Google Builds Custom Chip for Machine Learning

Google has built a processor called Tensor Processing Unit (TPU) just for artificial intelligence. The chip is designed to speed up machine learning. The name is inspired by the company’s TensorFlow open source deep learning framework. Mountain View has been rumored to have been developing its own custom chip, but Wednesday’s announcement marked the first time it confirmed such an initiative. Google Builds Custom Chip for Machine Learning According to PC World, Google CEO Sundar Pichai made the announcement on Wednesday, May 18, at the company’s I/O conference. He said that the chip delivers an order of magnitude better-optimized performance per watt than all third-party field-programmable gate array (FPGA) and (Graphics Processing Units) GPUs. The application-specific integrated circuit (ASIC) is specific to deep neural networks, which are software or hardware…


Link to Full Article: Google Builds Custom Chip for Machine Learning

Pin It on Pinterest

Share This