At the conference in Silicon Valley, TSMC showed developments that could increase the energy efficiency of the information process chips about ten times. For example, at the maximum load, Nvidia Yongali modern servers consume up to 1,200 Watts, which can be compared with the energy consumption of thousands of American houses in continuous work.
The basis of new solutions will be the transition from small modules to architecture – a variety of “chiplets” combined in an information process complex. Such a scheme allows you to combine different technologies in a package and improve performance.
To maximize the abilities of this architecture, developers are increasingly implementing artificial intelligence -based software. Cadence design systems and Synopsys, which are closely cooperating with TSMC, presented new tools that can perform complex design tasks faster and more efficiently than engineers.
According to TSMC, there were more successful solutions for some stages of the work of AI platforms within a few minutes, while engineers needed days.
However, the growth of the complexity of the chips is based on physical restrictions. Today, the main obstacle is the rate of data transfer between crystals. In order to solve this problem, optical compounds that should be adequately reliable for use in large -scale data centers are investigated.
Source: Ferra
I am a professional journalist and content creator with extensive experience writing for news websites. I currently work as an author at Gadget Onus, where I specialize in covering hot news topics. My written pieces have been published on some of the biggest media outlets around the world, including The Guardian and BBC News.
