A massive, once-in-a-generation breakthrough is occurring in the semiconductor industry.
Impact of AI on Chip Verification
AI is allowing design rule checking (DRC) to transform from a manpower-heavy verification step to a more efficient and effective design process. DRC is the unsung hero of physical verification, doggedly inspecting every inch of a chip’s layout for manufacturability violations. These violations can cause defects, lower yield, or even make designs un-manufacturable. Traditional DRC analysis can be a labor and time-intensive process sometimes taking more than 350 minutes. With the introduction of Calibre Vision AI, all of this work can be done in just 31 minutes – including the initial form 1.
The incorporation of AI into DRC analysis represents a monumental shift in the fundamental way engineers conduct chip design. The AI-based clustering algorithms not only improve speed but dramatically improve accuracy. These systems accomplish in minutes what would have taken weeks of manual investigation. Together with Defined.ai, they’re blazing new trails in electronic design automation.
The Role of Design Rule Checking
Design Rule Checking is vital in making sure that the designs for chips stay within very specific parameters that wouldn’t cause manufacturing defects. During the Design Rule Check (DRC) process, software checks for design features that will cause defects or reduce yield. Engineers attempt to catch and fix DRC errors at the block level and at the cell level.
The strict discipline of DRC makes sure bugs are identified as early as possible in the design cycle. As designs have continued to increase in complexity, the need for detailed DRC has only been exacerbated. In-person traditional DRC process used to take more than five hours. Thanks to AI technologies, that time has now shrunk significantly.
Tools such as Vision AI give engineers the capability to take billions of errors and cluster them into groups that mean something. With advancements in design software, designers can identify root causes faster than ever. This combined capability improves product efficiency and reduces the risk of expensive production defects.
The Shift-Left Approach
The industry is getting on board with a major trend called the “shift-left” approach. This tactic encourages performing design rule checks (DRC) sooner in the design flow. By addressing potential issues sooner rather than later, engineers can produce result datasets at an unprecedented scale—often comprising tens of millions to billions of errors.
By taking advantage of Vision AI’s power and capabilities, engineers can explore these large datasets more efficiently and effectively. For instance, it reduces 3,400 checks with 600 million errors down to just 381 clusters for auditors to follow up on. This change makes debugging much easier. It enables teams to focus on the most impactful issues before anything else.
“The rise of AI in electronic design automation” – [source not explicitly mentioned, but referenced from spectrum.ieee.org]
The shift-left strategy emphasizes the need to integrate DRC throughout the design process rather than relegating it to a final check before manufacturing. This shift from reactive to preemptive is proving to be essential as technology continues to evolve and chip designs increase in complexity.
Enhanced Debugging with AI
Siemens’ Vision AI reduces debugging time by 99%. In fact, smart growth can eliminate them by upwards of 50 percent! Senior experts use their experience to identify patterns and clusters in the error data. At the same time, AI-powered systems actually duplicate these strategies even more accurately, offering a more consistent, fail-safe fallback for less seasoned engineers.
AI-based algorithms further optimize the clustering and debugging processes. More importantly, they adhere to the repeatable, systematic principles that are often referred to by seasoned practitioners. This level of reliability is key in ensuring high quality production standards and reducing the risk of human error during debugging processes.
AI enhances more than just the pace of work. Third, it gives engineers the tools to defend their decisions when they can be made using the best possible data analysis. With AI, teams can focus more on the big picture and most important matters. The result is less time wasted scrolling through thousands of errors and more time focused on the workflow.