Get started with parallel application development
Introduction
Before you start
Debug your application
Optimize your code
Next Steps

  Log an issue

  Fork and edit

  Discuss on Discord

About this Learning Path

Skill level:  Advanced
Reading time:  30 min
Last updated:  29 Sep 2026
Skill level:
  Advanced
Reading time:
  30 min
Last updated:
  29 Sep 2026
Author:Florent Lebeau, Arm
Arm IP:Neoverse
Tags: Performance and Architecture AWS Graviton Microsoft Azure Cobalt Google Axion Oracle Cloud Infrastructure (OCI) Ampere Compute Linux Fortran GCC Linaro Forge gdb mpi Runbook
Author:
  • Florent Lebeau, Arm
Arm IP:
Neoverse
Tags:
Performance and Architecture AWS Graviton Microsoft Azure Cobalt Google Axion Oracle Cloud Infrastructure (OCI) Ampere Compute Linux Fortran GCC Linaro Forge gdb mpi Runbook

Who is this for?

This is an advanced topic for high-performance computing software developers writing MPI applications.

What will you learn?

After completing this Learning Path, you'll be able to perform the following tasks:

  • Debug and fix a parallel application
  • Profile and optimize your code
  • Use optimized routines for common math operations

Prerequisites

Before starting, you'll need the following prerequisites:

  • General knowledge about distributed parallelism (MPI)
  • Some understanding of C, Python, and Linux commands
  • An Arm computer running Linux. Cloud instances can be used, refer to the list of Arm cloud service providers .

Summary

AI-assisted

This summary was drafted with an approved AI-assisted workflow and reviewed by Arm contributors before publication. Human technical review remains part of the process so the final page reflects engineering rigor, accuracy, and Arm editorial standards.

Close
?
You’ll debug and profile an MPI matrix-multiplication application on Arm Linux with Linaro Forge. First, you’ll build C, Fortran, or Python variants that contain intentional defects, then configure debugging flags and AddressSanitizer. You’ll then identify and fix errors before rebuilding the application to confirm stable execution. Finally, you’ll profile unoptimized and optimized builds, compare equivalent math libraries, and use profiler output to assess each build and library choice.
Next