Installation

PPMD is developed to run on Linux 64bit machines. Other operating systems may work but are not actively developed upon at this time.

Dependencies

Required tools:

  • Python 2.7
  • mpi4py compatible MPI
  • A C compiler, preferably intel
  • CUDA Toolkit if targeting CUDA.

Required Python Packages

  • NumPy
  • MPI4Py
  • ctypes
  • cgen
  • PyCUDA (If targeting CUDA)

Optional Python Packages

  • matplotlib (if plotting is required)

To generate gpu code PPMD requires that the CUDA Toolkit is installed see further: CUDA.

Environment Variables

  • MPI_HOME: Used by the default CUDA and MPI compiler configurations to locate the desired MPI implementation.
  • BUILD_DIR: The directory used as a storage location for generated libraries, default /tmp/build.

Compilers

Compliers are defined in the compilers sub-directory of the config_dir directory found in the main library directory. Future versions will support loading configurations from alternative directories. Each complier is defined in a separate file and is identified by name in the main configuration file.

CUDA

The PPMD CUDA toolchain gpucuda expects to find a CUDA toolkit installed at the environment variable CUDA_INSTALL_PATH.

Table Of Contents

Previous topic

Introduction

Next topic

Examples

This Page