Accession Number : ADA163195
Title : Highly Concurrent Scalar Processing.
Descriptive Note : Doctoral thesis,
Corporate Author : ILLINOIS UNIV AT URBANA COORDINATED SCIENCE LAB
Personal Author(s) : Hsu,Peter Y
Report Date : Jan 1986
Pagination or Media Count : 122
Abstract : High speed scalar processing is an essential characteristic of high performance general purpose computer systems. Efficient concurrent execution of scalar code is difficult due to data dependencies and conditional branches. This thesis proposes a code scheduling heuristic call the decision tree scheduling (DTS) technique for general scalar code, and an optimal code scheduling algorithm called the simple loop scheduling (SLS) algorithm for a restricted class of innermost loops. Also proposed is a highly concurrent machine architecture that takes advantage of these scheduling techniques. Keywords: parallel processing; throughput; instructions; multiprocessors.
Descriptors : *CODING , *SCHEDULING , ALGORITHMS , DATA PROCESSING , OPTIMIZATION , HIGH RATE , COMPUTER ARCHITECTURE , THESES , PARALLEL PROCESSING , DECISION THEORY , THROUGHPUT , HEURISTIC METHODS , MULTIPROCESSORS , SCALAR FUNCTIONS , INSTRUCTIONS , LOOPS
Subject Categories : Computer Hardware
Distribution Statement : APPROVED FOR PUBLIC RELEASE