The compilation and optimization of a microprogram for a computer with a horizontal control word format is highly machine dependent. The compilation phase involves the detection of parallelism in a sequence of source statements followed by a mapping into a sequence of microoperations. The optimization phase involves combining these microoperations to form microinstructions in an optimal manner. One of the important subprocesses of the optimization phase is determining whether or not two or more microoperations can be executed concurrently.
This paper presents the Control Word Model for determining when two or more microoperations can be executed concurrently. The Control Word Model is a machine independent model of the semantics of the control words for microprogrammable computers. As will be demonstrated, the concurrency permitted is sometimes determined not simply by the hardware configuration of buses, register, memories, and ALUs but also by the format of the control word chosen by the designer. This observation has motivated the development of the Control Word Model.
[1]
Robert G. Barr,et al.
A Research-Oriented Dynamic Microprocessor
,
1973,
IEEE Transactions on Computers.
[2]
David J. DeWitt,et al.
A microprogramming language for the B-1726
,
1973,
MICRO 6.
[3]
Stephen S. Yau,et al.
On storage optimization of horizontal microprograms
,
1974,
MICRO 7.
[4]
Donald R. Oestreicher.
A microprogramming language for the MLP-900
,
1973,
MICRO 1973.
[5]
Masahiro Tsuchiya,et al.
An approach to optimization of horizontal microprograms
,
1974,
MICRO 7.
[6]
C. V. Ramamoorthy,et al.
Execution time (and memory) optimization in microprograms
,
1974,
MICRO 7.
[7]
Harold W. Lawson,et al.
The Datasaab FCPU microprogramming language
,
1973,
MICRO 1973.
[8]
R. G. Barr,et al.
AMP, a dynamic microprocessor
,
1973
.