Why I Like UVM Express

[nosnhojn]

Here's a few of my thoughts on UVM Express that Mentor announced last week and how I think it could work very well for teams that are interested in doing incremental development.

http://www.agilesoc.com/2012/02/24/why-i-like-uvm-express/

-neil

[SeanChou]

This idea seems good in lowering the entry barrier for UVM beginners. Are you one of them? Could you share which stage you have been and what is your suggestion to adopt it?

[nosnhojn]

I was a vmm expert turned uvm beginner. after seeing people struggle with vmm and knowing uvm is... well... bigger, the barrier I was expecting and indeed found was pure information overload. I don't think there's a better way to put it. uvm is big and overwhelming. imho, uvm on it's own does nothing to encourage focused learning/ramp-up. Not to say you can't have focused learning/ramp-up, just that the library itself doesn't make how to do that immediately apparent. there's also the fact that it includes a lot of optimizations under the hood, global methods/properties and macros (which I assume are the applied lessons learned through use of it's predecessors) which makes it difficult for beginners to trace control and see what's going on. In short, uvm is big, powerful and hard to understand.

to uvm express now... I think this is great for beginners. It suggests a ramp up path and starts from a very basic point (an interface bfm). It also suggests a way to focus on some pieces while temporarily disregarding others (look at functional coverage and leave random stimulus and sequences to later). I also think this is great for *experts* even though it seems mentor is suggesting otherwise. I think uvm express is a good mentoring tool for experts that could help close the gap between novice and expert verifiers. imho it also provides a better way to communicate and share code with designers (going back to the interface bfm again). of course it's also an idea which means you could apply it to any other library or use it with any other toolset - ideas are so much more portable than code - so anyone can apply it.

I'd now consider myself intermediate because I've used uvm a bit and spent several hours getting to know the phasing and component control. I also already knew about tlm so the communications mechanisms were familiar. I'm to the point where I can figure out new features/components as I need them. Starting again though, I'd seriously consider the route suggested in uvm express. I think it'll improve uvm from an accessibility standpoint and I hope there's other ideas like this in the pipe.

...or maybe there already are other ideas out there that I'm missing??

-neil