Concurrent threads for parallel electron tomography

Authors:
J. A. Álvarez;J. Roca;J. J. Fernández
Affiliations:
Department of Computer Architecture and Electronic, University of Almería, Almería, Spain;Department of Computer Architecture and Electronic, University of Almería, Almería, Spain;Department of Computer Architecture and Electronic, University of Almería, Almería, Spain
Venue:
ACS'06 Proceedings of the 6th WSEAS international conference on Applied computer science
Year:
2006

Citing 3
Cited 0

CHARM++: a portable concurrent object oriented system based on C++

OOPSLA '93 Proceedings of the eighth annual conference on Object-oriented programming systems, languages, and applications
Charisma: a component architecture for parallel programming

Charisma: a component architecture for parallel programming
Performance evaluation of adaptive MPI

Proceedings of the eleventh ACM SIGPLAN symposium on Principles and practice of parallel programming

Quantified Score

Hi-index	0.00

Visualization

Abstract

High Performance Computing (HPC from now) is a characteristic defined and developed with the aim of providing a feasible framework in which large-scale and grand-challenge applications can be run within acceptable time bounds. Electron tomography of large specimens sets up a field in which applications are time and resources consuming. Large scale reconstructions by means of WBP as well as voxel-based iterative methods were adapted to HPC by parallel computing. Porting of the reconstruction algorithms into the HPC framework was a task accomplished using MPI libraries and an experienced programmer due to the complexity of overlapping computation and communication tasks needed to achieve good performance. In contrast to MPI, AMPI (Adaptive MPI) provides a solution to easily port legacy MPI code into a multithreaded environment where overlapping is reached naturally. This paper shows an study of how AMPI behaves in our research group's cluster as well as results obtained when it is applied on the software developed by our group in order to solve the tomographic reconstruction problem.