Most popular Intel provides real-time people in Mi

  • Detail

Intel provides "real-time artificial intelligence" technology

artificial intelligence and deep learning in the Microsoft accelerated deep learning platform, which is expected to change the way humans interact with the world and enterprises operate, so that people can make smarter decisions, and Intel technology is realizing this change. Today, Intel will focus on how one of the world's largest cloud service providers uses Intel artificial intelligence technology to run complex deep learning models that implement a variety of applications, from intelligent face recognition and speech recognition to autonomous driving

Microsoft chose Intel Stratix? FPGA is the key hardware accelerator of its new deep learning acceleration platform (code named Project brainwave). This FPGA based deep learning acceleration platform can provide real-time artificial intelligence technology, enabling cloud infrastructure to process and transmit data as quickly as possible with ultra-low latency. In the cloud, the system needs to process live data streams (including videos, sensors or search queries) and quickly transmit data back to users, so providing real-time artificial intelligence becomes more and more important

stratix 10 FPGA and SOC FPGA use Intel 14nm process technology

in the blog, Microsoft introduced in detail how the project brainwave platform using Intel FPGA can flexibly handle the challenging deep learning model with unprecedented high performance. Microsoft demonstrated the FPGA based deep learning platform at the 2017 hot chips conference, which mainly demonstrated the latest achievements in the field of semiconductor technology. Microsoft is the first large cloud service provider to deploy FPGA in its public cloud infrastructure, and its technological progress through Intel Stratix 10 FPGA has accelerated the deep neural network (DNN), which can replicate the thinking of human brain in a similar way

artificial intelligence is a rapidly developing field, which requires a variety of technologies to effectively manage the needs of various workloads. Intel offers a wide range of technologies to drive market evolution, including Intel Xeon processors, Intel FPGAs and Intel nervana ASIC technologies. Compared with those dedicated deep learning hardware accelerators optimized for running a single workload, the high flexibility of Intel FPGA allows users to customize the hardware to meet specific workload requirements, and quickly reconfigure the hardware as the deep learning workload and usage mode change. Intel Stratix 10 FPGA combines hardened processor modules, which provide a high level of continuous performance and efficiency, and can be programmed flexibly according to users' customization

many silicon AI accelerators currently need to group multiple requests (called batch processing) to achieve high performance. Microsoft's project brainwave platform uses Intel Stratix 10 technology to demonstrate the performance of more than 39 trillion floating-point operations per second on a single request, establishing a new standard for real-time Artificial Intelligence Computing in the cloud. Stratix 10 FPGA establishes a new cloud performance standard for real-time AI computing, with record low latency, high performance and many characteristics without batch processing of AI requests

Doug burger, a famous engineer of next at Microsoft Research, said: we use the high flexibility of Intel FPGA to quickly integrate innovation, while providing performance equivalent to or higher than that of many ASIC based deep learning processing units. Microsoft is trying to deploy project brainwave in azure cloud, so that customers can finally run complex deep learning models with record high performance

the machine learning platform based on Intel technology transforms data into operational business intelligence through accelerated model training, fast scoring, and a highly scalable infrastructure. Combined with the latest Intel Xeon scalable processor ni0.053, Intel FPGA can be customized and programmed to provide low latency and flexible accuracy. Compared with a separate Intel Xeon processor, it has higher performance in deep learning reasoning per watt

about Intel

Intel is committed to expanding the boundaries of technology and making the best experience possible. For more information, for more information, please visit Intel China Center and the official website in new chemical materials industry 1, which has always been an active field of scientific and technological innovation in Shandong, and take multiple measures to play the role of market players such as venture capital

Intel and the Intel logo are trademarks of Intel Corporation in the United States and other countries (regions)

* other names and trademarks involved in the text belong to their respective owners' assets

Copyright © 2011 JIN SHI