Choice of model of the finite state automation
With the help of ZUBR package synthesis of six classes of finite state
automations A, B, C, D, E and F which may be effectively realized on
PLD, and also four combined models of ADE classes, AD, AE and BF is
possible(probable). As on inputs of all structural models of finite
state automations registers or latches, and on outputs - registers the
total number of models of finite state automations which admit
implementation on PLD, is equaled 50 may be installed. Therefore choice
of the most approaching model finite at its implementation on PLD is
not the trivial task. The task even more becomes complicated, when it
is necessary to select group of the most effective models for some set
of finite state automations, for example, meeting in the projected
digital system.
The most essential factors influential in choice of structural model of
the finite state automation at its implementation on PLD are the
following:
The System requirements showed to the finite state automation, are the
most important at choice of structural model. For example, if usage of
the finite state automation at the first level of the digital system on
which inputs signals act in arbitrary instants it is necessary to use
structural models with registers or latches on inputs is supposed. On
the other hand, if usage of the finite state automation at the last
level of the digital system models with registers on outputs, as a
rule, are required is supposed. For digital systems with a pipelining
of the data important that in each clock tick of synchronization at
each system level operations on data processing were completed.
Therefore in pipeline systems at all levels models with registers on
outputs are usually used. Especially closely it is necessary to concern
to usage of models with combinatorial inputs and outputs when output
functions of the finite state automation immediately depend on entry
variables. In such models any modifications on inputs will cause
changes values of outputs. For construction of models of finite state
automations the following architectural properties PLD are used:
(1) - possibility of the configuration of output macrocells PLD with
the trigger in a chain of feedback;
(2) - presence of two types of link (combinatorial and register)
between an entry buffer and internal logic PLD;
(3) - presence of one type of link (register) between an entry buffer
and internal logic PLD;
(4) - possibility of the configuration of output macrocells with a
register output(exit) and combinatorial feedback;
(5) - possibility of the configuration of entry macrocells with latches
as state elements.
For construction of separate models simultaneous presence of several
properties PLD is required. The PLD also concerns to architectural
properties possibility of buffering of input or output signals which is
taken into account at definition of cost of implementation of models.
The Purpose of choice of models of finite state automations depending
on temporary parameters is support of maximum speed of the digital
system. Thus it is necessary to take into account the type of the
digital system: sequential or pipeline. In case of the sequential
digital system also it is necessary to distinguish systems with
separate and common synchronization.