LLNL/Mitos

Name: Mitos

Owner: Lawrence Livermore National Laboratory

Description: Memory Sampling Tool using Linux perf_events

Created: 2015-01-19 22:12:58.0

Updated: 2017-11-20 08:41:38.0

Pushed: 2016-11-17 00:49:37.0

Homepage: null

Size: 194

Language: C++

GitHub Committers

UserMost Recent Commit# Commits

Other Committers

UserEmailMost Recent Commit# Commits

README

Mitos

Mitos is a library and a tool for collecting sampled memory performance data to view with MemAxes


Quick Start

Requirements

Mitos requires:

Building
  1. Make sure that Dyninst is installed and its location is added to the CMAKE_PREFIX_PATH environment variable.

  2. Run the following commands from the root of the MemAxes source:

    r build && cd build
    e -DCMAKE_INSTALL_PREFIX=/path/to/install/location ..
    
     install
    
Running
  1. Find the mitosrun command in the bin directory in the install directory.

  2. Run any binary with mitosrun like this to generate a folder of mitos output data. For example:

    srun ./examples/matmul
    

    The above command will run the matmul example and create a folder called mitos_###, where ### is the number of seconds since the epoch. The folder will contain:

    s_###/
    ta/
     samples.csv
    c/
     <empty>
    rdware.xml
    

    Where samples.csv contains a comma-separated list of memory samples, hardware.xml describes the hardware topology (using hwloc) and src is an empty directory where you can put the program source files for use in MemAxes.

    mitosrun can also be fine-tuned with the following parameters:

    ions]:
    b sample buffer size (default 4096)
    p sample period (default 4000)
    t sample latency threshold (default 10)
    

Authors

Mitos and MemAxes were written by Alfredo Gimenez.

Thanks to Todd Gamblin for suggestions and for giving Mitos a proper build setup.

License

Mitos is released as part of MemAxes under an LGPL license. For more details see the LICENSE file.

LLNL-CODE-663358


This work is supported by the National Institutes of Health's National Center for Advancing Translational Sciences, Grant Number U24TR002306. This work is solely the responsibility of the creators and does not necessarily represent the official views of the National Institutes of Health.