Gabe's Commentary: Accellera and the SystemVerilog standards process
By Gabe Moretti, EDA editor, EDN magazine -- EDN, June 10, 2004
Well-meaning attempts to forget some less-than-pleasant events of the past three years are unfortunately creating more confusion and uncertainty within the EDA community. This may in fact already have created more harm than the damage of telling the entire story can possibly have done or do now. Therefore I decided to take the responsibility to put current events in the perspective of recent history.
The controversy surrounding the channel chosen by the Board of Directors of Accellera to deliver SystemVerilog 3.1a to the IEEE for national and international standardization has its roots in events that have not been published to my knowledge. I think that once the EDA industry knows and understands want has really gone on, the Accellera choice will be much clearer.
I must begin by explaining what went on in the IEEE Standards Association. Until 2003 the IEEE Standards Association had only one procedure that could be followed in the approval of a standard proposal: individual voting by members of the IEEE Standards Association. Although other opportunities existed, there was also only one channel that outside entities traditionally used to initiate work to develop a new EDA standard within the IEEE. The channel was the Design Automation Standards Committee (DASC) of the Computer Society.
In 2003 the IEEE Standards Association formed the Corporate Advisory Group (CAG) in recognition of the contributions that Corporations make to the development of the profession and the development of standards. Many leading companies, both in the US and from Asia and Europe, have now joined this group. Membership is not just open to Corporations, but universities, industry consortia, and governmental entities are also welcomed to join the CAG.
In late 2003 Accellera joined the CAG because it provided a way to increase its ability to receive corporate input, raise its profile as the most effective incubator of new EDA standards, and potentially increase its membership. The CAG is an official standards sponsoring organization within the IEEE Standards Association, meaning that outside entities, like Accellera, now have an additional channel that can be used to propose a standard to the IEEE. With the formation of the CAG, the IEEE Standards Association also revised its standards approval voting procedures and accepted entity (read corporate) voting in addition to individual voting. A major reason behind this change is the realization that successful standards are born from commercial interests, and that corporations and consortia are the best exponents of such interests.
The recession that hit the electronic industry and thus EDA in 2001 had a very detrimental effect on the DASC. Its chair lost his job and his funding to do the DASC work, and the vice-chair was too busy attending to the financial interests of his small company to provide the required leadership. In a short period of time the DASC became an organization with no supervision and practically no control. This resulted in a significant lowering of the effectiveness and efficiency of its Working Groups (WG), and in the deterioration of the quality of communication between the DASC and outside organizations like Accellera.
One of the results was that the coordination of the work between the 1364 WG that was developing the next generation Verilog specification, and the Accellera SystemVerilog Technical Committee did not have a proper structural liaison. Individual technical members, working on both projects, like Stu Sutherland, have tried valiantly to keep the two organizations synchronized, but he and a few other supporters lacked the commercial and organizational capabilities to guarantee success.
With the recovery came new leadership for the DASC following election in the summer of 2003. Peter Ashenden, has made significant strides in developing and adopting new working rules, but as with most large and diverse entities, change does not come overnight. Since the DASC is still putting its own house in order, and since the CAG influence in the IEEE Standards Association is growing, and since Accellera has always had an entity voting mechanism, it was prudent and responsible for its Board of Directors to vote to use the CAG as the sponsor for the SystemVerilog standardization. Although the DASC operating procedures continue to improve, they are likely only to be adopted later on Friday June 11, two weeks before the SystemVerilog project will be approved by the IEEE. Therefore the prudent and responsible choice was to use a channel that is robust, growing in importance, and has a track record of efficient and timely standards approval.
Much has been said about the vote cast by Cadence opposing the choice of the Accellera board. I believe that two factors, neither of them malicious, are at the root of their choice. During most of 2002 and all of 2003, Cadence had no one on staff that had both the charter and the expertise to advise the company on a strategy related to standards development. This has been remedied recently by re-hiring Victor Berman. Victor has outstanding credentials as a respected member of a number of past DASC WGs, and his knowledge of modeling languages cannot be questioned. Victor became the Cadence representative on the Accellera board, and also the Vice Chair of DASC. That he was voted vice-chair practically immediately after becoming once again active in EDA is an indication of the lack of leadership that existed in the DASC.
Victor, therefore, has a personal interest, well-motivated as it may be, in seeing that the DASC remains the only channel used by Accellera in its dealings with the IEEE, and Cadence has commercial interests in making sure that the adoption of SystemVerilog is both successful and deliberately slower than what both Synopsys and Mentor can achieve given the efficiency with which the CAG can operate and the quality of the SystemVerilog 3.1a document.
It is important to acknowledge that developing a standard takes money and that many of the standards developed by the DASC received significant financial help from VI, OVI, and Accellera. The CAG has a fund raising mechanism that is built-in, and thus can quickly muster the resources required to insure that a standard is developed in the most effective possible manner.
It is easy to second-guess the Board of Directors of Accellera without knowing all of the reasons that motivated its choice. It is always preferable not to gratuitously display dirty laundry for the neighbors to see. But I believe that it is proper, when the EDA community shows so much confusion over such a critical issue as SystemVerilog, to tell the entire story in the most positive way I know how.
