Commentary: Why we don't have IP quality yet (by Larry Cooke, VSI)

Why we don't have IP quality yet
By Larry Cooke, EEdesign
July 24, 2003 (8:55 p.m. EST)
URL: http://www.eetimes.com/story/OEG20030724S0045

It is interesting to see how silicon intellectual property (IP) quality has recently become a hot topic in the IC industry. The Fabless Semiconductor Association recently decided IP quality was one of the top two technical issues their members need solutions for. Apparently, the IC industry didn't know that IP quality has been an issue that organizations like the Virtual Socket Interface Alliance (VSIA) and Virtual Component Exchange (VCX) have been working on for the better part of the last decade, so why all the fuss now?

The answer can be found in the quiet growth of the 3rd party IP industry. Seven years ago, when VSIA was founded, the founders predicted that there would be an explosive growth of 3rd party IP companies. That seemed to be the case for the next two years, but since then, the IP industry has appeared to be imploding. 3rd party IP companies were either going out of business or merging by the droves until today only a handful of independent "star IP" companies continue to exist, and most of the commodity IP is now available from the major EDA companies.

As a result, most observers thought, the 3rd party IP industry died. Quite the contrary! The overall IP industry has actually been growing at a low double-digit rate most of this time. In fact it has grown to the point where existing chip companies like Transmeda have switched to selling IP.

IP quality has now become an overall issue in the IC industry, because the predictions that organizations like VSIA made back in the mid 1990s have become reality. The combination of the convergence of electronic products, coupled with the relentless march to chips that can contain tens of millions of gates today, is making the integration of internal and 3rd party IP the only economically viable way to create the next generation chips. Now that the fabless semiconductor companies have woken up, they are clamoring for better IP quality. But the problem isn't as easy as these newcome rs think it is.

VSIA promoted use of their specifications — the required data and documents necessary to successfully transfer IP cores from developers to integrators — for years without success. First the IP providers were afraid they wouldn't meet the minimum requirements and refused to do it. Then, when compliance was watered down to disclosure only, they argued that their customers didn't want it, so nothing happened. Similarly, OpenMore, a highly touted measure of good design practices, initially met with great industry interest, but ultimately was only sporadically used by IP vendors to measure their IP, again because the customers didn't require it.

Ultimately the failure of these past efforts rests with the consumers of 3rd party IP, not the developers, and even today the fabless semiconductor companies want a way to demand quality without paying for it, and there's the problem. Measuring IP quality costs time and effort. Many of the large system and semiconductor companies have spent the last seven years creating in-house IP quality procedures, and a number of them claim it costs as much as 3 man-months to verify the quality of one single piece of IP.

Of course the IP companies aren't going absorb such a cost. They would go broke if they did. Furthermore, most internal IP verification is not done by the IP developers. That's like having the fox guard the hen house. So the industry will continue to pay lip service to IP quality, regardless of how important it is, until the IP consumers:

• are willing to agree on what constitutes IP quality,
• are willing to pay for measurement of IP quality, and
• are willing to support an on-going IP quality industry initiative.