The development and application of the Internet of things is based on the intelligent devices without the number of connected networks, and the data generated by these devices needs to be useful informationBig dataHandling capacity. But the original enterprise server and data processing equipment can no longer fully meet this computing demand. Recently, a speedcore that can be integrated into SOC? Embedded systemFPGAThe advent of IP has brought new impetus to the application of the Internet of things.
Embedded FPGA IP is introduced by achronix semiconductor company. Embedded FPGA as an IP integrated into SOC chip will have many inherent advantages. According to Robert Blake, President and CEO of achronix, one is to increase the Internet bandwidth by 10 times, the other is to reduce the Internet delay to 1 / 10, the third is to reduce the power consumption by 50%, and the fourth is to reduce the cost by 90%.
Specifically, in terms of power consumption, speedcore is directly connected to the SOC through internal wiring, eliminating the large-scale programmable I / O buffer visible in the external independent FPGA. Its power consumption is only half of the total power consumption of independent FPGA. At the same time, the chip area of speedcore can be customized according to the application requirements. stayInterfaceIn terms of performance, the independent FPGA is usually connected through a high latency SerDes architecture, while the speedcore IP interface has lower latency and higher performance. Speedcore connects to theASIC。
In terms of cost, the chip area of speedcore is much smaller than that of independent FPGA due to the omission of programmable input-output buffer architecture. At the same time, speedcore eliminates all support for independent FPGA peripheralComponents and partsThese components include power regulator,Clock generator, level shifter, passive component and FPGA cooling device.
In addition, embedded FPGA has higher system reliability and yield. Integrating the functions of FPGA into an ASIC can eliminate thePrinted circuit boardThe loss of reliability and yield caused by placing an independent FPGA on it.
With speedcore IP products, customers can customize the best chip area, power consumption and resource configuration for their applications. Customers can define LUT, embeddedstorageas well asDSPThe number. In addition, customers can define the aspect ratio of speedcore, the connection of input and output (IO) ports, as well as the trade-off between power consumption and performance.
At present, the advancement of the Internet of things puts forward new requirements for data computing and processing capabilities. In the past, in order to improve the performance of mobile phone, the processor of mobile phone has changed from single core to dual core, four core or even eight core. However, the performance of mobile phone will not be improved as much as the data of processor. How to improve the performance of mobile phone is not an effective solution. In addition, with the promotion of the Internet of things, more and more products realize the networking and product data. In order to turn the real data into valuable information, it needs big data computing ability. However, the traditional basic equipment can’t be better qualified for this kind of big data processing ability.
The emergence of embedded FPGA solves the problem of hardware acceleration by increasing bandwidth, reducing delay and reducing power. The technology is speedsteThe basis of r22i FPGA and speedcore FPGA products. All of achronix’s FPGA products are supported by its ace design tool, which integrates with synopsyssynplifSupport for y pro tools. Its business model is to use standard IP product business model, authorize and license speedcore technology; charge intellectual property royalties for devices using speedcore IP according to the shipment volume; and maintain ace design tools.
At present, embedded FPGA is mainly used in the middle and high-end market of the Internet of things to achieve network acceleration, cloud acceleration and wireless acceleration. Robert points out thatMiCrosoft’s Bing search has used embedded FPGA, which provides more than 30 times the data computing power. In the future, it will further explore the application field of this technology in the Internet of things.
Embedded FPGA is superior to FPGA in performance, power consumption and price. Whether it can replace independent FPGA has become a focus topic. In this regard, Robert said that it will not replace the independent FPGA, embedded FPGA will cover part of the market of the independent FPGA, and more importantly, it can continuously expand the application market of the Internet of things.