Editorial

Spending the Evening with "ALF" Helping to develop a standard is worth giving up a good meal. by Jonah McLeod

I made it through Wednesday at DAC as everyone else did and was looking forward to an elegant dinner at Mr. Stox Restaurant in Anaheim, Calif. Compass Design Automation was the host of this undisguised attempt to curry favor with the press. For my part, I was intent on letting them have their way with me. However, I got an eleventh-hour phone call from Georgia Marszalek, a representative of Open Verilog International (OVI), asking if I would moderate a "Birds of a Feather" meeting on "ALF."

I immediately thought of the obnoxious outer-space character in the inane sitcom of the same name, but Georgia informed me that the acronym meant "advanced library format." I was interested in the topic, thinking it might provide the meat of an editorial, so somewhat reluctantly, I agreed, deluding myself into believing I'd be free just as the gang at Mr. Stox was finishing cocktails and filing in for dinner.

David Burrows, group vice president at Viewlogic and vice chairman of OVI, introduced the topic to the interested attendees. Burrows explained that ALF is supposed to supplement the two existing file format standards, SDF (Standard Delay Format) and DCL (Delay Calculation Language). ALF aims to provide the ASIC library developer with one standard format for specifying parameters of ASIC library elements for synthesis, timing, power, and test analysis tools.

Wolfgang Roethig, a design engineer at LSI Logic, made the complete ALF presentation, pointing out the benefits of adopting the format: saving ASIC vendors considerable time, effort, and money. Today, Roethig said, it takes 60 people four months to develop a library. He predicted that ALF would save library developers 10 person-years of effort for each library.

Tool vendors would also benefit from ALF, he said. The format allows ASIC vendors to incorporate much more information about every library element than is currently provided--precise timing information instead of unit delays, accurate power consumption for each element, and the like. Aggressive tool suppliers will be able to make use of this information to give users an accurate model of each library element for design creation and verification.

The picture, though, is not entirely as it appears. One attendee pointed out that Synopsys was not actively supporting ALF. If the company chose to open up access to its tools, it could make ALF redundant. Another pointed out that DCL may not be entirely compatible with ALF. A third noted that ALF has a programming interface that needs a parser (currently under development).

An engineer from an ASIC vendor stressed a further drawback: ALF is yet another format he must support, thus increasing his work initially without any guarantees that it will eliminate any of the other formats currently being supported. Another attendee from an EDA tool vendor echoed those remarks. He added that being a programming language, DCL can accomplish much of what ALF proposes to do.

As with most standards presentations, nothing was resolved, while areas of contention were defined more clearly. The proceedings carried on well beyond the time that my associates were finishing their cocktails and filing in for dinner. I reconciled myself that I had been part of a worthwhile process while the gang at Mr. Stox chose rack of lamb, swordfish, or culotte steak for dinner, accompanied by a selection of four different California wines (two whites and two reds), followed by brandy or cognac and cigars.

However, I made the right choice--right?

To voice an opinion on this or any Integrated System Design article, please e-mail your message to miker@isdmag.com.

integrated system design  August 1997