Customer care
+34 913 67 09 92
Follow us

Customizing AI for EDA

Artificial intelligence (AI) and machine learning (ML) are transforming the field of electronic design automation (EDA) by offering new ways to optimize hardware design and solve complex problems. In a recent interview with industry experts, the benefits and drawbacks of AI and ML in EDA were discussed, shedding light on the current state and future prospects of this exciting technology.

According to Jackson, from an application development perspective, machine learning is currently the focus in EDA, with more general AI expected to come downstream. EDA companies are creating specialized applications that target specific problems, such as printed circuit board layout, functional verification, and exploration of power, performance, and area (PPA) for digital integrated circuit (IC) design. Different applications require different tools, and the goal is to leverage AI to deliver value to customers and optimize their hardware designs.

Sumner highlighted the importance of identifying the right problems to solve with AI in the near term. Data reduction is one area where AI can be applied to handle the vast amount of data generated in the field and extract relevant information for engineers. However, he emphasized that the focus is currently on ML techniques and that the base technology should work for everyone but can be customized for specific applications.

Pan pointed out that smaller companies do not necessarily need to create their own AI infrastructure, as there are already established platforms like PyTorch, TensorFlow, and Microsoft available from larger companies like Google and Facebook. Customization of these existing infrastructures can enable small companies to develop new algorithms and push the state of the art in their respective applications.

However, the panel also acknowledged that there are limits to how far AI and ML can venture outside of the box of predefined applications. Sumner noted that the tool is the framework for the capabilities of AI, and it should function as expected within its boundaries. Yu shared a real-world example where AI has not yet reached the point of helping engineers with complex debugging tasks, highlighting that there are still challenges to overcome in fully realizing the potential of AI in EDA.

Despite the challenges, the experts expressed optimism about the future of AI in EDA. Yu expressed high hopes that AI can accelerate hardware design and optimize designs in unique ways that were not previously possible. Jackson emphasized that large companies will continue to invest in AI internally to optimize their operations, while EDA companies will focus on providing more general solutions that can scale and be used across different companies. Pan also emphasized that AI has the potential to unleash the creativity of genius engineers and push the boundaries of what is possible in EDA.

In conclusion, AI and ML are already making a significant impact in EDA, with machine learning being the current focus. While there are challenges and limitations, there is optimism about the potential of AI in optimizing hardware design, solving complex problems, and pushing the boundaries of what is possible in EDA. As the technology continues to evolve, it is expected to play a crucial role in shaping the future of electronic design and accelerating innovation in the semiconductor industry