The South Korean giant was one of the first chip makers to use artificial intelligence to make chips. Samsung Use AI features in new software Synopsys, A leading chip design software company, used by many companies. Synopsys Chairman and Co-CEO Aart de Geus said: “What you see here is the first real artificial intelligence commercial processor design.”
Other, including Google and Nvidia, Has talked about using AI to design chips. But Synopsys’ tool DSO.ai may prove to be the most far-reaching because Synopsys works with dozens of companies. According to industry observers, the tool has the potential to accelerate semiconductor development and unlock new chip designs.
Synopsys has another valuable asset for making AI design chips: cutting-edge semiconductor designs that can be used to train AI algorithms for many years.
A Samsung spokesperson confirmed that the company is using Synopsys AI software to design its Econos Chips are used in smart phones, including private-branded phones, and other gadgets.Samsung has launched its latest smartphone, one called Galaxy Z Fold 3, Earlier this week. The company did not confirm whether the artificial intelligence-designed chips have been put into production, or what products they might appear in.
Throughout the industry, artificial intelligence seems to be changing the way chips are made.
A sort of Google Research Paper An article published in June described the use of AI to arrange Tensor chip It is used to train and run AI programs in its data center. Google’s next smartphonePixel 6 will use custom chips made by Samsung. A Google spokesperson declined to say whether artificial intelligence helped design smartphone chips.
Chip manufacturers include Nvidia and IBM Too Get involved in artificial intelligence-driven chip design. Other chip design software manufacturers, including Rhythm, Synopsys’ competitor, is Also developing artificial intelligence tools Help draw a blueprint for the new chip.
Mike DemlerLinley Group, a senior analyst who tracks chip design software, said that artificial intelligence is very suitable for arranging billions of transistors on a chip. “It applies to these problems that have become very complex,” he said. “It will become a standard part of the calculation toolkit.”
Demler said that using artificial intelligence is often expensive because it requires a lot of cloud computing power to train powerful algorithms. But he predicts that it will become easier to obtain as the cost of computing decreases and the efficiency of the model increases. He added that many tasks involved in chip design cannot be automated, so expert designers are still needed.
Modern microprocessors are extremely complex and have multiple components that need to be combined effectively. It usually takes weeks of hard work and decades of experience to outline a new chip design. The best chip designers have an instinctive understanding of how different decisions will affect each step of the design process.This understanding is not easy to write computer code, but you can use the following methods to capture some of the same skills Machine learning.
Synopsys and the artificial intelligence methods used by Google, Nvidia, and IBM use a machine learning technique called reinforcement learning to design chips.Reinforcement learning involves Training algorithms to perform tasks Through reward or punishment, it has proven to be an effective method of capturing subtle and difficult to codify human judgments.
This method can automatically formulate the basics of the design by trying different designs in simulation and learning which ones produce the best results, including the placement of components and how to connect them together. This can speed up the process of designing chips and enable engineers to experiment with novel designs more effectively.In a June Blog post, Synopsys said that a North American integrated circuit manufacturer used the software to increase chip performance by 15%.
The most famous is that reinforcement learning is used Deep thinking, Google subsidiary, developed in 2016 Alpha Go, A program that can master board games proficiently is enough to defeat world-class Go players.