There has been a lot of on-line discussion since DAC about whether the Universal Verification Methodology (UVM) is ready for prime time. My last three blog posts (1 - 2 - 3) addressed some of the reasons why I strongly believe that the UVM is ready and why you should consider it for your next verification project.
In the middle of my series, Mark Glasser from Mentor posted in his blog that "Eventually there will be a tipping point and UVM will become the obvious choice for a testbench methodology. That day is still in the distance." This was in direct contradiction to my statements that "I really believe that no one should have any reservations about moving forward with the UVM" and "I encourage [users to] make this seamless transition as soon as they can."
But then Ed Sperling referenced Mark's post and mine with the statement that the two blogs have "a similar perspective." Ed's a good guy, so if we're confusing him then we're probably confusing others. Let me clear the slate and attempt to shed some more light on this topic. Here are some key points:
- Cadence and Mentor have both pledged to support current Open Verification Methodology (OVM) users for as long as they continue to use the OVM. Most users will not transition methodologies in mid-project, so it will likely be 6-9 months before all teams migrate to the UVM.
- Transition from the OVM to the UVM is almost entirely painless; Accellera provides a conversion script that was well tested and is now being used in production by both vendors and end users.
- UVM 1.0 EA is solid and production-ready. Accellera has made a strong statement that only three new features should be considered in "early adopter" status. Everything else in the UVM is taken straight from the widely adopted OVM 2.1.1 release.
- In the same statement, Accellera also made a pledge for a high degree of backward compatibility between OVM 1.0 EA and future versions of the OVM. Mark seems worried that Accellera will not adhere to this commitment. But Accellera is us -- the vendors and users already relying on the UVM -- and it's up to us to ensure backward compatibility as we evolve the UVM.
- Cadence and Mentor both made unequivocal statements at DAC that all new development will be going into the UVM; no new releases of the OVM will happen.
- Ultimately, each project will make its own decision on when to migrate from the OVM to the UVM based on its own needs. That timing may depend on vendor VIP or tool support, although the whole OVM ecosystem is moving to UVM extremely rapidly so any concerns there won't be around for long.
- At DAC, I heard about multiple projects from multiple companies starting with the UVM, so I'm certain that there are at least a few dozen such projects already. The horse is out of the barn and running.
So here's the deal. I'm very proud of the impact that the OVM has had on the industry; in a small way the OVM team really did change the world. I love the OVM. My colleagues at Cadence love the OVM. Lots of other folks love the OVM. We're going to support it as long as needed.
But the UVM is the standard. The UVM is ready for prime time. The UVM is where all new technology will reside. The UVM is the future.
Visit UVM World and learn lots more.
The truth is out there...sometimes it's in a blog.