Once Cadence realized that it had used the IEEE channel in an improper manner in order to impose its Common Power Format (CPF) on the rest of the industry, the best course of action would have been to reset and use the IEEE channel properly. Instead it has made a mess of things that will either result in multiple "standard" formats, or in a much longer than anticipated progress toward a unified standard that will require the harmonization of at least three separate standardization initiatives.
What continues to amaze me is that after thirty-five years since the beginning of the standardization effort in the EDA industry with the EDIF format, professionals in our industry continue to misunderstand and abuse the standardization process. And this is not due to naivet; it is due to greed. While both history and common sense show that a good standard benefits all but benefits its original inventor most, companies, and in this case a consortium, try to optimize their short term revenues at the expense of the electronic design community. The two culprit in what seems destined to become the latest EDA soap opera are Cadence and Si2.
Cadence is understandably proud of the technical work done by its technologists in developing the CPF format and methods. As such it believes, with the pride only a mother could feel, that its baby is perfect and any alterations would weaken the format and slow its adoption. Thus it proposed that CPF be adopted by the IEEE without modifications and on the fastest track possible. Victor Berman, who is employed by Cadence and is the present Chair of the IEEE Design Automation Standard Committee (DASC) has great experience in the development of standards within that organization. He should have injected a required dose of realism into the Cadence plan. But, for whatever reason, this did not happen. The result was a major confrontation that almost killed the DASC and resulted in both its Secretary and Treasurer to be replaced, although they were personally blameless for the mess. Hide the mess by executing the innocents is a well-known action plan when things really look unsolvable. The result is that the DASC CPF effort is ongoing and could, in due time, result in an IEEE standard.
Since it is now obvious that there will not be a power standard from the IEEE in 2007 for procedural reasons, Cadence has now turned to its most friendly "independent" marketing arm: Si2. You can read the details in this article. To accommodate Cadence, Si2 has formed the Low Power Coalition (LPC) project to provide a harbor for CPF after January 31, 2007. Since Cadence already had the Power Forward Initiative (PFI) coalition organized and working, as reviewers of CPF, it is unclear at best why they would need another organization, unless they felt that the industry is now considering PFI as nothing more than its captive audience whose sole mission is to rubber stamp the format. The bottom line is that Cadence has now three active initiatives all aimed to make CPF a standard, either through its own consortium PFI, an Si2 project LPC, or the IEEE DASC. I know that the latter will use the technological knowledge of the participants to the working group to develop the standard, but I am not sure of the degree of independence the other two efforts will employ.
Si2, in fact, has been so connected to Cadence's agenda that an analyst from outside the EDA industry recently called me for advice on the power standardization mess and said that he knew of Si2 that, he stated, "was an industry organization founded by Cadence".
The primary goal of Steve Schultz, President and CEO of Si2, is to raise funds and increase the membership roster of the organization. Developing technology and possibly standards is, in his mind, the result of a larger budget and membership. One way to increase the amount of funds available to SI2 is to start a project. To be a project member, a company pays dues that are in addition to its annual membership dues, thus increasing the amount of money available to Si2. So it is understandable that Si2 was more than willing to start a new project: the LPC.
But, those who follow the EDA industry closely would ask, didn't Si2 join with Accellera to start the Unified Power Format (UPF) Technical Committee? Yes it did. So, why would Si2 start a project when it was already a founding member of another project with the same goal? Follow the money, as all good TV sleuths say. In addition, of course, LPC gives Si2 the freedom to follow Cadence's directions, at least until members like Magma have the ability to offer technical modifications and extensions to CPF after February 1, 2007. And, associating with Accellera, a consortium with a proven record of developing robust EDA standards, could only help SI2 self-promotion campaign.
As you would expect, most thinking people within EDA, have raised serious questions about LPC. You can read about it in this article. It is obvious that LPC has failed in its goal to rally the industry behind one effort under the auspices of an independent consortium. There are two fundamental reasons for this. The most obvious is that Cadence continues to prohibit any outside entity to modify the CPF document in any way until after January 31, 2007. Therefore there is no reason for any other EDA vendor or for any user company to be a member of an organization whose aim is to produce a standard until they will be allowed to actually work on one. The second reason is that Si2 is not a standard making organization. Si2's main claim to fame, in fact, is marketing the Open Access standard proposed by Cadence. The adoption of this standard has been much slower than expected due to lack of underlying technology and a few aspects of the data base that continue to be Cadence's proprietary, like skil based pcells.
The reasons Cadence is not contributing CPF to Accellera's UPF are twofold. Firstly Accellera requires the donation of an open document, something Cadence is not willing to do at least until February 1, 2007. Secondly, Accellera's Technical Committees are open to all interested parties and do not require any membership fees in order to contribute technically. It is true that only Accellera's members can vote on the proposed standard, but this does not preclude any company from contributing technically and if they see fit, join the organization and cast a vote. What seems clear is that the present situation is still unstable and that the creation of the LPC project solved nothing.