IPDPS24: Graph Analytics on Jellyfish Topology

The Artifact Description document (Artifact_description.pdf) in this repository contains the necessary information about our open-source GitHub-hosted code artifacts pertaining to our "IPDPS24" paper titled "Graph Analytics on Jellyfish Topology". It outlines how to build and run the applications to reproduce our results, as well as to conduct additional experiments using our implementations.

Repository Description:

This repository contains a collection of routing data organized into various subfolders.

Folder Structure:

Routing Data Subfolders:

Each subfolder follows the naming convention:

routing_data_(application_name)_(topology_name)_and_(equivalent_jellyfish)These subfolders contain routing results for a specific application on a given topology and its equivalent Jellyfish topology, across different input graphs.

File Naming Convention:

Files within these subfolders are named according to:

(input_graph_name)_(topology_name)_(topology_configurations)_(routing_algorithm_name)For example: orkut_dfly_17_9_par

input_graph_name: orkuttopology_name: Dragonfly (dfly)topology_configurations: Configuration details of the topology (e.g., number of switches per group, number of groups).routing_algorithm_name: Progressive adaptive Routing(PAR) - Name of the routing algorithm used. 

Scaling Subfolders:

strong_scaling: Contains data for strong scaling.weak_scaling: Contains data for weak scaling.

Graph-specific Data Subfolders:

bfs_rgg_data: Contains data for random regular graphs on Graph500 BFS application.minivite_rgg_data: Contains data for random regular graphs on Minivite (graph clustering) application.

File Contents:

Each file within the subfolders contains information such as stalls, hops, MPI time, estimated runtime, etc., corresponding to the routing results for the specified configurations.