Results

Keysight Technologies Inc.

15/04/2024 | News release | Distributed by Public on 16/04/2024 05:35

Sixteen Semiconductor Design Data Management Best Practices

Sixteen Semiconductor Design Data Management Best Practices

Key takeaways:

  • Traditional data management and version control systems suitable for software development cannot handle semiconductor design data well.
  • Purpose-built tools for semiconductor design data management (DDM) can enhance the design workflows across different EDA workflows.
  • Semiconductor design teams must choose scalable and secure DDM solutions to navigate increasingly complex design requirements and heterogeneous integration.

Semiconductor chips, the backbone of modern technology, power everything from smartphones to Mars rovers. According to Gartner, in 2023, worldwide semiconductor sales reached $533 billion, underscoring the industry's immense value. The field of semiconductor design, which includes the creation of both physical integrated circuits (ICs) and related software, accounts for a substantial share of R&D investments across all industries. With government efforts like the CHIPS Act, enhancing advanced IC (Integrated Circuit) design capabilities has become a strategic imperative.

However, semiconductor design is a lengthy and complex process, especially for advanced chips for automotive, defense, aerospace, and AI (Artificial Intelligence) applications. Key steps include design specifications, architectural design, functional design, logic design, circuit design, simulation, verification, prototyping, testing, and manufacturing. According to International Business Strategies (IBS), developing a large-scale 2nm-class chip could cost $725 million.

Today's IC design process spans multiple geographies and electronic design automation (EDA) workflows, which requires a systematic and secured approach for managing all the design files and data across disparate systems.

In this article, we will explain the basics of semiconductor design data management (DDM) and its importance. We will also explore the emerging challenges in this field, 10 best practices in semiconductor design data management, and key considerations for SoC (System on Chip) and chiplet design engineers. Finally, we will demonstrate how Keysight Design Data Management (SOS) can effectively help streamline DDM and enhance design collaboration across teams.

What is semiconductor design data management?

Semiconductor design data encompasses a broad range of information critical to both front- and back-end development and production of integrated circuits (ICs).

Semiconductor design data management is the systematic organization, storage, retrieval, and revision handling of all the design-related information throughout the engineering lifecycle.

Figure 1. A chip design flow and types of design data

This data is integral to various stages of semiconductor design and manufacturing. Here are the main types of semiconductor design data.

Design specifications

Design specifications include customers' functional specifications, which provide detailed descriptions of the functions that the integrated circuit (IC) must perform. It encompasses performance metrics such as speed, power consumption, and heat dissipation for its intended application.

Schematic design data

Schematic design data involves diagrams representing the design components using abstract, standardized symbols. Schematic diagrams focus on the functional description of the IC - demonstrating how the components are electrically connected.

Figure 2. Schematic design samples in Keysight EDA software

Layout data

Layout data in semiconductor design specifically refers to the detailed physical arrangement of various components in standard file formats such as GDSII or OASIS. These files include the IC layers, geometries, and design rules essential for the manufacturing process. This level of detail ensures that the physical realization of the schematic design meets all specified electrical and functional requirements.

Simulation data

IC designers rely on simulation data to predict the behavior of a circuit before it is physically fabricated. This process is crucial for verifying that designs perform as intended under various conditions (i.e., temperature, power, and frequencies) without the need for costly physical prototypes. Circuit simulation data include the following:

  • Test vectors: Sets of test conditions applied to the design.
  • Timing analysis: Analysis on the timing performance of the design.
  • Power analysis: Analysis on how much power the design consumes and how it affects power delivery under different conditions.
  • Thermal analysis: Analysis on how the design manages heat during operation.
  • EMI / EMC analysis: Analysis on potential electromagnetic interference (EMI) problems under the defined safety standards.

Test data

Test data involves checking IC performances during both the pre- and post-silicon validation phases, where the design is tested under a range of operating conditions.

PDK data

Process development kit (PDK) involves details about the semiconductor fabrication process such as process variations, lithography information, and materials used, which are critical for ensuring the manufacturability of the IC designs.

