The LatTIce acquisition, which has been delayed for a year, has finally come to an end. Due to Trump's veto, Canyon Bridge's offer to acquire LatTIce may be blown away. Although the sale is not successful, the pace of latTIce development must continue to move forward. According to its latest developments, latTIce is aimed at the emerging field of network edge. Let's take a look at the related content with the embedded Xiaobian.
There are already 6.4 billion devices connected in the current network, and 5.5 million new devices have been added. Therefore, the rise of the Internet of Things requires new methods of processing and analyzing requirements. Getting the most out of the Internet of Things requires a strong seamless connection between devices and the cloud while eliminating computational and privacy issues. The ability of cloud computing combined with IoT technology means that by 2018, IoT sensors and devices will surpass mobile phones as the largest access device. Complex algorithms for industrial and consumer applications enable voice and face recognition as well as machine learning functions to be rapidly evolving. However, data transfer to the cloud and back to each IoT device must cope with the inevitable network latency, and 45% of all IoT-created data will be stored, processed, analyzed, and approached or on the edge network. "Lattice is moving into the edge of the network. We are also investing in FD-SoI technology and accelerating through acquisitions," said Glen Hawk, chief operating officer of Lattice Semiconductor, on the microgrid.
“Before 2006, Lattice's revenue was mainly from the control of the PLD segment, which was very stable at around US$200 million per year. Since 2006, the interconnection market at the edge of the network has gradually increased. Network edge computing is a brand new market. Demand will be the main driver of growth in the future.†Glen Hawk pointed out that “Lattice can provide complete and highly advantageous solutions in the three areas of control, interconnection and computing at the edge of the network.â€
According to reports, in terms of control, it is now a solid foundation for Lattice to achieve stable revenue, with more than 4,000 customers in 2016 alone. “The long product life cycle and stable supply chain are the basis for meeting the needs of more customers. The rich system design experience helps customers achieve innovation and has been adopted by many customers in the network edge field,†said Glen Hawk.
In terms of network edge interconnection applications, Lattice's iCE series, CrossLink series, and wireless connection series FPGA products are widely used in smart speakers, ADAS, in-vehicle infotainment systems, surveillance cameras, machine vision, tablet computers, VR and other fields. “In the past year alone, we have seen companies from all over the world adopt our small size, low power consumption and low for AR/VR systems, robots, drones, machine vision, intelligent surveillance cameras and other products. Delaying the FPGA. This is just the beginning. We are eager to help innovate and design in the edge of the network."
The potential of network edge artificial neural network applications is unlimited. But the reality is that it's easy to generate ideas, but it's not that simple to implement. How do design engineers bring the benefits of artificial intelligence, neural networking, and machine learning to low-power network devices with limited resources?
“We see that power, price, and performance requirements are different in all types of applications at the edge of the network. Lattice has the power and performance advantages of neural networking and machine learning, so we lock in "1 trillion operations per second, 1W programmable area of ​​the programmable neural network application market." Glen Hawk said, "now in ADAS 360 ° surround view, license plate detection, AR / VR position tracking and other network edge computing applications They can meet their needs. For example, face tracking applications based on convolutional neural networks, using ECP5, power consumption is less than 1W; face detection applications based on binary neural networks, using iCE40 UltraPlus, power consumption is less than 5mW."
In his view, the parallel computing and programmable features of FPGAs are well suited for neural network computing, and are more suitable for a wide range of IoT markets than ASICs. The GPU is very efficient in training deep learning algorithm models, but the advantages of parallel computing cannot be exploited for small batches of data in reasoning. FPGAs have both pipeline parallelism and data parallelism, so processing tasks are less delayed and consume less power. In addition, FPGAs are programmable chips that are more flexible. At present, the deep learning algorithm is not fully mature, and the algorithm is still in the iterative process. If the deep learning algorithm changes greatly, the FPGA is software-defined hardware, which can flexibly switch algorithms and quickly enter the market.
In addition, Glen Hawk emphasized that market opportunities in the smart edge of network edge are expected to double by 2022. "With the growth of the network edge computing field, Lattice is expected to achieve more than $2 billion in revenue." He pointed out that "network edge interconnects maintain stable business growth, network edge computing accelerates future growth, and steadily grows to achieve stable returns. This is Lattice's next growth and enterprising approach."
Stator and rotor laminations are an important part of motors and generators. For small-size and large batch laminations, we usually use the progressive punching. The advantage of progressive punching is that it can be mass-produced in a short time. But the cost of the progressive punching die is high and manufacturing period of progressive punching die is long. Usually the outer diameter is under 300mm.
Stator And Rotor Lamination By Progressive Punching,Rotor Lamination,Motor Rotor Stator,Motor Rotor
Henan Yongrong Power Technology Co., Ltd , https://www.hnyongrongglobal.com