Download.pdf

It outlines a series of steps to create stability and flexibility in stimulus generation sequence classes, reduce the footprint of testbench changes, automate sequence/test creation, and improve the ability to absorb modified verification requirements.

The approach involves creating a sequence configuration class that can be modified by the test writer at runtime, allowing the same sequence to be reused by multiple directed tests. A factory is used to override the sequence configuration class type in some or all tests if desired.

The text also introduces the concept of ultimate sequence flexibility, where sequence configuration parameters are redefined as an executable microarchitecture specification for the entire sequence. A signal generator class is defined and used to automate the creation of sequences, introducing random delays between signal transitions and using (min, max) parameter ranges to constrain delays.

The final approach is concluded to have been successfully used as part of a shared VIP, allowing for the creation of stable and flexible sequences using UVM configuration data base. The test writer can create targeted random tests without changing underlying sequences, and sequences/signals can share code using the signal generator class.

Key components include:

- Microarchitecture specification
- Sequence configuration class
- Factory for overriding sequence configuration class type
- Signal generator class
- Executable microarchitecture specification
- UVM configuration data base

The approach enables the creation of flexible and stable sequences, reduces the need for testbench changes, and improves the ability to absorb modified verification requirements.