The U.S. has brought down the ban hammer once more by preventing the sale of EDA tools for China, which are considered an integral component of semiconductor design. Without this paraphernalia, companies like Xiaomi will be limited to TSMC’s 3nm process, with little to no hope of materializing its in-house 2nm SoC. Likely realizing that additional export control rules will be enforced on the region in the future, the Chinese Academy of Sciences (CAS) has successfully unveiled a chip design system called QiMeng that utilizes artificial intelligence to expedite semiconductor development and minimize human involvement.
Scientists have already designed two processors using QiMeng, but they deliver processing capabilities of solutions that existed a multitude of generations ago
The term QiMeng means ‘enlightment’ and according to the South China Morning Post, the CAS’ Laboratory of Processor and the Intelligent Software Research Centre divisions are said to have introduced the autonomous integrated design system. Using Large Language Models, solutions designed by QiMeng can match the performance and efficiency of those developed by human experts. For instance, a self-driving car could take weeks for a team of qualified individuals to make, but would take just a handful of days for the automated system.
QiMeng features three interconnected layers, with the base sporting domain-specific large processor chip model, the middle part housing a hardware and software design agent, and the top part flaunting various processor chip design applications. Using these layers, Chinese scientists have successfully designed two processors. The first one, QiMeng-CPU-v1, touts the capabilities of Intel’s 486 chip, which is more than 36 years old, and the second, QiMeng-CPU-v2, is equivalent to ARM’s Cortex-A53 design.
The details that were published in a research paper last week and open-sourced on GitHub mention the challenges in the semiconductor design phase, such as the lack of advanced fabrication technology, limited resources, and a diverse ecosystem preventing scientists from using QiMeng to its full potential. The goal to improve efficiency, reduce costs, and shorten the development cycle can be obtained, but the lack of cutting-edge machinery means that China will be limited to previous-generation lithography.
Huawei’s partner SiCarrier has earlier been reported to have unveiled a slew of machines at SEMICON that could potentially take on Dutch giant ASML, and the last time the company’s name was mentioned from our side, it was seeking a funding round of $2.8 billion to accelerate its ambitious goals of narrowing the technological gap. Huawei has also managed to develop 14nm EDA tools to mass produce the 7nm Kirin 9020, but to keep competition alive with the U.S., injecting massive sums in research will be mandatory to even the playing field.
News Source: QiMeng
About the author: Omar Sohail is a reporter and analyst for Wccftech's mobile section, specializing in the technology and business of the mobile industry. His expertise lies in the intricate hardware supply chain, covering developments in semiconductor manufacturing, chip lithography, and camera sensor technology.
Follow Wccftech on Google to get more of our news coverage in your feeds.
