Overview

HTTPS SSH
  Heat map software for analysis of SWF workloads
  ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 


Background
=========================================================
This software includes utilites to generate heat maps from workloads 
traces and to generate comparison plots between workloads.

Examples of real world usage can be found in:

 * David Krakov and Dror G. Feitelson, “High-Resolution Analysis of Parallel Job Workloads”.
   Job Scheduling Strategies for Parallel Processing, May 2012
   Soft copy available at http://www.cs.huji.ac.il/~feit/parsched/jsspp12/p10-krakov.pdf
   
 * David Krakov and Dror G. Feitelson, "Comparing Performance Heatmaps”.
   Job Scheduling Strategies for Parallel Processing, May 2013
   Soft copy available at http://www.cs.huji.ac.il/~feit/parsched/jsspp13/krakov.pdf

This software was tested and used for traces obtained from the 
paralllel workloads acrchive only (http://www.cs.huji.ac.il/labs/parallel/workload/)
but is compatible with any trace in the SWF format.



Installation
=========================================================
To use, following minimal package versions are required:
 - Python 2.7
 - Matplotlib 1.2.1
 - Numpy 1.6.2
 - Scipy 0.11.0


Usage
=========================================================

To generate hexagonal heatmap from log:
  python plot_local_heatmap_log.py  --help 

To generate square (grid) heatmap from log:
  python plot_grid_heatmap.py --help

To compare two logs
  python compare.py --help
  
In all scripts when choosing output file name the extension sets its
format. Everything matplotlib supports goes, specifically: png, jpeg, pdf, ps
  


TODO
=========================================================

 * Specifiying color map not implemented
 * Does not handle newlines or trailing spaces well in logs




Disclaimer
=========================================================

See the LICENSE file for copyright information. 

I'm not very available for bug fixing, but I would be glad to push 
patches in if they come.