Samsung continues to surprise us. It recently released a statement saying that the next generation of mobile phone chips will be designed using artificial intelligence.
As reported Wired, Samsung will use the AI function (DSO.ai) provided by Synopsys to develop the next-generation Exynos processor.
FYI, Exynos chips are used in Samsung smartphones and tablets (mainly in the Korean and European markets).
Synopsys is one of the world's largest providers of chip design (EDA) software. The chairman of this company said that DSO.ai is the first commercial artificial intelligence software for processor design.
However, it should be noted that Samsung does not delegate all the work to artificial intelligence. Instead, it will use reinforcement learning to automatically search the design space to find the best solution.
How are mobile phone chips made?
First, the chip must complete its logic design portion, which is completed by human engineers. After that, the manufacturer begins to lay out and design routes to determine the location of each transistor and how to connect them.
However, modern chips typically have billions or even tens of billions of transistors. This is why design and testing usually take 20 to 30 weeks.
Faced with "innumerable views," the final layout design must strike a compromise between three goals: productivity, power consumption and area (i.e. PPA). A chip design can have 10⁹⁰⁰⁰⁰ possibilities.
Engineers instinctively understand how different designs make chips. But this understanding is difficult to write down in computer code. It's like reinforcement learning.
Note: Reinforcement learning trains algorithms through rewards or penalties. DSO.ai is what they need. This approach is similar to AlphaZero.
AlphaZero learns to play Go and Chess with AI games. DSO.ai learns to make optimal decisions using a large stream of computer-generated data and finds more reliable design solutions in a shorter training time.
DSO.ai has significantly improved design speed. Synopsys said the tool increased chip frequency by 18% in some cases, reduced power consumption by 21% and reduced engineering time from six months to one month.
Moreover, AI will continue to learn on its own to improve its capabilities. Therefore, the longer he works, the smarter he becomes. This means that it will develop chips faster in the future.
In addition to Synopsys, several companies are also developing their own AI tools. Among them, the most famous are Google and Nvidia. Additionally, another EDA vendor, Cadence, has also recently released an AI tool.
