The Power Of Standards

The semiconductor industry only needs one power format, so why are there still two?


By Barry Pangrle
It’s often said that the wonderful thing about standards is that there are so many to choose from. As an industry, EDA seems to have a short memory as VMM and OVM (now becoming UVM), VHDL and Verilog, and more recently UPF and CPF. In cases where one standard suffices, it is horribly inefficient to create multiple “standards.” It is a waste of effort and resources for EDA companies that have to maintain multiple versions of code to perform the same functionality, and it is inefficient for customers because IP shared among teams or acquired from other companies through mergers or acquisitions may use different formats. If one format doesn’t eventually prevail, the whole industry pays for the added inefficiency.

How do we end up with multiple standards in EDA? Typically, someone believes they will be able to hold a market captive by creating their own proprietary standard. Customers like choice, and given that some of the semiconductor companies that are customers of the EDA industry have annual revenues that are a multiple of the whole EDA industry’s annual revenue, customers have the resources to actively work to prevent one vendor from completely controlling a format specification. In regards, to UPF and CPF, I’ve had one customer describe them as the same in every intent but different in every detail.

UPF was started through Accelera and UPF 2.0 (also known as IEEE Standard 1801-2009) is a newer and updated version that was approved by the IEEE a little over a year ago now. Having the backing of the IEEE, it has quickly gained momentum in the industry with two of the three largest EDA companies backing it and reports from customers that the third company has agreed to also support UPF. At this point there is no compelling reason for supporting another standard besides IEEE 1801 and several strong reasons for not supporting another standard for power. For the sake of the industry, let’s hope that the IEEE standard quickly prevails and that customers will demand that all relevant vendors support it sooner rather than later.

—Barry Pangrle is a solutions architect for low power design and verification at Mentor Graphics