Application-specific arithmetic circuit design for a particle tracking application

The main bottleneck in system LSI performance is, in many cases, the processing overhead that occurs when arithmetic calculations are performed multiple times. The circuit size and processing time required for multiplication, division and other basic arithmetic operations are often especially large....

Full description

Saved in:
Bibliographic Details
Main Authors: Kambe, T., Takehara, K., Tsukiyama, S.
Format: Conference Proceeding
Language:English
Subjects:
Acceleration
Correlation
Educational institutions
Fluids
Logic gates
Pipeline processing
Random access memory
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The main bottleneck in system LSI performance is, in many cases, the processing overhead that occurs when arithmetic calculations are performed multiple times. The circuit size and processing time required for multiplication, division and other basic arithmetic operations are often especially large. In this paper we propose an application-specific arithmetic circuit design methodology and apply it to the design of a correlation coefficient calculation circuit for a particle tracking application. Particle tracking velocimetry (PTV) measures velocity fluctuations in fluids and gases and has been applied to a diverse range of flows. However, the algorithms are computationally intensive so to achieve real-time processing speeds implementation in hardware is essential. The computational overhead is accelerated by applying code refactoring, memory access optimization, and variable length pipelining. The processing speed and the circuit size of the design are evaluated. We show that the hardware solution is 1,000 times faster than the software solution and is able to achieve real-time processing speeds.
DOI:10.1109/LASCAS.2013.6518978