Have you heard of FPGA yet? Then you must have not updated your knowledge in the enterprise IT field for a long time. Today I will talk to you about what is FPGA? What are the main application scenarios of FPGAs? Some people say that FPGA is the future to replace traditional CPU and GPU, do you believe it?
FPGA, the full name of Field-Programmable Gate Array (Field-Programmable Gate Array), originally appeared as a semi-custom circuit in the field of application specific integrated circuits. It has a certain degree of programmability and can perform data parallel and task parallel computing at the same time. There is a more obvious efficiency when dealing with specific applications.
In fact, Intel, Tsinghua Unigroup, Inspur and other companies have all begun to deploy FPGAs. As early as the SC2015 conference, Inspur teamed up with Altera and iFlytek, China’s largest intelligent voice technology provider, to jointly release a set of deep learning-oriented platforms based on Altera Arria 10 FPGA.
Tsinghua Unigroup is another company that hopes to have direct access to the latest FPGA technology through acquisitions. Following the failure to acquire Micron and the acquisition of the hard drive manufacturer WD, Tsinghua Unigroup may acquire shares of Lattice Semiconductor in the United States. Get into the FPGA market and make a layout.
So, what is the reason why FPGA attracts so many manufacturers crazy?
From the perspective of application scenarios, we can see that after Google’s Alpha Dog defeated the human Go champion, deep learning has gone from the altar, and more and more people are beginning to realize that deep learning may change their lives in the future. Become the direction of future technology development; and FPGA design tools make it more compatible with upper-layer software frequently used in the field of deep learning. FPGA is a major technology that helps deep learning.
However, if FPGAs are the future successor to traditional CPUs and GPUs, it is a bit of an exaggeration. Regardless of whether the CPU and GPU technologies are mature and have a complete ecological chain, the structure of the CPU and FPGA is also different. The CPU has processes such as control instruction fetching, decoding, etc., and the processing is credible and has the ability to handle all kinds of strange instructions.
In contrast, FPGAs cannot handle all kinds of unseen instructions as flexibly as CPUs, and can only process input data and output according to a fixed mode. This is why FPGAs are often regarded as an expert-specific architecture. .
Different from CPU, FPGA and GPU have a large number of computing units, so their computing power is very strong. When performing neural network operations, the speed of both will be much faster than the CPU. However, the instructions natively supported by the GPU due to the fixed architecture of the hardware are fixed, while the FPGA is programmable.
FPGA's application fields are mainly deep learning and neural network algorithms, while traditional CPUs pay more attention to "general purpose". Although GPUs pay more attention to computing speed, their instructions are still fixed. The emergence of FPGAs is so popular all over the world because of its programmability, which gives FPGAs a unique advantage in the field of deep learning. It is not surprising that Google has developed its own chip called TPU in order to develop deep learning. Just as Holzer, the head of Google's data center, said: Google's research and development of its own chips is to solve the problems that the provinces are solving.
Some people believe that when market demand changes, technology will develop accordingly. When deep learning becomes a hot field, the FPGA that best matches it will also become the focus of manufacturers' pursuit.
Modbus Card,Modbus RTU Protocol,RS485 Port,RS232 Port
Shenzhen Unitronic Power System Co., Ltd , https://www.unitronicpower.com