Loading...
Please wait, while we are loading the content...
Similar Documents
Sample Sort using the Standard Template Adaptive Parallel Library
| Content Provider | Semantic Scholar |
|---|---|
| Author | Tanase, Gabriel Bianco, Mauro Rauchwerger, Lawrence Amato, Nancy M. |
| Abstract | The Standard Template Adaptive Parallel Library (stapl) is a parallel library designed to make developing software that takes advantage of parallelism on multi-processor machines easier. Written to be a superset of the Standard Template Library (stl) for C++, stapl is intended to be straightforward to use, efficient, and portable. Since stapl's core library consists of ISO Standard C++ components, it must provide the ability to sort. We present a parametrized Parallel Sample Sort algorithm, developed with stapl, that provides a choice in the method of sampling, the over-sampling ratio, and the over-partitioning ratio. Unlike many other Sample Sort algorithms, it covers the situation where the load on each processor is not balanced at the beginning of the algorithm without requiring an initial re-balancing step. In addition, we evaluate the general scalability and the conditions under which two versions of the parametrized Parallel Sample Sort algorithm perform well and poorly. This research was supported in part by the Computing Research Association CRA-W Distributed Mentor Program “Sample Sort ...”, Berlin et al. TR07-002, Parasol Lab, Texas A&M, Spring 2007 1 |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | https://parasol.tamu.edu/people/jberlin/sample_sort_tr.pdf |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