IP (Intellectual Property) libraries

Semiconductor intellectual property (IP) core refers to reusable units of logic, cell, or chip layout design. It includes IP data such as specifications, source code (if applicable), integration guidelines, and licensing information. Effective semiconductor design data and IP management is critical as companies own more internally developed or third-party IPs. Engineering teams must manage the IP update, publish, and consumption across different EDA workflows to maximize the value of IP reuse. They also need to keep track of critical IP for different applications, use cases, and compliance requirements.

Documentation and Bill of Materials (BOM)

IC designers must keep track of necessary documentation such as design specifications, standards documentation, user manuals, and operational guidelines that help understand and implement the designs. A bill of materials (BOM) is a comprehensive list of parts, items, documents, and other materials required in a design.

Metadata

Metadata helps in managing and organizing the semiconductor design process. For instance, IP metadata could involve details such as version histories, origins, process technologies, and licensing details.

The importance of semiconductor design

Before exploring semiconductor design data management, it is vital to recognize the significant role of design in the trillion-dollar semiconductor industry.

Semiconductor design enhances the efficiency of every aspect of our lives, from transportation to manufacturing. It also enables emerging technologies like artificial intelligence (AI), 6G, and autonomous driving. Simply put, when there is a chip design breakthrough, all applications that use chips can benefit from it.

To improve chip performance, integrated circuit (IC) design teams must consider various trade-offs such as performance versus power efficiency, simulate design performance (i.e., timing, signal integrity, power, thermal, and EMI), and verify functionality before prototyping.

As a result, IC design is an extensive effort requiring large teams of highly skilled engineers, each specializing in different domains, to collaborate for years to finalize a design ready for production.

With new chip architectures and heterogeneous integration, design data grows exponentially from concept to tape-out. With a single chip housing billions of transistors and various heterogeneous IP blocks, design data management (DDM) becomes an increasingly strategic process to detect issues earlier in the design stages, optimize overall system performance, and decrease time to market.

Try Keysight DDM SOS for free

Key stakeholders in semiconductor design management

In the complex landscape of semiconductor design, companies can be categorized into four main groups based on their roles and functions within the design data flow.

  • Fabless companies focus on chip design while outsourcing manufacturing to third-party foundries.
  • Integrated device manufacturers (IDMs) handle both design and manufacturing, typically through in-house fabrication facilities.
  • Original equipment manufacturers (OEMs), including automakers and cloud computing providers, are increasingly designing chips tailored to their specific needs.
  • EDA / IP vendors offer electronic design automation (EDA) tools, third-party IP cores (processors, memories, interfaces, and security), and services to assist in chip design.

The increasing demand for more complex integrated circuits (ICs), particularly systems on chips (SoCs), incentivizes IDMs and fabless companies to have more domain-specific design centers and processes. Distinctive design teams often use different EDA tools that best fit their specific needs. Without robust, centralized design data management, such diversification poses significant challenges for holistic design integration.

Demo: Keysight SOS' integration with all major EDA tools

Four Critical Semiconductor Data Management Trends

The semiconductor industry is experiencing key digital transformation trends that impact design data management (DDM) strategies for both large enterprises and startups.

Shift-left and more simulation data

As systems grow more complex, adopting "shift-left" principles-which move testing and design validation earlier in the development cycle-allows teams to address potential issues through simulation before engaging in expensive prototyping.

A 2023 survey by Keysight reveals that over three-quarters of electronic design teams anticipate spending more time on simulations, with one-fifth noting a greater than 50% increase.

Figure 3. Over 70% of design teams expect to increase simulation in the next 24 months. (Source: Dimensional Research)

To process and analyze massive simulation data efficiently, it is critical to switch to more scalable DDM solutions to accelerate data access and minimize network storage burden.

AI-enhanced design

Designers can more effectively meet design targets by leveraging AI-based EDA tools. These tools automate routine design tasks such as design space exploration, enabling engineers to focus on more sophisticated decisions. However, reliance on machine learning and other AI algorithms results in a substantial increase in design data throughout the engineering lifecycle, often overwhelming traditional data management systems.

