Synopsys Insight 

Insight Home   |   Next Article

Issue 3, 2011

Partner Highlight
Interoperable Process Design Kits (iPDKs) Gain Momentum

Many of the top-tier foundries have now embraced the benefits of interoperable process design kits. Jingwen Yuan, a strategic alliance manager at Synopsys and president of the IPL Alliance, outlines the benefits for foundries and design teams and brings us up-to-date with the latest developments.

The use of interoperable process design kits (iPDKs) is gaining ground. iPDKs are sets of foundry-verified data files (Figure 1) that include schematic symbols, component description formats (CDFs), callbacks and parameterized cells (PCells) for analog and mixed-signal design. The IPL (Interoperable PDK Libraries) Alliance, an industry-wide collaborative effort to create and promote standards for iPDKs, is now working with all the top-tier foundries.

The benefits of interoperability are compelling for both foundries and analog design teams. Without iPDKs, each foundry has to create multiple tool-specific PDKs for different EDA tools. The numbers are staggering. It is not unusual for a foundry to have to develop, distribute and maintain thousands of PDKs and technology files to support their processes.

In the past, the use of proprietary PDKs limited design reuse and prevented designers from adopting best-in-class tools. Proprietary formats can stifle EDA innovation, but the availability of iPDKs has changed that. One iPDK supports multiple OpenAccess EDA tool flows, so foundries can spend less time and money developing and maintaining design kits and invest more resources in advancing their process technologies.

Figure 1. Main components of a PDK with OpenAccess database
Figure 1: Main components of a PDK with OpenAccess database

From a process technology standpoint, iPDKs are much better able to support complex design flows and IP reuse, as well as foundries' advanced processes and complex parameterized layout cells (PCells).

Benefits for Design Teams
End users are enjoying the benefits of iPDKs today. Analog and custom IC design teams benefit from fast access to the latest design kits and tools because foundries can release iPDKs for their latest processes sooner, with immediate support for a broad range of design tool flows. iPDKs offer advanced functionality across multiple EDA vendor tools, improve design accuracy, shorten design cycle times, and promote design reuse. They are proving especially valuable for design teams using multiple-vendor flows, where the standardization of PDKs removes bottlenecks in the flow.

Design teams and foundries dedicate significant resources toward migrating their current IP to next-generation process technologies. IP that is based on an iPDK takes much less effort to migrate to a new process than IP that involves multiple technology files and tool flows. In addition, design teams and foundries that use the iPDK approach do not lock-up their critical IP in a proprietary set of formats. This makes it easier for them to share their key analog mixed-signal building blocks both internally and externally.

IPL's Approach
Since its inception in April 2007, IPL has focused on some key principles to guide its work. It uses existing standards wherever possible and aims to validate them with prototype iPDKs and flows before releasing open standards for industry adoption.

Traditionally, foundry PDKs are developed and validated by foundries before being distributed to customers. PDKs belong to foundries and are foundry intellectual property (IP). IPL works closely with foundries to understand the requirements for standardizing PDKs and to validate that the iPDK standard meets the foundries' needs.

In addition to its work with foundries, IPL brings together EDA vendors and the design community in order to ensure that its standardization efforts meet the needs of the complete ecosystem.

Process kits have many components and there are benefits to having more experts involved in the standardization process. However, a proliferation of multiple standards would be a step backward in the industry's efforts to streamline process kit development. To that end, IPL is currently working with the OpenPDK Coalition, an organization founded in 2010 by Si2, with the broad aim of defining open standards to enable greater efficiency in PDK development, verification and delivery. The current agreement between IPL and OpenPDK will see IPL's iPDK standard become a component of the OpenPDK's standard when it becomes available. The iPDK is available now and in use at many of the top-tier foundries, those who have invested in iPDK do not have to worry about the risk of any compatibility issues between iPDK and standards that the OpenPDK Coalition will announce in the future.

Technical Features
iPDK is based on the OpenAccess (OA) database and uses standard languages such as Tcl and Python to ensure interoperability among all EDA vendor tools. iPDKs include a comprehensive set of APIs to enable customization, support advanced PDK features and provide an interactive environment for PDK development.

PCells written in Python (PyCells) have significantly fewer lines of code and provide tremendous performance improvements compared to relative object-based SKILL PCells. PyCells support advanced features such as abutment, stretch handles and DFM rules. A free tool, PyCell Studio from Ciranova and associated IDE provides an interactive environment for PyCell development and efficient PyCell debugging. High-level Python APIs provide process porting capability within PDKs.

Because the interoperable PDK is based on OA, no stream-in or stream-out or any shape or data translation is needed for kits to work with the leading physical verification tools, which cuts down the effort required to validate PDKs.

Standardizing Constraints
iPDKs are helping to drive standardization, which benefits the entire industry. However, without a similar level of standardization for constraints, there is a danger that design teams could still get locked in to using proprietary formats.

IPL's latest efforts have focused on defining standards for design constraints. IPL Constraints 1.0 defines an open standard of the specifications for a set of design constraints, enabling design teams to enter constraints and design intent once, and use them across multiple EDA tools in their design flows, improving productivity and quality of results while reducing time to market.

IPL Constraints 1.0 defines both the syntax and schema for open, interoperable design constraints. Design constraints are design-specific and distinct from foundry constraints, which are based on semiconductor process design rules that are imposed by a foundry's manufacturing technology. Design teams can apply constraints to many different areas, including placement, routing, timing, clock, symmetry, matching, power, floorplanning, pin, electrical, and more.

Digital designers often characterize analog design as "black magic" because of its difficulty and complexity. Analog and custom design teams still do much of their design with less automation than their colleagues working in the digital domain. However, interoperable PDKs and interoperable design constraints give analog and custom design teams a significant productivity boost as they enable design teams to more easily reuse blocks and gain access to the latest tools, flows and technology processes.

Leading foundries around the world see the value in investing in iPDKs, which is why TSMC, TowerJazz, Dongbu HiTek and LFoundry have joined IPL, and GlobalFoundries has chosen to work with Synopsys to deliver iPDKs for its latest processes.

More Information

About the Author

Jingwen Yuan
Jingwen Yuan is a strategic alliance manager at Synopsys and president of the IPL Alliance. Jingwen works with foundry and EDA partners to create and promote interoperable PDKs. This is a collaborative industry effort to create PDK standards that accelerate innovation in custom and analog design. Jingwen began her career as an analog design engineer at MicroLinear (now RFMD). She then joined Cadence Spectrum service. She held various engineering, account management and marketing positions at Cadence and Oridus, an Internet collaboration start-up, before joining Synopsys.

Having read this article, will you take a moment to let us know how informative the article was to you.
Exceptionally informative (I emailed the article to a friend)
Very informative
Somewhat informative
Not at all informative