TITLE:

Rethinking the Union of Computed Tomography Reconstruction and GPGPU Computing

DATE:

Friday, March 15th, 2013

TIME:

3:30 PM

LOCATION:

GMCS 214

SPEAKER:

Edward Steven Jimenez.
Sandia National Laboratories.

ABSTRACT:

This talk will discuss the utilization of the massively multi-threaded
environment of graphics processors (GPUs) to improve the computation
time needed to reconstruct large computed tomography (CT) datasets
and the arising challenges for system implementation. Although many
GPU-based CT reconstructions exist in the medical community, many
of these implementations are better suited for smaller application (less
than 10 Gigabytes). Previous work has shown that CT reconstruction
on GPUs becomes an irregular problem for large datasets (10GB-4TB),
thus memory bandwidth becomes a signi?cant bottleneck for industrial
applications. Optimizing the GPU kernel only relocates the bottleneck
to the system storage. This problem can be addressed by considering a
combination of load-balancing, dynamic task allocation, hybrid comput-
ing, and MIMD-type environments. Sandia National Laboratories is a
multi-program laboratory managed and operated by Sandia Corporation,
a wholly owned subsidiary of Lockheed Martin Corporation, for the U.S.
Department of Energy’s National Nuclear Security Administration under
contract DE-AC04-94AL85000. This document serves only to present the
title and abstract for the presentation.

HOST:

Dr. Jose Castillo.

DOWNLOAD:

PPTX File