The most popular and comprehensive Open Source ECM platform
Let us know more
Artificial Intelligence: Chip Manufacturers Target Specialized AI Chips
Chip manufacturers are beginning to introduce specialized computer chips designed to specifically target Artificial Intelligence (AI) algorithms. While some AI results have been stunning, the reality is that AI algorithms require very intensive computations and it isn’t uncommon for AI programs to run for days. By targeting and optimizing the computations required for the mathematics used regularly be many AI algorithms, the new chips bring needed performance improvements.
Chip companies developing AI chips now include NVIDIA, AMD, Intel, Graphcore, Qualcomm, and Adapteva, but technology companies that have traditionally focused on software and services, like Facebook, Google, Amazon, Microsoft, and Baidu are also developing AI chip technology.
Allied Market Research estimates that the global AI chip market size was $2.4 billion in 2017 and will grow to $37.8 billion by 2025.
What is notable is that there is huge variety in the types of chips being developed. Each is targeted in accelerating performance for specific types of problems. Gary Brown, for example, director of marketing at Movidius, an Intel company, said that “not every AI chip is equal. Each chip can handle different intelligence at different times. Our chip is visual intelligence, where algorithms are using camera input to derive meaning from what’s being seen. That’s our focus.”
Nvidia’s GPU chips have been generic accelerators for a wide variety of algorithms, from image processing to machine learning. Currently, the pendulum now seems to be swinging towards creating highly specialized chips. But it’s likely that as technology improves the cycle will again move back again towards development of more general processing chips.
