Researchers have shown in their latest research that artificial intelligence can design sophisticated and sophisticated chips within just a few hours. One of these models has recently designed a chip that the perfect scientists do not understand.
The results of this study are currently published in the journal Nature Communications and show significant details of the power of artificial intelligence in design. Artificial intelligence not only took a more efficient approach to chip design, but the software’s approach was so different that it was very unlikely to be invented by the human circuit designer.
Artificial intelligence can also work in chip design
In this challenge, the researchers challenged the ability of artificial intelligence to design MM-Wave chips. These types of chips are commonly used in 5G modems and smartphones. The complexity and need of shrinking is one of the biggest challenges facing human designers in the design of these chips, but artificial intelligence has been able to solve this problem.
Currently, manufacturers rely on a combination of human expertise, custom circuit design and the use of stabilized patterns. Then every new design must pass through the slow optimization process based on testing, because it is often so complex that one cannot fully understand what happens inside the chip.
Deep learning -based artificial intelligence models can use the reverse design method, according to researchers at Princeton School of Engineering and Indian Institute of Technology. In this method, the output is specified and the algorithm finally specifies the inputs and parameters. Artificial intelligence also considers each chip a single product, not a set of elements that must be combined.
One of the professors of electrical and computer engineering at Princeton said the method of artificial intelligence is so complex that humans could really not understand it, but the chip performance was higher than the existing chips. In the end, it was found that artificial intelligence has a very high creativity in chip design that surprised researchers.
RCO NEWS