https://github.com/matthiasdiener/numalize

Detect memory access patterns of parallel applications
https://github.com/matthiasdiener/numalize

c-plus-plus parallel-applications pin r

Last synced: about 1 month ago
JSON representation

Detect memory access patterns of parallel applications

• Host: GitHub
• URL: https://github.com/matthiasdiener/numalize
• Owner: matthiasdiener
• License: gpl-2.0
• Created: 2015-03-13T14:49:35.000Z (almost 10 years ago)
• Default Branch: master
• Last Pushed: 2019-02-07T19:07:14.000Z (almost 6 years ago)
• Last Synced: 2024-08-03T19:09:38.685Z (4 months ago)
• Topics: c-plus-plus, parallel-applications, pin, r
• Language: R
• Homepage:
• Size: 143 KB
• Stars: 19
• Watchers: 3
• Forks: 7
• Open Issues: 2
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

• awesome-numa - Numalize - A memory tracing tool to detect communication and page usage of NUMA applications (Observation and profiling tools)

README

        
# Numalize: Detect memory access patterns of parallel applications

Numalize is a memory tracing tool to detect communication (i.e. accesses to shared memory areas) and page usage of parallel applications that use shared-memory APIs (e.g. OpenMP and Pthreads).

It is based on the Intel Pin Dynamic Binary Instrumentation (DBI) tool (https://software.intel.com/en-us/articles/pintool/). 

## Requirements 

- Intel Pin installation (https://software.intel.com/en-us/articles/pintool/). The ```Makefile``` tries to find your Pin installation in ```/home``` and ```/opt```. You can override the path manually in the top of the ```Makefile```. Note that Pin version 3.x is currently required.

## Usage

Compile numalize:

    $ make

Generate communication pattern:

    $ ./run.sh -c -- ./binary

    

    

Generate page access pattern:

    $ ./run.sh -p -- ./binary

    

## Publication

Numalize is described in:

- Matthias Diener, Eduardo H. M. Cruz, Laércio L. Pilla, Fabrice Dupros, Philippe O. A. Navaux. **“Characterizing Communication and Page Usage of Parallel Applications for Thread and Data Mapping.”** Performance Evaluation, 2015. http://dx.doi.org/10.1016/j.peva.2015.03.001