MiniTB is a simple yet powerful responsible development platform (RDP) that provides design and verification engineers with an alternative to complex verification methodologies like UVM.
That’s what should have been part of the initial announcement I made releasing MiniTB a few months ago. Originally, I envisioned it a smoke testing platform for RTL design engineers. Leaving it at that, though, I think underestimates the power of the simple platform Jean-Marc and I put together and the success people can create with it. MiniTB is not just a framework for smoke testing RTL, it’s a responsible development platform that addresses many of the concerns people have with current methodologies and techniques.
Complexity is always the first complaint people have when it comes to UVM. Second is how the addition of OO programming constructs to SystemVerilog has become the wedge that’s been driving design and verification engineers apart for the last decade. MiniTB intentionally stresses a level of simplicity and inclusiveness that have been optimized out of methodologies like UVM – slowly and deliberately.
MiniTB is powerful in that it gives you the flexibility to control your own destiny without locking you in. While it neither provides nor enforces complex methodologies around stimulus generation, configuration or communication, it also does not preclude you from leveraging existing methodologies or creating your own as you see fit. In short, MiniTB does not attempt to impose solutions upon functional verification engineers, it simply provides the framework and run-flow in which people are free – design and verification engineers alike – to create solutions that make sense to them.
If you’re a design or verification engineer that is tired of the complexity, try MiniTB.
-neil
Looks good. I have a feature suggestion – something that has extremely useful in title=”RTL verification using Python”>Cocotb:
Output a JUnit XML summary of test results for integration into Jenkins.
Coupled with auto-test discovery simply writing a new test is sufficient to automatically include it in the regression. Since tests are so quick to execute we can run a full regression every check-in. Almost instant feedback when you break something – great for agile development!