Low Power Blogs

A Look Behind the Si2 CPF 2.0 Release

The long awaited new version of the Common Power Format, CPF 2.0, was released by the Silicon Integration Initiative (Si2), an industry standards organization, today. Here are several interesting observations from this latest release.

First of all, this new release is a big step forward for interoperability between IEEE 1801 (Unified Power Format 2.0), the other industry power intent format, and CPF. In 2009 the format working group of the Si2 Low Power Coalition (LPC), which is responsible for maintaining CPF, released a document called Interoperability Guide for Power Format Standards, which describes clearly how to map between an IEEE 1801 command/option to a CPF counterpart. It also shows which commands/options of 1801 are declared as non-interoperable due to missing constructs in CPF and the methodology differences between the two formats.

As a result, the LPC format group quickly started to work on the extension of CPF to close the gaps identified by the interoperability guide. According to the Si2 press release, the major improvements in this area include the following, among many others:

  • The generic mode concept, which combines both power mode and functional mode for a more comprehensive and robust power intent specification.
  • More flexible isolation and level shifter rule specification.
  • Link with the Liberty format for global supply net connection based on pg_type attributes.

The release also has many important and useful extensions based on the requirements from the member companies of the working group. For example, it introduces a power design concept to enhance the hierarchal design flow. Also, a new extension of simulation control adds significant flexibility for the user control of simulation corruption semantics, especially on the non-synthesizable behavioral constructs such as initial statements in Verilog.

In May 2010, after the Si2 LPC's call for technology contribution, Cadence made a contribution to the extensions. Some of them were accepted by the format working group. The contribution includes the enhancement of macro-models to improve their usability for I/O pads, and support for analog/mixed-signal IP with power management features. The Cadence contribution also includes extensions to support some new types of low-power IP such as a special clamp cell, power and ground level shifter, complex global cells, and isolation/level shifter cells that can be placed anywhere in the design.

The CPF 2.0 specification acknowledges LPC Format working group members, including representatives from Synopsys, Calypto, AMD, IBM, LSI, Si2, and Cadence. Clearly, this shows the strong interest across the industry to drive the CPF standard forward.

Cadence has been an active member of the format working group with a strong commitment to supporting CPF in the tools and flows of the Cadence Low Power Solution. The new CPF extension delivers the language support requested by many of our customers who had successfully adopted a CPF based low power design methodology. As a result, Cadence is committed to work with our key technology partners to enhance our tools and flows based on the CPF 2.0 release based on the priority requested by our customers.

In the next a few months we will run a series of blogs to dive into some specific CPF 2.0 enhancements and to show how it can benefit designers.

Pete Hardee

 


Share Your Comment

  • You're welcome Karen, pleased we got that straight!

    Pete

  • Anonymous
    Anonymous

    Pete,

    I'd like to clarify our participation. A Virage employee (now a Synopsys employee) was a participant on the LPC. During the transition period as Virage merged with Synopsys, a Synopsys employee contributed some verbal comments (none substantive) during some LPC calls, but did not vote on the CPF 2.0 specification.

    Thanks for the opportunity to make everything clear,

    Karen

  • Hi Karen,

    Thanks for your comment, but I must admit to being puzzled. The Si2 CPF 2.0 specification lists two members of the format working group affiliated to Synopsys on page 4. Please check it out by downloading it at http://www.si2.org/?page=811. I also understand that the specification was approved by representatives of all the affiliations listed there.

    Pete

  • Anonymous
    Anonymous

    Pete,

    Please be aware and let your readers know that Synopsys was *not* a member of the LPC. We contributed to UPF/IEEE 1801 but not to CPF. It's not correct to include us in the list of representatives.

    Thank you,

    Karen

Add Yours

Post (Login required)