Information Technology Reference
In-Depth Information
The 100% coverage level is established by a sufficient amount of items per group
(i.e., test cases) whose corresponding applied stimuli and observed responses match
the parameter group characteristics. The larger the number of items is considered,
the stronger the functional verification process will be (Tasiran & Keutzer, 2001).
19.5.1
Coverage Analysis
Hekmatpour suggests that the functional verification should be carried out by fol-
lowing several phases such as planning, reference model implementation, coverage
analysis, and others (Hekmatmpour & Coulter, 2003).
The coverage analysis is an important phase for certifying the test bench robust-
ness. After the test bench application, in case of evidence of coverage holes, the
stimuli generation should be redirected, and the verification should restart until no
missing coverage aspects are found.
Under random stimuli, the coverage evolution, in terms of time, presents a fast
growth in the initial phase of the test bench application, and then it follows a saturation
tendency if higher levels of coverage are reached as a result of an increased occurrence
of redundant stimuli.
The functional coverage saturation effect has motivated two types of techniques
known as closed-loop test benches and reactive test benches. One example of a
closed-loop test bench technique, is the stimuli filtering technique, which is based on
the observation that simulating the reference model is much faster than performing it
on the RTL model of the DUV, and it shows that important time can be saved without
much computational expense or development effort. Although a basic functional
technique is a good example for a reactive test bench technique.
19.5.2
Evaluating Functional Approach
The importance of the functional verification strategy is shown in the success of the
verification process. With respect to coverage point of view, random stimulation is
a big source of redundant cases (i.e., stimuli that do not increase coverage). Con-
sequently, the effective and appropriate use of random stimulation requires using
techniques to modify the generation patterns according to the desired coverage.
19.5.3
Verification Methods Summary Table
Table 19.2 summarizes and compares some of the verification methods.
19.6 COMPARISON OF COMMERCIALLY AVAILABLE VERIFICATION
AND VALIDATION TOOLS
Many code generation, partial process automation, or test bench generation tools
require the use of additional software tools for patching through to another root
software platform to complete an uninterrupted tool chain. In this section, a brief
Search WWH ::
Custom Search