Google has reinvented this technology and works in the same way as the internet. Google has designed its own computer chip for the drive works. The CEO of the company SUNDAR PICHAI has said that the company has designed an ASIC, or application-specific integrated circuit that’s specific to work in neural nets. These are the networks of hardware and software that can learn specific tasks by analyzing the vast amount of data. Google uses neural nets to find objects and faces in Photos, to understand the language you speak on android phones, and translate text from one language to another. This technology is even transforming the Google search engine.
TENSOR CHIP:
Google called its chip the Tensor Processing Unit or TPU because it underpins Tensor flows. tensor flow is basically a software engine that drives its deep learning services.
Google released its tensor chip under an open-source license which means that anyone outside the company can use it or modify it. Google has not predicted that it will share the designs for the TPU, but anyone on the outside can make use of Google’s machine learning hardware and software via various Google cloud services.
Google is just one of the many companies incorporating deep learning into a wide range of internet services. Facebook, Microsoft, and Twitter are also taking part in this AI-driven transformation. Typically these internet giants drive their neural nets with chips called graphic processing units, or GPUs made by companies like Nvidia. But some including Microsoft are also exploring. The use of a chip is called field programmable gate arrays, or FPGAs, which can be programmed for specific tasks.
TPU:
A TPU board can fit into the same slot as a hardware drive on the massive hardware racks inside the data center that powers Google online services, the company says, adding that its own chips provide an order of magnitude better-optimized performance per watt for machine learning than the hardware options. TPU is tailored to machine learning applications allowing the chip to be more tolerant of reduced computational precision, which means it needs lesser transistors per operation. Because of this they can squeeze more operations per sec into silicon, and use more powerful machine learning tools. So users can get more intelligent results. The worry for intel is that in showing a willingness to build its own chip for deep learning. Google may expand the effort to design its own central processing unit, which is the heart of computer
