Low Power Blogs

Low power design was undoubtedly one of the themes of DAC this year -- especially at the Cadence booth.

We drew lively interest on the DAC floor with our low power demo station, which was continuously busy especially on the free Monday. We were showing a new demo explaining how advanced low power techniques ("advanced" means the kind of techniques you need a power intent format to describe -- more on this later) tends to introduce two kinds of problems we see people struggle with:

• How can I get predictability early in the flow to decide which low power techniques will actually meet the power specification?

• Low power techniques introduce a whole new level of complexity, so how can I implement those techniques and verify that I did them correctly, ensuring no bug escapes?

The demo neatly showed how a comprehensive Cadence solution, in this case highlighting Incisive Enterprise Simulator, RTL Compiler, Conformal Low Power and Encounter Power System, solves these twin issues. The demo sparked lively conversations with a good sample of San Diego and Southern California's rich population of power-aware designers, as well as dedicated travelers from further afield.

Meanwhile, in the suites, I asked the attendees in each low power session how many were already using a power intent format, either UPF (Unified Power Format) or CPF (Common Power Format), as a gauge of who was already using advanced low power techniques. Typically, about half the room would put their hands up. This led to two further trends:

• On asking those with their hands up how many were using both UPF and CPF, almost all the hands stayed up! Strong confirmation of the need for format interoperability and methodology convergence, which just happened to be one of the subjects on which we presented.

• For those that didn't put their hands up, I confirmed the trend that many designers not in the traditional mobile applications are realizing that their next designs are likely to need advanced low power techniques anyway, either because they are moving to latest process nodes where leakage is an issue, or because they need to control the thermal profile for cost and reliability reasons, in the face of increasing performance of their devices.

In the Cadence Theater on Tuesday, one low-power related presentation drew great interest - Sorin Dobre of Qualcomm presented his thoughts on power formats entitled "Power Intent Methodology Convergence: A Case Study of Hierarchical Flow". Sorin is one of the engineers who would raise his hand to using both CPF and UPF, and his presentation described the need for methodology convergence for power intent formats, and especially the need to extend IEEE 1801 (UPF 2.0) with CPF's hierarchy capabilities. Which is quite fitting, since one week earlier, Si2 announced the contribution of relevant parts of the CPF specification to IEEE 1801 with a view to achieve exactly that.

On Wednesday, the final day for the exhibition floor, with the agreement of Texas Instruments India, I presented a paper in the Cadence Theater on their behalf, first presented at CDNLive! India, detailing the work of Rakesh Hariharan, Prabhu Bhairi, and Nithin Maiya, in using Palladium to speed up RTL power-aware verification by up to 1000x. That also drew a sizable, engaged crowd.

In these busy three days, San Diego lived up to its reputation as a hot-bed of advanced low power design. In forthcoming blogs, look for further explanation of the need for, and issues with, methodology convergence as we drive that issue forward with the standards groups.

Pete Hardee