:

• Using LoadLeveler for batch queuing
• IBM's parallel environment
• Performance Tools (High Performance Toolkit, Eclipse User Interface) including a demonstration.

Different modern Krylov Subspace Methods (CG, GMRES, BiCGSTAB ...) as well as highly efficient preconditioning techniques are presented in the context of real life applications.

Hands-on sessions (in C and Fortran) will allow users to immediately test and understand the

• basic constructs of iterative solvers
• the Message Passing Interface (MPI)
• the shared memory directives of OpenMP.

This course is organized by the University of Kassel, the high performance computing centre of Stuttgart (HLRS) and IAG.

Members of universities and public research institutes outside Germany: 120 EUR
others: 400 EUR
(includes handouts and drinks at coffee breaks, will be collected on the first day of the course, cash only)