Nvidia Corp, the leading designer of computer semiconductor chips in the world and whose chips used in latest technologies such as artificial intelligence, unveiled new research demonstrating how AI can be used to better designing of semiconductor chips.

To develop working chips, designers must decide where to place tens of billions of tiny on-off switches called transistors on a wafer of silicon. The precise positioning of the transistors has a significant impact on the chip’s cost, speed, and power consumption.

Chip design experts employ advanced design software from companies like Synopsys Inc and Cadence Design Systems Inc to optimize transistor placement.

Nvidia issued a study on Monday demonstrating how it may use a combination of artificial intelligence algorithms to find better ways to put large groupings of transistors. The article intended to improve on a 2021 paper by Alphabet Inc’s Google, the findings of which were later criticized.

The Nvidia study took an existing effort produced by University of Texas researchers using a technique known as reinforcement learning and put a second layer of artificial intelligence on top of it to achieve even better results.

The work is essential, according to Nvidia chief scientist Bill Dally, because chip manufacturing gains are slowing, with per-transistor prices in future generations of chip manufacturing technologies now greater than earlier generations.

That contradicts Intel Corp co-founder Gordon Moore’s famed promise that processors would always become cheaper and faster.

“You’re no longer actually getting an economy from that scaling,” Dally said. “To continue to move forward and to deliver more value to customers, we can’t get it from cheaper transistors. We have to get it by being more clever on the design.”

(Adapted from ThePrint.in)