↩ en technology NGINX This computer chip simulates the future faster than physics

This computer chip simulates the future faster than physics

supercomputer simulates future faster than physics feat.

As we move into the future, computer companies are making their chipsets more and more powerful. For example, Apple’s latest M1 chip has 16 billion transistors, making it extremely fast compared to other SoCs such as Intel’s i9 chip. But no matter how fast Apple’s new computer chips are, I’m convinced they won’t be able to simulate a “faster-than-the-laws-of-physics” future. It turns out that there is actually one supercomputer chip that can perform the aforementioned tasks.

This 462 square centimeter chipset, called the Cerebras CS-1 chip, is currently the world’s largest chip and contains approximately 1.2 trillion transistors . This is a 75,000% increase in transistor count when compared to Apple’s latest SoC.

It has been called “the world’s most powerful AI computing system” and was developed by the organization’s Energy Technology Institute for the U.S. Department of Energy . So researchers recently tested Cerebras CS-1 by simulating combustion inside a power plant.

They fed the supercomputer with all the necessary data about the current operating conditions of the power plant. The researchers then waited for the 1.2 trillion transistors to analyze the data and simulate the future state of the power plant.

Cerebras CS-1 captured more than one million variables, including 3D air movement and temperature fluctuations , and analyzed them to show the future potential of the power plant faster than in real time.

“CS-1 is the first system to demonstrate sufficient performance to simulate more than one million fluid cells faster than real time. It simulates a power plant based on data about current operating conditions. Using CS-1 to do this means we can know what will happen in the future faster than the laws of physics would produce the same result,” read a blog post by the researchers.

Looking to the future, the Cerebras team plans to release a second generation CS chip with 2.6 trillion transistors. Until then, existing CS-1 chips will be used to train neural networks to simulate advanced simulations of real-world scenarios.