|
USM LINKS Computer Science
WEB PAGES CMInfo
RESOURCES Math Resources |
SYMPOSIUM
MINISYMPOSIA
|
December 4-5, 1998
University of Southern Mississippi
Hattiesburg, Mississippi
B. Kasche, D. Fey and W. Erhard
In this paper we show the development of a 3 dimensional optoelectronic hardware approach to realize an arithmetic logical unit. For that we show the main ideas of the low level algorithm. We will introduce several concepts and evaluate them with regards to the highest throughput. At the end will will focus on application of our 3d approach, especially on algorithm for medical image processing.
Introduction
Optics is said to become one of the most important components for computing hardware
in the near future. This fact is motivated by the problems which are generated by
using pure electronics for data processing with a high demand on
communication. Optoelectronic can help to overcome those difficulties.
Optics and electronics have to form a synergetic union in order to combine the
advantages of both of them efficiently. To fulfill this aim the chosen algorithms
play an important role, too. That means only algorithm, which are absolutely well
adapted to the hardware, in our case to the 3d optoelectronic hardware, are suitable to
achieve high performance computing.
In our approach we designed an arithmetic logical unit. For that we designed an
integer and a fix point unit to be able to calculate standard functions for a data,
which is given in a floating point representation. By using optics we could
employ the third dimension for data processing.
In this paper we emphasize the fix point unit. We have developed several approaches
which are based on the so called Bit- and CORDIC algorithm belonging to
the class of add and shift algorithm. These kind of low level algorithm were
developed further and we could push eight of them into a unique structure.
Out of all developed concepts we want to determine that processing method which is
the most efficient. These concepts are characterized by the art of data
processing. We used a bit serial and several bit parallel methods. One of the last
is a method using a redundant number representation. By using the redundancy it is
possible to add any two numbers within a constant time.
The whole process of developing several processing methods and exploiting them to
realize high performance computing is done with the main attention to an
optoelectronic realization.
The design process of a synergetic relationship between optics, electronics and last but not least between the low level algorithms are to be continued to the application algorithms. These algorithms are called high level algorithms. As an example we will introduce algorithms which are necessary for 3d medical image processing. We will show, that we are able to process 3 dimensional dataset by using our approach.
To obtain more information about the meeting send e-mail to: fscc98@pax.st.usm.edu.