Enterprise Design Management Engineered for SoCs

In my initial look at ClioSoft’s design management system created from the ground up for the semiconductor industry, I made the opening case for managing and reusing IP across an ASIC design organization. Let’s for a moment say we agree on the need for an enterprise software package to do design management (we’ll dive into that idea more in our next installment). Why has ClioSoft chosen to develop their approach, and how does that help you?

First, we should focus on the operative word in the above paragraph: enterprise. Teams used to write their own configuration management tools, now an archaic practice. There are plenty of open source revision control systems out there that, at least in theory and with some modification, could handle some of the aspects of hardware IP design management for a small design team. I’m sure some teams have tried this approach, grabbing open source tools and making changes to suit specific needs. I’m also sure that as the needs grow, that idea looks less and less attractive – not because of the tool itself, but because of the level of effort around modifying and maintaining it. From my last post:

“A software configuration management tool such as Subversion [or Perforce] handles source code and metadata needed for compilation in a high-level language such as C or Java. What these tools lack is any knowledge of a hardware design flow or the ability to handle multiple file formats (likely with different EDA tools) and relationships involved when working with hardware IP. Extending these software-centric tools to integrate with typical EDA tools would be a massive undertaking.”

Ah, but that is the premise of open source. It’s meant as a basic platform for value-add. Some software vendor out there has already grabbed Perforce or Subversion, made the modifications to the database, and created what looks like a commercial hardware design management tool. They may have even done some integration work with various EDA design flows. The problem with such an approach is that when you support features of multiple software-based revision control systems, you need to support the different features of all of them. This makes the layer built on top of these software packages thin, and quality as well as robustness becomes questionable.

Comparing these efforts to the enterprise-ready alternative is kind of like comparing MySQL to Oracle Database 12c. Both are databases. Both can be queried dynamically. Both can be modified. One is free and the other isn’t. However, only one is designed for and trusted to handle enterprise-level distributed data and application flow, tuned for performance and scale and usability.

ClioSoft has spent the effort over the past seventeen years to create a robust, easy to use, enterprise-level SoC design and IP management application, not just a database to store stuff. They have optimized SOS, their design management solution, to meet the real needs of ASIC teams working with a range of IP and EDA tools. They own the entire ecosystem and can rapidly turn maintenance requests and feature modifications, as opposed to possibly having to wait on an open source community working with an underlying software configuration management database and tool to review, code, and approve changes. With a single system you can control your own destiny and create features backed by the proper infrastructure to ensure robustness and ease-of-use. It’s an Apple-like approach, with ecosystem ownership applied to design management tools.

There is more to the picture than functionality. ClioSoft has also engineered its performance with attention to details. Caching is smart, and network usage has been optimized. Large hierarchical objects are represented as composite objects for easy viewing and access. Instead of broadside copies of entire work areas created by the software-based revision control systems, SOS creates smart links – a huge advantage avoiding the performance hit of copying big blocks of design data. In addition, speed counts in EDA. With its native multi-stream hyper transport technology, ClioSoft’s SOS delivers that speed. It’s easy to see how a tool not designed for hardware design/IP management in EDA could quickly run into very ugly performance and storage issues.

In most organizations, configuration management data outlives the designers. (Anyone remember these “data cartridges” from the Unix era, or even more antiquated formats?) Large companies tend to want control of their process and don’t want to be locked into a proprietary commercial vendor. Again, one can hear the open source siren in the background. Here’s the reality: when you’ve made any modifications to a database, you’re locked in. When that day comes that you have to get the data out and into another tool – and it will – what happens? ClioSoft allows data export with a read-only license to provide confidence that hardware IP management data moves forward if needs change.

Isn’t it overkill? Not really – let’s consider the bigger picture. More and more IP blocks comprise an SoC today. Those blocks are coming from many different sources, with many different locations where human capital resides. IP can be analog, digital, RF, or mixed-signal. Collaboration with increased efficiency in creating IP is a must. Even a small team is likely to track and exchange hardware IP design data amongst themselves or with contractors, third party IP houses, foundries, software providers, and other parts of an ecosystem that could be anywhere on the globe. A small team might need to coordinate with external design teams, such as in working with customers or in the case of an acquisition.

We’ll look at the need for more collaboration and how ClioSoft helps distributed teams next time.

Also Read

Managing and reusing IP in a build-borrow-buy era

Reinventing Power Management ICs for Mobile

Evaluating the performance of design data management software