Google has admitted that the chip it claims is the world’s smallest is, in fact, only half the size of the one pictured. The company was quick to credit everyone else in its design team and its data centre boss.

Designing a chip for its Soli system-on-chip (SoC) required solving some very tricky problems. One of them was that the team was being forced to do away with inefficient silicon transistors in favour of incredibly thin transistors, known as TFTs. According to Google’s Algori Suppu-Touretsu, who is in charge of the chip project, the challenge in designing a SoC on such a tiny chip was “like the movie Blade Runner.”

For a long time before Google was even founded, the company had been making its own semiconductors based on designs from chip giant ARM. Google’s Toughest Semi was in fact based on ARM’s 24-nanometer technology. Google switched to 20nm, the super-small node used by Intel, in 2010. For Google, this issue was truly a logistical one: It wanted to get off the Intel gun quickly and at its new size.

It didn’t really help Google’s cause that the processing technology behind what it calls the Google UX was at the same time in flux. So when Google started to work on its Soli chip — which is due to be delivered to partners in 2019 and showcased in 2020 — it needed a solution that would allow it to be much more efficient than its low-power infrastructure technology, the Power VRX2.

Via TechSpot

Spread the fake ai universe

Leave a Reply

Your email address will not be published.