Efficient HPC Communications: Profiling and Tuning MPI Applications

A cluster’s compute time is precious because its job queue is never ending. So when it’s your application’s turn on the cluster, you want to effectively use all of the compute resources while your applications runs.

In MPI applications, communication patterns and latencies are paramount to using the available compute resources. Thus you need to ask yourself the following question:

Is your MPI application performing optimally on your cluster?

Intel technical expert Dmitry Sivkov will help you uncover the answer, including:

  • Understanding MPI application behavior
  • Quickly finding application bottlenecks
  • Achieve high performance for parallel cluster applications

Get the software

  • Intel® VTune™ Amplifier (Available standalone, and also as part of Intel® Parallel Studio XE.)
  • Intel® Trace Analyzer and Collector (Available as part of Intel® Parallel Studio XE. Try it now for 30 days free.)
  • Intel® MPI Library (One of five free Intel® Performance Libraries.)
Dmitry Sivkov, Technical Consulting Engineer, Intel Corporation

Dmitry has 15 years’ experience in high performance computing (HPC) and data analysis. With Intel since 2011, he is responsible for helping customers successfully use Intel® Software Development Tools to build, debug and deploy edge-to-cloud applications and solutions.

Dmitry holds a PhD in Mathematics. He is based in Russia.

Performance varies by use, configuration, and other factors. Learn more at www.Intel.com/PerformanceIndex.