Coriolis VLSI CAD Tools
The patterns module of Stratus is a set of Python classes and methods that allows a procedural description of input pattern file for the logic simulator. The Stratus Pattern method produces a pattern description file as output. The file generated by Pattern method is in pat format, so IT IS STRONGLY RECOMMENDED TO SEE pat(5) manual BEFORE TO USE IT.
From a user point of view, Pattern method is a pattern description language using all standard Python facilities. Here follows the description of the Pattern method. A pat format file can be divided in two parts : declaration and description part. The declaration part is the list of inputs, outputs, internal signals and registers. Inputs are to be forced to a certain value and all the others are to be observed during simulation. The description part is a set of patterns, where each pattern defines the value of inputs and outputs. The pattern number represents actually the absolute time for the simulator. Similarly, a Pattern method can be divided in two parts : declaration and description part. Methods related to the declaration must be called before any function related to the description part.
The first thing you should do in this part is to instantiate the class Patwrite to have access to all patterns declaration and description methods. The constructor of this class take as parameters the name of pattern output file and the Stratus cell that is described (see PatWrite [patwrite]). Then, this part allows you to declare the inputs, the outputs, and internal observing points (see declar[declar] and declar_interface [declar:sub:interface]).
After all signals are declared, you can begin the description part (see pattern_begin [pattern:sub:begin]). In this part you have to define input values which are to be applied to the inputs of the circuit or output values which are to be compare with the values produced during the simulation. (see affect [affect], affect_any [affect:sub:any], affect_int [affect:sub:int] and affect_fix [affect:sub:fix]). Pattern method describes the stimulus by event : only signal transitions are described. After each event there is a new input in the pattern file (see addpat [addpat]). Last thing you should do in this part is to generate the output file (see pattern_end [pattern:sub:end]).
This class is used to create patterns for Stratus models. Currently it only supports Alliance “.pat” pattern format. Patterns time stamps are in the “absolute date” format, “relative date” isn’t allowed. Legal time unit are ps (default), ns, us and ms. The constructor takes as parameters the pattern output filename and an optional reference to Stratus cell.
Adds a connector from a Stratus model to the pattern interface. Writes the corresponding connector declaration in the pattern file with name, arity and direction automatically extracted from the connector properties. Supported Stratus connectors are:
Adds all the connectors from a Stratus model to the pattern interface. Write the corresponding connector declaration in the pattern file with name, arity and direction directly taken from the connector proprieties.
Affect a string value to a connector.
Affect an integer (CA2) value to a connector. Convert the 2’s complement value to the corresponding binary value. The binary size is taken from the connector arity. If the connector is an output, the binary value is preceded by “?”.
Affect a fixed point value to a connector. Convert the floating point input value to the corresponding fixed point value with word_length=connector.arity() and integer_word_length=iwl. If the connector is an output, the binary value is preceded by “?”.
Disable comparison between this connector value and the one calculated during simulation.
Adds a pattern in the pattern file.
Mark the end of the interface declaration and the beginning of the test vectors.
Mark the end of the test vectors and of the patterns file.
Pattern method for an addaccu
def Pattern(self):
# initialisation
pat = PatWrite(self._name+'.pat',self)
# declaration of ports
pat.declar(self.ck, 'B')
pat.declar(self.load, 'B')
pat.declar(self.input, 'X')
pat.declar(self.output, 'X')
pat.declar(self.vdd, 'B')
pat.declar(self.vss, 'B')
# use of pat.declar_interface(self) has the same effect
# description beginning
pat.pattern_begin()
# affect vdd and vss values
pat.affect_int(self.vdd,1)
pat.affect_int(self.vss,0)
# first pattern : load an initial value
pat.affect_int(self.input,5)
pat.affect_int(self.load,1)
pat.affect_int(self.ck,0)
# add the pattern in the pattern file
pat.addpat()
# compute next event
pat.affect_int(self.ck,1)
pat.addpat()
# compute 22 cycle of accumulation
pat.affect_int(self.load,0)
for i in range(1,22):
pat.affect_int(self.ck,0)
pat.addpat()
pat.affect_int(self.ck,1)
pat.affect_int(self.output,i+5)
pat.addpat()
# end of the description
pat.pattern_end()