Oak Ridge National Laboratory (ORNL), United States of America
The rise of artificial intelligence (AI) and machine learning applications in computing, the end of Dennard scaling and the looming end of Moore's law have driven the computing community to investigate new types of computer hardware targeted specifically at AI and machine learning computation. Two new types of computers in this class of hardware, neural accelerators and neuromorphic computers, have seen a rise in popularity in recent years. Neural accelerators like Google's TPU have focused primarily on accelerating today's deep learning computation, while neuromorphic computers like Intel's Loihi take a more brain-inspired approach and look to the future of AI and machine learning computation. These new types of computer hardware offer significant advantages over traditional computing approaches, including accelerated neural network-style computation and significantly more energy efficiency. This talk will introduce the fundamental computing concepts of these two new types of computer hardware, highlight some of the initial performance results of these systems and discuss how each type of system will fit into the future landscape of HPC.