Skip to content

Conducted a comparative analysis on scaling results for graph algorithms like single source shortest path (SSSP) algorithm using multiple runtime systems thereby reducing time to solution for graph processing with message driven runtime systems (like Charm++, HPX5), parallel boost graph library, or Graph500

Notifications You must be signed in to change notification settings

prateek22sri/Runtime-system-benchmarking

Repository files navigation

runtime_system_analysis

Graph problems are one of the most important problems used for benchmarking purposes. Among other graph problems, single source shortest path problem(SSSP) is relatively more difficult compared to other graph problems like breadth first search. There are many frameworks or runtime systems that are dedicated to solve such problems. These systems use different architectures, different methods for parallelization, different means for synchronicity and so on. All these parameters responsible for the performance of these systems individually make these comparisons heterogenous. Here the author aims to achieve a comparison for the performance of these runtime systems which will provide a baseline and help understand the innate differences among them. The comparison is done by assuming these runtime systems are individual black boxes with a fixed input and output. The metrics calculated in the process quantize the differences or benchmark each runtime systems in reference to SSSP.

To read more check the file "Comparative Analysis of different AMT systems using SSSP.pdf"

About

Conducted a comparative analysis on scaling results for graph algorithms like single source shortest path (SSSP) algorithm using multiple runtime systems thereby reducing time to solution for graph processing with message driven runtime systems (like Charm++, HPX5), parallel boost graph library, or Graph500

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published