Intel Launches New AI Hardware from Cloud to Edge to Accelerate AI Development

During the 2019 Intel AI Summit, Intel demonstrated a series of new product advances designed to accelerate the development and deployment of artificial intelligence systems from the cloud to the edge, welcoming the arrival of the next wave of artificial intelligence. Intel demonstrated the NNP-T1000 for training and the NNP-I1000 for inference.Intel® Nervana™ Neural Network Processor (NNP). As Intel's first dedicated ASIC chip for complex deep learning for cloud and data center customers, Intel Nervana NNP has ultra-high scalability and ultra-high efficiency. Intel also released the next-generation Intel® Movidius™ Myriad™ Vision Processing Unit (VPU) for edge media, computer vision and reasoning applications.

“As AI advances, both computing hardware and memory are reaching a tipping point,” said Naveen Rao, vice president and general manager of Intel’s Artificial Intelligence Products Group. “Specialized hardware, such as the Intel Nervana NNP and Movidius Myriad VPU, is essential if we are to continue to make great progress in this area. With more advanced system-level AI, we will move from the ‘data to information’ phase to the ‘information to knowledge’ phase.”

With the launch of these products,Intel’s AI solutions portfolio has been further strengthened and is expected to generate more than $3.5 billion in revenue in 2019. The breadth and depth of Intel’s AI portfolio are the highest in the industry, enabling customers to develop and deploy AI models across all devices and at all scales, from the cloud to the edge.

The new Intel Nervana Neural Network Processor, now in production and delivered to customers, is part of a system-level AI solution. The solution provides a complete software stack developed with open components and deep learning frameworks to fully exploit hardware performance. The Intel Nervana Neural Network Training Processor (Intel Nervana NNP-T) strikes a balance between computing, communication, and memory, allowing for near-linear and energy-efficient scaling for both small clusters and the largest pod supercomputers. The Intel Nervana Neural Network Inference Processor (Intel Nervana NNP-I) is energy-efficient and low-cost, and its flexible form factor makes it ideal for running high-intensity multi-modal reasoning at real scale. These two products are targeted at cutting-edge AI customers such as Baidu and Facebook, and are custom-developed for their AI processing needs.

Misha Smelyanskiy, director of artificial intelligence system co-design at Facebook, said: "We are very excited to work with Intel to deploy faster and more efficient inference computing using the Intel Neural Network Processor for Inference (NNP-I). At the same time, our latest deep learning compiler Glow will also support NNP-I."

In addition, the next-generation Intel Movidius VPU is scheduled to be available in the first half of 2020. With its unique and efficient architecture advantages, it will be able to provide industry-leading performance: compared with the previous generation VPU, the inference performance is improved by more than 10 times, and the energy efficiency can reach 6 times that of competing products.Intel also announced the new Intel® DevCloud for the Edge, which together with the Intel® Distribution of OpenVINO™ toolkit addresses a major pain point for developers: the ability to try, prototype, and test AI solutions on a variety of Intel processors before purchasing hardware.

Advancing deep learning reasoning and applications requires extremely complex data, models, and technologies, so different considerations are needed in architecture selection. In fact, most organizations in the industry have deployed artificial intelligence based on Intel® Xeon® Scalable processors. Intel will continue to improve the platform through features such as Intel® Vector Neural Network Instructions (VNNI) and Intel® Deep Learning Acceleration Technology (DL Boost), thereby improving the performance of artificial intelligence reasoning in data centers and edge deployments. For many years to come, Intel Xeon Scalable processors will continue to be a powerful cornerstone of artificial intelligence computing.

Intel customers with the most advanced deep learning training needs require performance to double every 3.5 months, and this type of breakthrough can only be achieved with a range of AI solutions, such as Intel AI solutions. Intel has the ability to consider computing, memory, storage, interconnect, packaging and software to maximize efficiency and programmability, and ensure the key capabilities to scale deep learning to thousands of nodes, thereby expanding the scale of the knowledge revolution.