Efficient hardware-acceleration for sparse-matrix computation

Abstract

Sparse Matrix to Vector Multiplication (SMVM) are very prevalent in many commercial, industrial and national security mission~s software. This proposal could offer unprecedented performance for these missions. With the increasing amount of data collected, more efficientways of processing this data are needed. Another benefit of this advanced Sparse Vector Matrixhardware accelerator is that it is compatible with all levels of compute platforms from embedded systems to super computers. This means that the results of this proposal will have broad impact on USG programs within both DOE and DOD. The results of this proposal could also be applied to commercial applications.

Document Details

Document Type
DoD Grant Award
Publication Date
Mar 03, 2017
Source ID
N000141712225

Entities

People

  • Eugenio Culurciello

Organizations

  • Office of Naval Research
  • United States Navy
  • University of Virginia

Tags

Fields of Study

  • Computer science

Readers

  • Defense Technology Research and Development.
  • Distributed Systems and Data Platform Development
  • Linear Algebra