Intel® MPI Library

Developer Reference for Linux* OS

Download PDF
ID 768732
Date 10/31/2024
Public
Document Table of Contents
Document Table of Contents x
Developer Reference for Linux* OS
Developer Reference for Linux* OS x
Introduction Command Reference Environment Variable Reference Miscellaneous Notices and Disclaimers
Introduction x
Introducing Intel® MPI Library Notational Conventions Related Information
Command Reference x
Compiler Commands mpirun mpiexec.hydra cpuinfo impi_info mpitune_fast Workload Managers  Support MPI Options Support
Compiler Commands x
Compilation Command Options
mpiexec.hydra x
Global Hydra Options Local Hydra Options gtool Options
Environment Variable Reference x
Compilation Environment Variables Hydra Environment Variables I_MPI_ADJUST Family Environment Variables Tuning Environment Variables Process Pinning GPU Support NIC Pinning Environment Variables for Fabrics Control Environment Variables for Memory Policy Control Environment Variables for Asynchronous Progress Control Environment Variables for Multi-EP Other Environment Variables
Tuning Environment Variables x
Autotuning
Autotuning x
I_MPI_TUNING_AUTO Family Environment Variables
Process Pinning x
Environment Variables for Process Pinning Interoperability with OpenMP* API
GPU Support x
GPU Pinning GPU Buffers Support
Environment Variables for Fabrics Control x
Communication Fabrics Control Shared Memory Control OFI*-capable Network Fabrics Control
Miscellaneous x
Java* Bindings for MPI-2 Routines
Developer Reference for Linux* OS

mpitune_fast

This utility tunes the Intel® MPI Library to the cluster configuration using the Autotuning functionality.

Syntax

mpitune_fast <options>

Arguments

Options Description
-c | --colls Set custom collective operations to tune, delimited by commas.
-d | --results_dir <path-to-results-dir>

Set custom directory for tuning results, host files, and logs. Default: the current working directory.

-h | --help Display the help message.
-n <n> Specify the number of nodes. This can be a comma-delimited set of values to set up several launches.
-pd <path-to-performance-results-dir>,

--perf_results_dir <path-to-performance-results-dir>

Set a custom directory for validation performance results.

Default: performance_results_<timestamp>.

-ppn <n> Specify the number of processes per node. This can be a comma-delimited set of values to set up several launches.
-val <path-to-tuning-file>,

--validate <path-to-tuning-file>

Run the tuning file validation cycle to validate the existing tuning file.

Description

The mpitune_fast utility allows you to automatically set up the Intel MPI Library and launch with Autotuning enabled and configured for your cluster configuration.

The tool iteratively launches the Intel® MPI Benchmarks utility with the proper autotuner environment and generates a .dat file with the tuning parameters for your cluster configuration.

After generation the tuning file, set it as I_MPI_TUNING_BIN: 

$ export I_MPI_TUNING_BIN=./tuning_results.dat
$ mpiexec <args>

Workload Managers  Support

mpitune_fast supports Slurm and LSF workload managers. It automatically defines job allocated hosts and performs launches.

MPI Options Support

The following MPI options are available within the utility:

Options Description
-f <filename> Specify a file containing host names.
-hosts HOSTS, --hosts HOSTS Set the host names, delimited by commas. Example: --hosts host1,host2

Example (default launch): 

$ mpitune_fast -f ./hostfile

Example (customized launch):

$ mpitune_fast -ppn 8,4,2,1 -f ./hostfile -c alltoall,allreduce,barrier

See Also

Autotuning

MPI Tuning in the Intel MPI Library Developer Guide

For available configuration options, refer to mpitune Configuration Options.

Cluster-Wide MPI Tuning Using Intel® MPI Library

Parent topic: Command Reference
Level Two Title