The Relationship between DSP and FPGA in Embedded Development

fish001

The Relationship between DSP and FPGA in Embedded Development [Copy link]

The so-called single-chip microcomputer focuses on control and does not support signal processing. It is a low-end embedded processor. ARM can be regarded as an upgraded version of the low-end single-chip microcomputer, which supports operating system management, more interfaces such as network cards, and stronger processing capabilities; FPGA is a programmable logic device that focuses on timing and can build almost all digital circuit systems from small to large. DSP mainly performs complex digital signal processing, such as FFT. Usually a complex system can be composed of one or more of microcontrollers, ARM, FPGA, and DSP, each with its own advantages and disadvantages.

DSP is usually used for calculation-intensive tasks, and FPGA is used for control-intensive tasks. Many people use DSP for high-level algorithms and FPGA for peripheral control circuits.

When Xilinx introduced the integration of FPGA and DSP in Beijing last year, it seemed to sweep the DSP and FPGA markets. But later in practice, it was found that the resistance was not small. The key was that it was difficult to match the proportion of FPGA and DSP. We know that FPGA is good at various controls (including signal processing with very high timeliness, such as baseband filtering in mobile communications, which is difficult to handle with C5000 and C6000 unless multiple chips are parallel), and DSP is good at various signal processing operations, but how big is the control circuit in a system, and how many intensive operations are there? The variables are very large, and we can't solve this problem with one chip. For computationally intensive systems, if we use this chip, we may need other DSPs, so we need to understand a variety of DSP programming and interface design. Does this run counter to our original intention?

The above are some introductions to basic concepts. Now I will give a popular introduction. In DSP, you are a software designer and the hardware is completely solidified. All you have to do is to optimize its functions on this fixed hardware platform. Generally, TI's DSP involves most basic task scheduling between BIOS operating systems, as well as algorithm improvement and optimization. The key advantages of DSP include its shorter development time for new and complex algorithms, and the flexibility to run multiple algorithms.

For FPGA, you are a hardware designer, and FPGA is a blank sheet of paper. What you write and draw on it is up to you. For the same FPGA, the functions implemented by a novice and an expert will be very different. The biggest advantage of FPGA is the hardware implementation and the efficiency gain achieved through parallel processing. When using FPGA, you spend most of your time not designing and optimizing algorithms, but logic design and timing constraints, etc.

Relationship between FPGA and DSP:

1. DSP focuses on core algorithm processing, while FPGA focuses on peripheral control processing.

2. DSP is written in C language, and the language execution is serial processing, which is relatively inefficient.

FPGA focuses on parallel processing and has high efficiency; it also has intersection logic (peripheral interface, communication, etc.);

FPGA development areas:

Application areas: electronic design, communications, automobiles, military industry

Not suitable for: consumer products (mobile phones)—FPGA power consumption is high;

FPGA technical difficulties:

1. Requires a specialized hardware language to develop

2. FPGA is more flexible, the design is determined by engineers and requires testing and verification