ARM Rides AI Wave With New Machine Learning Group

There is a ton of attention being paid to artificial intelligence and machine learning in the tech industry at the moment, encompassing a variety of fields, everything from digital assistants in smartphones and other gadgets, to self-driving vehicles, and more. It is no surprise, then, that chipmaker ARM has gone and created a new machine learning group to accelerate its efforts in AI as it continues to develop increasingly sophisticated processors.

ARM confirmed the new division to EE Times, though it appears a bit gun shy to get into specifics, such as how many people will comprise the dedicated AI group or how much the company is investing into it. Speaking vaguely, Rene Haas, president of ARM's intellectual property group, said it is a "big team in hardware and software." It will be led ARM fellow Jem Davies, who built a reputation working on ARM's media blocks.

ARM

"It's clear we will do [machine learning] in CPUs, GPUs, and special purpose cores, but we are not announcing anything yet," Davies said.

Critics point out that ARM has fallen behind the competition, trailing companies such as Cadence, Ceva, and Synopsys in AI development. But the way Davies sees things, advanced AI is in its infancy, and perhaps even over hyped at the moment.

"Workloads change. [Neural nets are] just another workload. In various descriptions they are becoming the new black," Davies said. "Have byte codes came and went, MP3 decode at sub-milliwatt power had a lifetime. MPEG look like it would be general purpose, but trade-offs drove it to being a specialized chip. It's early days."

Nevertheless, some analysts feel ARM could be as much as three years behind its rivals. On top of that, there are a whole bunch of semiconductor startups receiving funding for AI research and development. ARM has some catching up to do, hence the new dedicated division.

"It's such early days that it's not clear what will win. We are putting a lot of resources on it, but leaving our options open," Haas added.

Via:  EETimes
Show comments blog comments powered by Disqus