SystemVerilog Assertion
Part 2: Sequence - An Introduction
Prev: Sequence Layer - Introduction |
Next: Sequence Repetition Operator
Sequence and Clock
One of the most important aspects of concurrent assertion (and thus of
sequences) is that it always works at a clock edge. All
expressions are evaluated at an edge and all actions corresponding to
the values of those expressions are also carried out at an edge. If an
expression changes value at an intermediate point of a clock cycle,
concurrent assertion is blind to that change. This has been done so
that a concurrent assertion samples only stable values of all variables.
Implied Clock
So, it is now natural for you to ask which clock the sequence
Sequence1 (reproduced below from our earlier example) for
calculating ##5 uses since there is no clock in the sequence at all?
sequence Sequence1;
~reset ##5 req;
endsequence
The same question holds true for Sequence2. The answer to this is that
both this sequences use something known as 'implied clock' - a clock
that is supplied to these sequences when they are instantiated by an
element of layers above the sequence layer (property layer, for example).
Whichever clock that higher level element has, the sequence inherits the
same.
Using Clock inside a Sequence
Although we are already using an implied clock in Sequence1, a clock
reference in a sequence does not always have to be implied. If you know the
name of the implied clock, you can use it directly inside your sequence.
sequence Sequence3;
@(posedge clk_1) // clock name is clk_1
s1 ##2 s2; // two sequences
endsequence
In the above example, as we saw in a previous example, Sequence3
uses two subsequences s1 and s2. While subsequences can be
nested (s1 uses subsequence s2, s2 uses subsequence
s3 and so on), beware of the cross-coupled subsequences - for example,
s1 uses s2 as its subsequence and s2 uses s1
as its subsequence - that causes infinite looping.
A sequence also can have multiple clocks, a subject that we will come back
to in the later part of this series.
Sequence Operations
Now we will discuss operations that can be performed on a sequence or
between two sequences. Just as x+y, where x and y are integers, produces
a new integer (their sum), these operations produces a new sequence
from one or more existing sequences. We are already familiar with one
such operator - the cycle delay operator '##' that 'joins' two sequences
(or one sequence and an event) with a number of clock cycles between
them.
There are three main categories of sequence operations as Figure 2
shows.
| Category
| Operators
| Associativity
|
| Repetition
| [* ] [= ] [-> ]
| -
|
| Cycle Delay
| ##
| Left
|
| Match
| throughout, within, intersect, and, or
| right for throughout, left for others
Figure 2. Sequence Operations
|
In this Part 2, we will discuss the repetition operators. In the
next part, we will look into the match operators.
Prev: Sequence Layer - Introduction |
Next: Sequence Repetition Operator
|
