Document Type
Conference Proceeding
Publication Date
2003
Department
Marine Science
Abstract
The Naval Research Laboratory’s Code 7440 Production Enhancement Team at the Stennis Space Center has been tasked to develop ways to speedup hydrographic data processing at the Naval Oceanographic Office [Depn02]. This paper presents the final development of a parallelized version of the Pfm_loader application customized to run on a Beowulf cluster.
This paper specifically covers software development efforts in early FY02 [Sarn02 ]. In a series of algorithms, called Schemes D, E, and F, a parallel algorithm previously implemented in Scheme C (see Figure 1) has been integrated with an improved version NAVO’s of the Pure File Magic (PFM) library. Former versions of this software not mentioned, namely Schemes A and B, were used as a design platform for the software architecture found in Scheme C. The main goal of this work was to increase the writing rate of binned data to a physical disk.
The final software version is Scheme F. The final parallel code of Scheme F achieves the best speedup for the largest available test dataset. The simple runs (no filtering) exhibited a speedup of 8 as compared to the original, serial algorithm. Runs with swath filtering showed a top speedup of 8. Runs with area filtering reached a speedup of 6.5. Runs with both swath and area filtering showed a top speedup of 7. In all cases, data strongly suggest that greater speedups could be achieved for larger than tested input datasets.
Section 2 of this paper is devoted to Scheme D. Section 3 presents results on an implemented threaded version of Scheme D. Section 4 deals with results for Scheme E. Section 5 describes Scheme F. Detailed results are presented on the performance of Scheme F when swath and area based filtering are enabled. Section 6 presents results illustrating the effect of the physical disk’s I/O throughput rates on speedup results. Section 7 discusses results and future plans.
Recommended Citation
Depner, J. et al, Dealing with Increasing Data Volumes and Decreasing Resources, Oceans MTS/IEEE, 2002. Sarnowski, K., Miller, M.J., Layne, G., Final Project Report FY02, Hydrographic Research Center, 2002.