Document Type
Thesis
Date of Award
5-31-1989
Degree Name
Master of Science in Electrical Engineering - (M.S.)
Department
Electrical Engineering
First Advisor
John D. Carpinelli
Second Advisor
Edwin Hou
Third Advisor
Irving Y. Wang
Abstract
Simultaneity is a key to high speed and large computation. Assuming hardware components of a given speed, it is the only remaining consideration in achieving raw speed. Simultaneity can be shackled by dependencies, however, and years of hardware and software work have been devoted to understanding the types of dependencies and how they can be obeyed or removed from a computation.
In this thesis, the principles of the data flow machines and the structure of the data flow languages are illustrated. Various methods of achieving high speed are discussed. These methods include streams, Lenient Cons (I-structure), and dependence-driven computation, which is an alternative approach.
Recommended Citation
Yu, Chung-Lin, "Methods of achieving high speed on data flow machines" (1989). Theses. 2977.
https://digitalcommons.njit.edu/theses/2977
