Intel® MPI Library

Potential issues with the RDMA translation cache in the Intel® MPI Library

The Intel® MPI Library comes with a cache that helps to speed up the translation of memory addresses in between the MPI library and the underlying DAPL fabric. As the Intel MPI Library documentation mentions: "The cache substantially increases performance, but may lead to correctness issues in certain situations."

  • Linux*
  • Intel® MPI Library
  • IMPI
  • RTC
  • Parallel Computing
  • Tracing and Correctness Checking

    Intel® MPI Library provides tight integration with the Intel® Trace Analyzer and Collector, which enables you to analyze MPI applications and find errors in them. Intel® MPI Library has several compile-time options to simplify the application analysis.

    Intel® Trace Analyzer and Collector is available as part of the Intel® Parallel Studio XE Cluster Edition. Before proceeding to the next steps, make sure you have the product installed.

    Fault Tolerance Support

    Intel® MPI Library provides extra functionality to enable fault tolerance support in MPI applications. The MPI standard does not define behavior of MPI implementation if one or several processes of an MPI application are abnormally aborted. By default, Intel® MPI Library aborts the whole application if any process stops.

    I_MPI_FAULT_CONTINUE

    Turn on/off support for fault tolerant applications.

    Syntax

    I_MPI_FAULT_CONTINUE=<arg>

    mpitune Utility

    mpitune

    Use the mpitune utility to find optimal settings for the Intel® MPI Library relevant to your cluster configuration or your application.

    Syntax

    mpitune [options]

    Options

    -a \"<app_cmd_line>\"

    --application \"<app_cmd_line>\"

    Subscribe to Intel® MPI Library