Perspectives on Developing EDA Standards

Last week, my colleague Paul McLellan published a blog post on the standardization work underway within the Portable Stimulus Working Group (PSWG) of Accellera Systems Initiative. He did a nice job of describing portable stimulus as the PSWG defines it and showing the broad scope of users and verification platforms that the standard proposes to address. It should be clear that this is far from a simple problem and no surprise that the development of a robust and usable standard is a multi-year effort.

Since I am often asked “what is the status of the portable stimulus standard?” and “why is it taking so long?” I would like to share some of my personal thoughts based on my experience helping to develop various standards over the years. Although I am secretary of the PSWG, let me be clear that I am speaking neither for the group not for Accellera overall in this post.

I’ll start by establishing my credentials. I began working on industry interconnect standards back in my days of developing silicon IP. At one point I was on the Board of Directors for the 1394 Trade Association, a member of the Steering Committee for the PCI Special Interest Group (PCI-SIG), and a regular participant at “plugfests” for the then-new USB standard. I was also active for several years in the working group that developed the IEEE 1500 Standard for Embedded Core Test.

My deep involvement in the early days of the commercial IP business also led me to chair the Functional Verification Working Group of the Virtual Socket Interface Alliance (VSIA) and to serve as an alternate representative on the board of the Virtual Component Exchange (VCX). As I moved into the EDA world, I spent some time in the Accellera subcommittee defining assertions for the SystemVerilog standard. I was also very active in my previous role at Cadence in developing the Open Verification Methodology (OVM) de facto standard although only indirectly involved in its transformation to the Universal Verification Methodology (UVM) by Accellera.

What have I concluded from all this experience in various standards organizations? Well, mainly that good standards take time. Working groups are typically a mix of users and vendors who have strong technical opinions and passionate interest in the topics being discussed. A few standards have essentially rubber-stamped existing technologies, but most are the result of merging and aligning ideas from multiple proposals. This must be done carefully and with technical due diligence to avoid Frankenstein standards composed of bits and pieces that don’t quite fit.

I’ve observed that politics almost always plays a role. Most members of working groups are employees of companies that may have a lot at stake in the outcome of the standards effort. For example, companies that have submitted proposals are often motivated to steer standardization their way. Members occasionally have to vote following corporate guidance rather than their own technical preferences. The flip side, of course, is that these companies are providing significant resources (their employees and perhaps other aid as well) to help drive standards. Their support is critical, even if it sometimes comes with strings attached.

Finally, I have learned that the best working groups have broad representation from everyone who will be stakeholders in the new standards. That doesn’t necessarily mean a huge number of people; too many participants can bog the meetings down. Having a few qualified, dedicated members from a few vendor and user companies may be enough as long as the full scope of the proposed standards is represented.

So where does the PSWG stack up compared with my previous experience? Certainly the group has a good mix of companies represented. All three major EDA vendors and several smaller ones have been active. User companies range from silicon to systems, including some of the largest electronics companies in the world. However, not all of these members are as active or as vocal as they could be. As ADI representative Dave Brownell has noted, additional users are always welcome since they will ultimately be the consumers of the portable stimulus standard.

I don’t believe in revealing the inner workings of standards bodies, so I will just note that of course politics has played some role in some aspects of the PSWG. As shown by the recent update tutorials at DVCon Europe and DVCon India, the group has worked through these issues to make a lot of progress toward a standard. It’s not out yet, and perhaps it is taking longer than some would like. As I said earlier, good standards take time.

I can imagine ways by which the group could have forced convergence more quickly, but I believe that would have happened only by suppressing good ideas and at the expense of the best possible standard. We at Cadence will continue to devote several of our best experts to the PSWG and keep you posted on the progress. Once again, I invite you to join the group if you have the time and experience to participate.

Tom A.

The truth is out there...sometimes it's in a blog.