How a Chip May Help Microsoft Beat Google

 

Nobody reigns supreme in the fickle world of technology. Look no further than names that in the span of just of a few years gathered more dust than Pompeii: Sony, Blackberry, Yahoo. Relative newcomers like Google gave Microsoft a run for its money.

And yet, the Redmond company remains a software giant. And a revolutionary chip is introducing may catapult it again to the very frontlines of software vanguard. That would be thanks to a new breed of programmable chips.

“Field programmable gate arrays, or FPGAs already underpin Bing, and in the coming weeks, they will drive new search algorithms based on deep neural networks,” reports Wired. This is “artificial intelligence modeled on the structure of the human brain, executing this AI several orders of magnitude faster than ordinary chips could.”

On your clock, that would be 23 milliseconds instead of four seconds of nothing on your screen. “FPGAs also drive Azure, the company’s cloud computing service,” Wired says. “And in the coming years, almost every new Microsoft server will include an FPGA.”

Google took a different approach. Despite being massively expensive, it designed its own processor for executing neural nets, the tensor processing unit.

The trade-off Google chose is speed over long term flexibility. Google’s TPU would recognize with no delay a voice command on a smartphone. But Wired warns that its sacrificing long-term flexibility for speed. With FPGAs, Microsoft has its eyes on the long term. An FPGA isn’t as fast as Google’s TPU, but Microsoft can reprogram the silicon as needs change. And it can reprogram not only for new AI models, “but for just about any task.”

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.