Optimizing Design Data Management with Advanced Version Control
To power AI innovations like Large language models (LLM), semiconductor companies face unprecedented pressure to deliver more powerful processors and memory chips.
As the complexity of design requirements surges, so does the strain on traditional design data management (DDM) solutions used to store critical information. As a result, massive design data is being left untapped today as organizations lack an efficient way to analyze it.
As with other data-intensive businesses, robust data management tools are essential to help design engineers store and share the data properly and turn it into valuable insights.
From version control and traceability to network storage optimization, this blog delves into key strategies for unlocking the full value of semiconductor design data and IPs in the AI era.
4 Unique Challenges of Design Data Management
Managing semiconductor design projects entails navigating a complex landscape of technical challenges and collaboration needs. Version control software such as Git, while highly effective for software development, encounters specific challenges when applied to today’s integrated circuit (IC) design environments.
No. 1 Challenge: Large binary files
Semiconductor designs, especially in the analog, custom, and packaging domains, generate large binary files by specialized graphical editors. Traditional version control systems for text files struggle to handle these large binary files efficiently.
Figure 1. Traditional data management systems lead to inefficient network storage for IC design
For instance, Git requires users to clone the entire repository, including all historical revisions, which can be prohibitively expensive in terms of network and storage infrastructure costs.
No. 2 Challenge: Heterogenous design flows
The IC design process is increasingly heterogeneous, involving various specialized design tools for different phases, such as synthesis, place & route, simulation, formal verification, and timing analysis. An effective design data management (DDM) must integrate seamlessly with these tools to streamline diverse workflows and ensure data integrity throughout the design cycle.
No. 3 Challenge: Collaborating across distributed teams
The global nature of semiconductor design requires effective collaboration between distributed teams. DDM is pivotal in enabling team members to access any design files they need quickly, share real-time updates, and keep their work in sync despite geographical separations. However, the decentralized nature of traditional data management systems can hinder such collaborations.
No. 4 Challenge: Intellectual Property (IP) protection and traceability
Given the landscape of geopolitical risks and export controls, implementing strict access controls on sensitive semiconductor design data is indispensable. A robust design data management (DDM) system must ensure that only authorized personnel have the capability to access or alter design data and Intellectual Properties (IPs). Contrarily, the Git model falls short in providing the necessary level of access control critical for Integrated Circuit (IC) design projects.
Moreover, the ability to meticulously track designs and IPs is vital to adhering to the rigorous functional safety standards required in the automotive and aerospace & defense sectors. Should any issues arise during the testing phases, it's crucial for IC designers to trace the problem back to its origin efficiently.
Figure 2. Keysight IP Management (HUB) is ISO 26262-certified, offering enterprise-wide IP catalog, traceability for complete management control
8 key considerations for choosing DDM solutions
Selecting a solid Data Management System (DDM) for integrated circuit (IC) design is no small feat. Here are 8 considerations to ensure you’re making an informed choice that will elevate your team's efficiency by maximizing the value of your data.
Storing large files
IC designs often include very large files, ranging from hundreds of megabytes to several gigabytes. With teams of engineers accessing, adding, modifying, and deleting data, the strategy for network storage takes paramount importance.
For instance, Keysight Design Data Management (SOS) enables more cost-effective workspace management while enhancing data availability with the following innovations:
- Central repository systems to store all data in one place
- Create symlinks to a central cache to prevent duplicates in different work areas
- All unused revisions in the cache is automatically cleaned on a regular base
Figure 3. How Keysight DDM (SOS) optimizes network storage
Merging binary data
IC design, especially in analog and custom designs, generates massive binary files such as schematics and layouts. Unlike text files that can be automatically merged, such binary data necessitates a centralized data repository with features such as edit locks to prevent errors from manual merging.
Incorporating advanced version controls
Just as software developers rely on tools to compare versions of files to identify changes, IC designers need to review differences between versions of schematics or layouts.
Figure 4. Keysight’s ADS version control featuring Cliosoft
Three main changes include:
- Tracking changes between versions: Identifying overall version changes requires analyzing the changes that occur within each format.
- Catching unexpected changes between versions: Expect the unexpected. Each revision inherently introduces changes, many of which are anticipated. However, it's the unforeseen changes that potentially lead to problems much later in the flow. Keysight’s SOS has incorporated advanced version control features to spot these anomalies early on.
- Spec changes with each revision For each revision made to the design IP, performance and specs may change either intentionally (e.g., targeted overall power improvement of the IP), or as a side effect of changes (e.g., timing or power differences from layout modifications). By tracking spec changes accurately, designers can ensure that the revisions align with the required performance benchmarks.
Facilitating collaborative processes
The IC development lifecycle is a collaborative effort involving multiple engineers (design engineers, verification engineers, layout engineers, etc.) across multiple sites. A centralized repository facilitates daily, or even more frequent, synchronization of work, thus enhancing collaboration and visibility across multiple teams.
Empowering geographically distributed teams
IC design teams are often spread across different locations. With the help of Keysight’s SOS, multiple engineers can collaborate on a project, sharing data safely between different geographies, allowing clear terms for handing off – like between schematic and layout designers. In addition, everyone on the team knows when changes to any design files or IP have happened.
Managing co-managed files
Designers often store design objects as collections of co-managed files. A suitable configuration management system should recognize these as related components of a single design unit, rather than as individual files with programmatically generated names. Since different engineers have different types of expertise, it is often necessary to put access controls in place. For instance, an effective DDM solution should ensure that a layout engineer doesn’t accidentally modify a schematic or vice versa.
Identifying design hierarchies
Today’s IC design involves intricate hierarchical structure of design blocks, each instantiating lower-level blocks. A DDM system should have the finesse of a skilled architect, capable of navigating and managing these layers effectively to streamline the design process.
Integrating with multiple tools
It's crucial for DDM to be seamlessly integrated with EDA, debugging, and project management tools throughout the workflows. This integration is critical for ensuring that every change is accurately recorded by the design tools.
Why now is the time to modernize design data management
The transformative power of AI is reshaping not just our work environments but the very fabric of IC design flows. EDA vendors like Keysight are at the forefront of this AI revolution, embedding AI / ML into various design and simulation tasks to accelerate time-to-market. Solid design data management may hold the key to unlocking the full potential of AI's capabilities for design teams.
With advanced version controls and network storage optimization, IC designers can manage vast design data effortlessly, ensuring seamless updates and releases. System integrators gain full visibility into updates and designs, streamlining the IP selection, integration, and verification. Meanwhile, design managers can leverage unified, real-time analytics to guide their teams through changes, resolve conflicts, and approve releases.
Keysight, EDA, and data
By bringing Cliosoft into its EDA portfolio, Keysight boosts its intelligent electronic design automation (EDA) software with comprehensive design data and IP management as essential components in building the foundation for more productive, data-driven workflows. Keysight is also extending Cliosoft's capabilities to include test data in a fabric that provides customers with a robust link between design and test. To learn more about how to scale efficient, collaborative design management across sites with Keysight DDM (SOS), request a demo today.