Chiplets and heterogeneous integration

The shift toward chiplets and heterogeneous integration reflects a strategic move to enhance design flexibility and performance while reducing costs. These technologies allow for the use of highly specialized components that can boost performance further. Effective design data management (DDM) is critical to seamlessly integrating across various EDA tools, manage diverse data files for multi-disciplinary workflows, and maintaining traceability throughout the complex design hierarchy.

Secure design data

With increased scrutiny on the security of semiconductor designs, there is a growing emphasis on advanced DDM tools, storage, and data encryption techniques. Implementations like strict access controls and geo-fencing are essential to meet export control requirements and enhance cybersecurity, ensuring that sensitive design data and intellectual properties (IPs) remain protected against unauthorized access.

Importance of design data management software

Effective semiconductor design data management software (DDM) confers multiple advantages.

Enabling multi-team collaboration

DDM is crucial to multi-site collaboration efficiency, facilitating knowledge sharing, and accelerating semiconductor innovation. For instance, a centralized data repository makes it more convenient and efficient to share data across distributed teams. Even as they work independently in various locations, versioning policies ensure that all teams are working with the most up-to-date information. This approach minimizes problems like task duplication, conflicting changes, and errors due to outdated design files.

Integrating workflows

Design data management enables seamless integration across EDA tools to avoid disrupting the existing workflows of design engineers. Design data management also connects key business systems like enterprise resource planning (ERP) and product lifecycle management (PLM) for seamless data flow across various stages of product development from design to manufacturing.

Figure 4. Core capabilities of design data management software

Enhancing design traceability

Semiconductor designs go through multiple revisions and iterations. Design data management systems track design data from concept to tape-out to provide the latest status to technical and project management stakeholders. For example, safety-critical industries like automotive must follow strict regulations and standards like the International Organization for Standardization (ISO) 26262. Design data management systems such as Keysight SOS enable compliance by maintaining traceability and an auditable trail for every design decision, file change, and approval.

Driving IP reuse efficiency

Design data management simplifies the reuse of existing IP blocks, speeding up the design of new products, improving efficiency, and reducing mistakes by using previously verified components.

Data security and access control

Protecting sensitive design data and IP from unauthorized access is critical. Design data management systems incorporate stringent security measures and access controls to safeguard the data, ensuring that only authorized personnel can access or modify the designs.

Emerging challenges in semiconductor design data management

The integration of advanced semiconductor design data management (DDM) systems is critical for addressing six pervasive challenges in the industry.

No.1 Siloed design data systems

design teams often use varied design data systems tailored to their specific needs for capturing design data. This diversity can lead to difficulties in design compatibility and integration across teams, complicating the collaborative design process. In the realm of IC design, EDA tools, simulators, and test software result in large binary-format files. Version control systems meant for software development like Git provide poor user experiences and inefficientworkflows for such hardware design data formats.

No. 2 Version control

IC components frequently require changes for different applications, making it challenging to maintain consistent integration at the system level. Without effective revision control, the incorrect selection of design branch variants can lead to the reintroduction of previously design flaws.

No. 3 Network storage

Managing vast volumes of design data across various stages of the design lifecycle becomes increasingly difficult as network storage demands go up.

The traditional version control system such as Git requires users to clone the repository. As a result, every IC designer on the project needs to create a copy of all the large design files and all the revisions. This approach becomes impractical due to the storage implications, especially when considering the costs of high-availability network storage in a corporate setting.

No. 4 Communication barriers

Effective communication about design data is crucial for high-quality product, yet it is often impeded by factors such as geographic distances, different time zones, language barriers, cultural differences, and variations in skill sets among engineers and teams. Collaboration across distributed sites can become inefficient if design data management systems can't ensure reliable data centralization, data consistency, and low latency. A DDM system must enable custom workflows to facilitate information exchange, approval processes and discussions.

