more likely smartphone or monitor for which you are viewing this web page was assembled in China. Prior to this, the engineering team used years of research and development (R&D) to improve the product, improve aspects such as cooling, durability, and develop new technologies. But in some years it is much more likely that chips (be CPU or GPU) equipping your smartphone, a computer or Smart TV be created intelligence artificial (AI).

NVIDIA uses AI to design its chips

Billy Dalli, Chief Scientist and Senior Vice President of Research at NVIDIAdisclosed at the conference that the company uses AI to improve and speed up the design of your GPUs. The green side of the force itself (one of the nicknames Hardware referring to NVIDIA) there is a link in the topic artificial intelligence and high performance computing. The company uses its own technology AI and yours high performance boards deliver GPUs more efficient.

According Dallya AI able to perform 92% design chipcreating the best communication routes, avoiding data overload and specifying the best positions for components GPU. OUR AI creates a design in a few days, while the work of a person takes an average of a year. Dally emphasizes that the use machine learning in development chip does not affect the work of employees, allowing them to devote themselves to 8% development, which AI can’t do. That 8% takes much less research time, which forces engineers to speed up other research work as well.

OUR design From chips made AI it’s also 12% smaller than those made by humans. So expect three things for the future: your GPU will have layout made artificial intelligenceincreasingly improved performance and reduced size.

Google also uses machine learning for its TPUs.

On the left, a drawing made by an engineer. On the right, made by an AI. Source: nature reproduction – Mirhoseini A,. other

According to Googlein a study published in nature In 2021 development chip through artificial intelligence compatible with or superior to those produced by humans. The fact that the work of several months is done in a few hours will speed up evolution. chips TPU (tensor processor units) of the company, in addition to the possibility of using other chips gives Google. OUR nature emphasizes that the methodology can balance the future end of Moore’s law. The law says that the number of transistors on a chip doubles every two. How much machine learning from Google does not eliminate the physical limitations of enabling more transistors, AI will bring new ways to improve productivity chips.

Samsung employee accused of stealing sensitive data from company chips
The worker allegedly stole information about the manufacturing process of the 3nm and 5nm process technology.

…..

Via: Neowin Source: HPC Wire