[ad_1]
Samsung is using it Artificial intelligence an extremely complex and subtle process for designing a state-of-the-art computer french fries.
The South Korean giant is one of the first chip makers to create AI chips. Samsung uses AI functions in new software Synopsis, a leading chip design software company used by many companies. “What you’re seeing here is the first of any real commercial processors made with AI,” says Aart de Geus, president and CEO of Synopsys.
Others, among others Google and Nvidia, They talked about designing chips with AI. Synopsys ’tool, called DSO.ai, may have the widest reach because Synopsys works with dozens of companies. The tool has the potential to accelerate the development of semiconductors and unlock new chip designs, according to industry monitors.
Synopsys has another valuable asset for working with chips designed by AI: the design of cutting-edge semiconductors that can be used to train an AI algorithm.
A Samsung spokesman confirmed that the company uses Synopsys AI software to design Exynos chips, which are used in smartphones, including its branded phones, as well as other gadgets. Samsung unveiled its latest smartphone, a folding device called Galaxy Z Fold3, earlier this week. The company has not confirmed whether the chips designed by AI are still in production or what products may appear on them.
Throughout the industry, the way AI chips are made seems to be changing.
A Google research paper the AI for organizing components through the AI published in June was described Tensioner chips that it uses AI programs to train and run programs in its data centers. Google’s next phone, Pixel 6, will have a custom chip manufactured by Samsung. A Google spokesperson does not say whether AI has helped design the smartphone.
Chips included Nvidia and IBM there are also Engaged in AI-driven chip design. Other authors of chip design software, among others Cadence, A competitor to Synopsys, are Also developing AI tools to help map plans for a new chip.
Mike Demler, According to a senior analyst at the Linley Group who designs chip design software, artificial intelligence is well-suited for organizing billions of transistors through a chip. “It addresses those issues that have become massively complex,” he says. “It will become a standard part of the computational tool.”
Demler says the use of AI is often expensive because it requires a large amount of cloud computing power to train a powerful algorithm. But he hopes it will be more affordable as the cost of computing goes down and the models become more efficient. He adds that many of the tasks involved in chip design cannot be automated, so expert designers are still needed.
Modern microprocessors are incredibly complex, with many components that need to be combined efficiently. Sketching a new chip design typically requires weeks of effort and decades of experience. The best chip designers have an instinctive way of understanding how different decisions will affect each step of the design process. This understanding cannot be easily written into computer code, but some of the same skills can be used machine learning.
The AI approach used by Synopsys, Google, Nvidia, and IBM uses a machine learning technique called reinforcement learning to work on the design of a chip. It involves the learning of reinforcement training an algorithm to perform a task through reward or punishment, and has proven to be effective in subtle and difficult to codify human judgment.
The method can automatically design the basics of a design, including the location of the components and how to relate them, testing different designs in the simulation and learning what the best results are. This can speed up the chip design process and allow an engineer to experiment with new designs more efficiently. In June blog post, Synopsys said a North American integrated circuit manufacturer has improved its chip performance by 15% using software.
Most popularly, reinforcement learning was used DeepMind, A subsidiary of Google, to be developed in 2016 AlphaGoA program that is able to dominate the Go game table, the age of defeating a world Go player.
[ad_2]
Source link
