New Standard!

What the IEEE 1801-2013 Unified Power Format Standard does, and what’s next.

popularity

It’s been a little over four years since the first IEEE 1801 standard was officially published in March 2009, but the standard can trace its roots back to years before that date. On May 30th, the IEEE released a press announcement for the newest version of the standard, IEEE 1801-2013 (a.k.a. UPF 2.1). It takes a considerable amount of effort and attention to detail to produce a solid standard and the working group committee should be congratulated on getting this new version out to the industry. The industry now (at last) has a power format standard that is backed by all of the major EDA vendors with Cadence’s recent first time announcement of support for the IEEE 1801™ standard (commonly referred to as UPF, Unified Power Format).

The IEEE also has made a free copy of the standard available for personal use. If you have any interest in this area at all, I encourage you to take a few moments and download the latest version.

Several members of the working group committee gave a four-hour workshop presentation on Sunday, June 2, at the 50th Design Automation Conference. The presenters included working group chairman John Biggs (ARM), co-chair Erich Marschner (Mentor), Qi Wang (Cadence), Jeffrey Lee (Synopsys) and Sushma Honnavara Prasad (Broadcom), with the workshop having been organized by Pete Hardee (Cadence). It looks as if Cadence is really serious about supporting this new version of the UPF standard which should be good for both the EDA and semiconductor industries. The presenters then participated in a follow-up panel discussion sponsored by Accellera and moderated by our editor, Ed Sperling, on the morning of Monday, June 3.

John, Erich and Sushma all stressed the advancements made in clarifying the standard and resolving ambiguities from the previous version. In addition to the normative section, there is a new informative annex that provides more examples and usage guidelines. Models for retention registers have been cleaned up and the simulation semantics simplified. Domain boundary definitions have been tightened up with clearer specifications for “lower boundaries” to aid in the handling of soft and hard IP. In addition to the new set_retention command, there’s a new set_repeater command to aid in AON (Always On) buffering.

Qi and others have mentioned the ability to attribute pins/cells in the library using “define” commands as an alternative to Liberty and a more structured-approach to macro cell modeling. Some of this is carry-over from the vast portions of the CPF format that Si2 donated to the IEEE for use in 1801. A number of UPF 1.0 commands also have been deprecated, which should help move engineers to better methodologies supported by the new standard and ease CPF users into using UPF quicker.

So, what’s next? It now looks as if attention will move to higher up in the design abstraction space with early modeling and estimation possibly being a key aspect of a future UPF (3.0) standard. In all, the hard work of the committee members in producing the new standard should be a boon to the industry.