Skip to content
No description, website, or topics provided.
C++ Python Shell CMake
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.

MONET Blue Waters

Tool Description

Monet is a congestion characterization toolkit for toroidal networks. We have only released the following: * Network hotspot extraction and characterization tool, which extracts CRs at runtime; it does so by using an unsupervised region-growth clustering algorithm: The clustering method requires specification of congestion metrics (e.g., percent time stalled (PTS) or stall-to-flit ratios) and a network topology graph to extract regions of congestion that can be used for runtime or long-term network congestion characterization. * Congestion vizualization toolkit: This allows to visualize the congestion clouds in the network along with applications. Please launch the jupyter notebook located here. See below references.

Monet was used to characterize the Blue Waters congestion dataset obtained using LDMS. The results and methodology are explained in detail in our NSDI 2020 and HOTI paper.



Refer to for license details.

Steps for using the tool

Create Docker

Build dockers for PCL (Point Cloud Library) and Spark using:

Build PCL Segmentation

Build executable for custom PCL region growth segmentation:

Launch Monet Batch Processing

Runs raw OVIS files through the data processing pipeling and produces congestion regions.
Execute the following command:
./ [host-data-directory] [host-output-directory] [date] [timestamp]


  • host-data-directory : directory containing OVIS files in YYYYMMDD format
  • host-output-directory : directory to output region files
  • date : date that needs to be proccessed (YYYYMMDD format)
  • timestamp : starting timestamp

Output Format

[host-output-directory]/region contains output of the region growth segmentation and region duration algorithms.

There are two CSVs in the output dir:

  • YYYYMMDD : Contains information about the congestion region size, duration and average stall values.
  • YYYYMMDD_edges : Contains information about transitions between regions.
You can’t perform that action at this time.