Lab exercise: Molecular Dynamics on Intel® Xeon Phi™ Coprocessor

Through this series of labs (intended to be run on the Linux* operating system), we will use a simple implementation of an nbody pairwise computation using the Lennard-Jones Potential from molecular dynamics, as a representative example to get a basic understanding of how to port and optimize applications for the Intel® Xeon Phi™ coprocessor.

This article is part of the Intel® Modern Code Developer Community documentation which supports developers in leveraging performance in code through a systematic step-by-step optimization framework.

Throughout the exercises, we encourage you to try the labs yourself before peeking at the answer key provided at the end of each Lab document.

As a first step, you will need to modify ~/Labs/Scripts/setup to match the paths where your software and Intel® MPSS are installed

Here is a list of documents that will help you get started on the Intel® Xeon Phi™ coprocessor:

The labs cover the following concepts:

  • Lab 1:
    • Running a serial implementation of the MD code on the Intel® Xeon® host
    • Running the serial implementation natively on the coprocessor
    • Identifying hotspots in the serial host workload
    • Parallelizing the workload using OpenMP
    • Offloading work to the coprocessor
  • Lab 2:
    • Parallelizing the workload using Intel® Cilk™ Plus
    • Offloading work to the coprocessor using the shared memory model
  • Lab 3:
    • Exploring the vectorization report
    • Profiling your application using Intel® VTune™ Amplifier XE
  • Lab 4:
    • Writing a SIMD enabled function (formerly known as an elemental function)
    • Using Intel Cilk Plus array notation

For questions, please visit our Intel® Xeon Phi™ Coprocessor landing zone and Community Forum.

有关编译器优化的更完整信息,请参阅优化通知