Design For Configurability

Just because you can build great IP doesn’t mean you’ll be able to re-use it.

popularity

I admit it was a bit of a surprise to me to hear from a leading IP provider of the missteps that still befall design teams today as they seek to reuse IP, but it’s a little like rubbernecking. How do you not look?

According to Grant Martin, chief scientist at Tensilica, “The biggest thing that people still don’t think about at the beginning of designing some new function is designing it to be configurable and designing it to indeed be able to move across different processes and different use models in different kinds of devices.”

“An IP vendor, and especially a configurable and extensible one like us, designs all of that configurability in and builds the process for specifying and generating IP. We do all that up front because that is the lifeblood of our business. It’s absolutely essential for us to do it and very often, even though we and other types of configurable IP represent a model, there are still design groups that don’t seem to look at those other models when they start off on a project. They find that their own blocks may not be as portable and reusable or have the right level of configurability at the end,” he continued.

Martin added that if you do think about it at the beginning and if you concentrate on the processor generating IP, you can get substantial savings in terms of the amount of effort to reuse IP across devices and across processes and generations

As a design engineer, what is your experience with building IP for reuse?

~Ann Steffora Mutschler