Spezifikation eines Benchmarks für Daten-intensive Computing

From Lsdf
Revision as of 12:05, 2 October 2013 by Tao (talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)

Zurück zur Themenliste

This work can also be extended to a master thesis.

Description

The project DASH [1] aims at providing a C++ template library to support exascale data-intensive computing. The library provides functions for processing large data as well as complex data structures, and facilities for multi-layer data locality. For validating the outcome of the project, several realistic applications will be developed. However, a standard benchmark suite is also required.

Tasks of the Bachelor thesis

Within the Bachelor thesis such a benchmark suite will be developed. The applications in the suite are chosen from existing standard benchmarks [2][3][4]. These standard applications will be first evaluated to see whether they are suitable for evaluating the DASH library. In the second step, selected applications will be ported to DASH, i.e., modified using the DASH library functions. The concrete tasks of the thesis include:

  • Understanding DASH and standard benchmarks.
  • Evaluation of the standard benchmark applications.
  • Modifying selected applications.
  • Testing the applications on DASH platforms.
  • Write-up thesis.

Requirements

The work requires background knowledge about parallel and distributed computing and programming skill in C/C++.

References

[1] DASH: http://www.dash-project.org/
[2] SPEC benchmark suite: http://www.spec.org/benchmarks.html
[3] NAS benchmark suite: www.nas.nasa.gov/publications/npb.html‎
[4] SPLASH benchmark suite: http://www.capsl.udel.edu/splash/

Contact

Dr. Jie Tao: jie.tao@kit.edu