No. 5 IP and data security

Sensitive and valuable IP must be protected from state and industrial espionage as well as secured against backdoor attacks and data theft. Moreover, the uptick in geopolitical tensions and export controls in recent years requires advanced geo-fencing and access control capabilities of DDM.

No. 6 Compliance requirements

The complexity of international and industry standards means that compliance must be built into design data management workflows. As more design teams tap into chiplet-based design projects, tracking a diverse set of design components with different manufacturers, standards, and licenses becomes more time-consuming and challenging.

Five best practices for improving design data accessibility

To enhance accessibility and manage the complexities of semiconductor design data, companies can adopt five crucial best practices.

  • Centralize repositories: By storing design data in a centralized design data management system, companies create a single source of truth. This ensures that all design teams, regardless of their geographical locations, access consistent and accurate data, thereby reducing siloed information and enhancing design productivity.
  • Enhance data access speed: Setting up cache servers and enabling auto-synchronization at each site enable design engineers to access large graphic files faster. This approach ensures that all sites maintain a low-latency view of the design data and important design changes become immediately available across all locations.
  • Automated data backups: Implementing automated backups for the entire protects against data corruption and loss. Automated testing of the restoration and cleanup workflows ensures that the system can quickly recover, when necessary, while minimizing network storage.
  • Monitor and adjust network storage continuously: Actively monitoring network storage and adjusting storage strategies across sites allows companies to optimize performance and avoid potential bottlenecks. This continuous adjustment ensures that DDM systems are not only efficiently used but also scaled appropriately to meet varying demands.
  • Utilize cloud solutions: Cloud storage solutions can complement on-premises options, offering advantages like scalability, disaster recovery, durability, and various data synchronization solutions.

Seven best practices for simplifying version control

Enhancing version control in semiconductor design processes is crucial for managing complex design. Here are seven best practices specifically aimed at refining version control.

  • Centralize design data: Centralizing design data onto a single platform ensures that everyone accesses the most current files, while still maintaining the availability of older versions for audit trails or reviews. This not only simplifies access but also optimizes storage capacity by saving only the differences between versions.
  • Adopt structured versioning policies: Implement structured versioning policies, such as semantic versioning. Structured versioning is vital for managing engineering change orders (ECOs) efficiently and ensuring design traceability to comply with industry standards like ISO 26262, which governs functional safety in automotive systems.
  • Automate check-in and check-out triggers: Establish workflows that activate upon check-ins and check-outs to prevent concurrent modifications, which can cause conflicts or data loss. These triggers should also automate data validations to facilitate the verification of recent changes.
  • Incorporate visual diff tools into design workflows: Given the highly visual nature of semiconductor designs, incorporating purpose-built visualization tools like Keysight Visual Design Diff helps engineers graphically identify changes in schematics and layouts. This visual comparison aids in quickly understanding modifications and their impacts on the overall design.

Figure 5. Keysight Visual Design Diff (VDD)

  • Streamline approval workflows: Automatically trigger workflows for the review and approval of versions before they are published. This approach ensures that all changes undergo thorough scrutiny efficiently, reducing the risk of errors progressing through to the final product.
  • Implement security controls: Enhance version control with robust data security measures. Implementing access controls through defined roles and permissions for each user helps manage who can read, modify, review, and approve changes. This step is critical for protecting sensitive and valuable intellectual property (IP) from unauthorized access or theft.
  • Integrate with design and development tools: Seamlessly integrate the version control system with other design and development tools used throughout the semiconductor design process. This integration ensures that all tools are working with the most current data, streamlines workflows, and reduces the risk of errors due to data mismatches or outdated information.

By following these best practices, semiconductor companies can significantly improve the precision and efficiency of their version control systems, leading to more reliable and higher-quality semiconductor designs.

Learn more about Keysight Visual Design Diff (VDD)

Four best practices for enhancing design traceability

Enhancing design traceability in complex IC designs is essential for maintaining the integrity and accountability of the design process. Here are four best practices that can effectively enhance the traceability of semiconductor design data.

