1 / 17

The Grid from a User´s Perspective: The Cluster Finder Use Case

This article explores the use of grid computing for improving code speed, specifically in the context of the Cluster Finder use case in the study of galaxy clusters. It discusses the benefits, challenges, and opportunities for parallelization and distribution, as well as the desire for support in code portability.

wlofton
Download Presentation

The Grid from a User´s Perspective: The Cluster Finder Use Case

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. The Grid from a User´s Perspective: The Cluster Finder Use Case Art Carlson AstroGrid-D, Heidelberg, 24 July 2006

  2. The bottom line • The main thing I want the grid for is improving my top speed. • I know I have to parallelize my code myself, but I want some support doing it. • I don‘t want to have to port my code to more than one environment.

  3. Why galaxy clusters?

  4. Dark Matter + Plasma + Galaxies

  5. Cluster Finder Workflow

  6. Likelihood maps

  7. No shortage of data or questions • More sky: 140 sq-deg → 10,000 sq-deg • Deeper surveys: 2000 times more data • Parameter studies: e.g. Cluster model • More sophisticated analysis: e.g. More spectral/energy channels

  8. Where does the Grid come in? • Speed – but sprinting, not marathon • Get it by • Parallelization • Distribution • Pleasant side effects • Collaboration • Logistics • Monitoring and restart

  9. Parallelization: regions of the sky

  10. Parallelization: regions of the sky

  11. Parallelization: parameter scans

  12. Parallelization: reassembly

  13. Parallelization: data streams

  14. Distribution: file management • Read from data base or source file or cache • Written locally or remotely, in one file or several • In ASCII or binary or data base format • I don‘t care about any of that

  15. Distribution: execution Many opportunities for trouble: • Architecture • Compiler • System routines • Libraries Less so with Java, more so with Fortran

  16. Distribution: execution • Installation – the tried and true method, but too much work and a shame • Metadata – also a lot of work and not foolproof • Self-tests – recognizes trouble early and reliably, but doesn‘t solve the problems • Virtual environment???

  17. The bottom line • The main thing I want the grid for is improving my top speed. • I know I have to parallelize my code myself, but I want some support doing it. • I don‘t want to have to port my code to more than one environment.

More Related