Title :
CT image reconstruction using Larrabee
Author :
Baer, Matthias ; Maass, Clemens ; Kachelries, M.
Author_Institution :
Inst. of Med. Phys. (IMP), Univ. of Erlangen-Nurnberg, Erlangen, Germany
fDate :
Oct. 24 2009-Nov. 1 2009
Abstract :
CT image reconstruction is a computationally very demanding task. Recently a new many core platform codename Larrabee (LRB), which is currently developed by Intel, has been described by Seiler et al. In this work we want to present different approaches on how to use the new LRB platform in CT image reconstruction. Therefore we will give examples of LRB optimized code sections which can be used instead of standard code for CT image reconstruction. Since multithreaded programming on LRB is similar to other multicore platforms we will here focus on describing the use of LRB´svector instruction set. Additionally some major facts about the LRB platform and the LRB intrinsic instruction set will be given.
Keywords :
computerised tomography; image reconstruction; medical image processing; CT; Larrabee; image reconstruction; multithreaded programming; vector instruction; Biomedical imaging; Code standards; Computed tomography; Computer architecture; Engines; Hardware; Image reconstruction; Multicore processing; Nuclear and plasma sciences; Yarn;
Conference_Titel :
Nuclear Science Symposium Conference Record (NSS/MIC), 2009 IEEE
Conference_Location :
Orlando, FL
Print_ISBN :
978-1-4244-3961-4
Electronic_ISBN :
1095-7863
DOI :
10.1109/NSSMIC.2009.5402422