  • Organize design data hierarchically: Hierarchical folder structures that mirror the project's workflow or organizational structure help to streamline file search and retrieval.
  • Use informative metadata: Metadata tagging of files enables granular search based on specific attributes like project name, date modified, and design parameters. Metadata enhances the granularity of file searches, allowing team members to quickly find files based on specific attributes, thereby reducing the time spent navigating through vast amounts of data.
  • Implement labeling: Track the changes made to design files, including who made the change, when it was made, and why. Mandate the use of labels, tags, and change descriptions for every change. It fosters transparency and accountability throughout the design process, facilitating easier reuse and reviews.
  • Adopt consistent file naming conventions: Adopt a standardized, descriptive, concise naming convention across all projects, incorporating details like the project name, version, file type, and revision numbers. Consistent naming prevents confusion and ensures that files are easily identifiable and traceable.

Optimize semiconductor design data management with Keysight

Keysight Design Data Management (SOS) implements all the mentioned industry best practices, effectively enhancing efficiency and data security across the engineering lifecycle.

Figure 6. Keysight Design Data Management (SOS)

With over two decades of experience, Keysight SOS has become a critical tool for thousands of semiconductor design engineers, safeguarding and streamlining their design data flows from concept through to tape-out.

Here is a detailed look at how Keysight SOS stands apart in managing the unique challenges of IC design compared to traditional software configuration tools used in software engineering.

Optimal data management for large design files

Unlike software configuration tools like Git, Subversion, and Perforce, which are tailored to manage numerous small text files, Keysight SOS is specifically optimized for handling large IC design files such as GDS (Graphic Design System) files or simulation waveforms. Keysight SOS uses efficient storage methods to archive large files optimally, ensuring that the storage space is used judiciously without compromising access speed or data integrity.

EDA tool integration

As design projects increase in size and complexity, Keysight SOS scales seamlessly to meet the expanding requirements of multi-disciplinary IC design workflows. The system's flexibility allows it to be tailored to distinctive design processes and EDA tools, ensuring that it fits perfectly within any organization's specific needs.

Keysight SOS integrates with all major EDA tools like:

Symbolic links to caches

Keysight SOS enables symbolic links that point to actual files stored in the main repository, reducing the need to duplicate data in each user's workspace. This system efficiently handles file checkouts and updates, significantly saving network storage space and enhancing scalability.

The Sparse Populate feature further reduces storage requirements for users. Design projects contain data blocks, like IP libraries or product development kit (PDK) data, which are only needed on a read-only basis. Instead of creating symbolic links for each file in these blocks, Keysight SOS creates a single symbolic link to the block's top-level directory. Users can then modify any required files and check the updated files back into the repository.

Figure 7. Keysight Design Data (SOS) maximizes network storage saving

Efficient IP reuse

Keysight SOS simplifies how teams manage and reuse proven IP blocks in their projects. Rather than duplicating these IP blocks, it creates references to them. This strategy conserves considerable storage and improves the trackability and traceability of IPs.

Enhanced security measures for EDA data

Keysight SOS provides stringent security measures such as access control, geo-fencing, and encrypted data transfer to protect sensitive design data against unauthorized access and potential security threats.

Advanced revision control

The system offers advanced revision control capabilities, including release and derivative design management, which are essential for managing multiple versions and iterations of design files efficiently. In addition, Keysight SOS interfaces smoothly with issue tracking software, making it easier to track changes, manage issues, and maintain a coherent record of all design-related activities throughout the project lifecycle.

Additionally, Keysight SOS further enhances design productivity through features like:

  • hierarchical and visual diff tools
  • automated audit report generation
  • real-time metrics and analytics for the design process
  • analysis of changes between releases or over time
  • propagation of fixes and releases up the design hierarchy
  • integration with ERP, PLM systems, and higher-level product data management (PDM)

In this article, we explored semiconductor design data management systems and the best practices that can mitigate the challenges of managing and versioning that data.

Contact us for expert guidance on design data management and quick demos of our design data management solutions.