A few words about standards

But, one may ask, if standards are good, why worry about how they got developed or how they are managed? Because if the process is not transparent and inclusive, one large company can in fact dominate the creation and contents of a standard. And what about management of standards? If not done properly, a standard could be arbitrarily modified or could fall into neglect.

EDA standards development: the past
In the pioneering days, yes I am talking about the '70s, there was little need for standardization, with the exception of schematic drawing interfaces. Thus the most famous standard, EDIF, was developed by a committee whose membership included both EDA vendors and users under the ANSI umbrella. The process worked well, most of the committee members learned a lot, and , as can be seen by its longevity, EDIF has been a success.

Then came the '80s and early 90's. Those were the maturing years, when the IEEE formed the DASC (Design Automation Standards Committee) and put in place the process, policies, and oversights very similar to those now asked by IBM of every standards body. VHDL is the most famous standard developed in those days. Cadence (why is it that they always need to be different?), went its own way and declared Verilog to be a standard, although it took Venk Shukla to see the error and form Open Verilog International (OVI) to give the language at least a flimsy appearance of openness. OVI would fail the present IBM test because the standard was controlled by Cadence and the rest of the OVI members were in fact not much more than a users group. Of course OVI was not alone in mudding the standards waters. VHDL International (VI) contributed to the confusion by forming as a VHDL promoter organization, which some people confused as a standard making organization because its members did work to justify the creation of other standards in support of VHDL. VI would also fail the present IBM test, because it did not have the mechanism to create and maintain a standard. But VI and OVI were a very good prototype for the organizations that today are responsible for the majority of EDA standards: the feeder organizations.

The present
The IEEE has taken a strong central role in the development of standards, both for EDA and other electronics industries, and the vast majority of EDA standards today have either been developed within the IEEE or are managed and maintained by the IEEE. As I said before, all standards developed within the IEEE do conform to the IBM requirements.

But only a small part of EDA standards are fully developed within the DASC. In most cases, another organization, called a feeder organization, identifies the need and begins the work to develop a new standard. Once the work is mature enough to produce a document that describes a proposed standard, all rights are offered to the DASC or other pertinent IEEE bodies. If the IEEE determines that the proposed standard is needed it will then accept the work and proceed to refined it, insure openness, and proceed to balloting, and if approved, management of the standard. Many feeder organizations are US based, although the European electronic Chips and Systems design Initiative (ECSI) in Europe and Japan's Electronics and Information Technologies Industries Association (JEITA) EDA/TC, are also instrumental in standards work.

When I think of organizations who develop proposed standards and then pass them on to the IEEE for final standardization and maintenance, I think of Accellera, OSCI, and SPIRIT. Accellera was formed almost ten years ago by merging OVI and VI and expanding the mission to include not just hardware description languages, but all aspects of EDA design. OSCI and SPIRIT are younger than Accellera, but have practically the same procedures when it comes to standards development. All three develop standards which, in most cases are then offered to the IEEE for final standardization and for maintenance. As feeder organizations they must maintain the inclusiveness, openness, and governance rules of the IEEE in order to have their standards accepted. The IEEE, in turn, works with both ANSI and the International Electrotechnical Commission (IEC) to market and maintain the standard worldwide. Most people in the EDA industry do not know that in many countries you cannot call something a standard unless it is an IEC standard.

Some of you might have noticed the SI2 is not on the list of IEEE feeder organizations. This is not an accident, it is by design. Although and SI2 standard could become an IEEE standard, the process would be different than in the case of the other three organizations mentioned before. The reason is that the process of developing and maintaining an SI2 standard is not open in the IEEE, and IBM, semantics.

The fundamental problem is that since SI2 derives income from its standards, like OpenAccess for example, it maintains the ownership of copyrights and trademarks of its standards and controls and maintains the standard according to its own internal rules which are not compliant with the rules of the IEEE, the IEC, ANSI, or, for that matter, the IBM requirements just made public. In its procedures, SI2 is not malicious or wrong: just different. It serves the interest of its member companies, as it should. In the case of OpenAccess, it is in the interest of Cadence to insure that changes to that standards are approved by Cadence, or it could find itself in the awkward position of having the majority of its tools become not compliant with its own standard. For this reason, SI2 standards are to be considered "de facto standard" or proprietary conventions. But they cannot be called standards since they lack the openness and independent oversight of IEEE, ANSI, or IEC standards.

The EDA industry should welcome the IBM posture since it will strengthen both the development and the maintenance process of its standards. The position of IBM clarifies, not modifies, its method by which it will contribute to various industries in the future.