FPGAs (Field Programmable Gate Array)
1. What are FPGAs (Field Programmable Gate Array)?
FPGA (Field Programmable Gate Array) is a semi-customized integrated circuit that dynamically configures logic functions through hardware description languages (such as Verilog/VHDL). Its core feature is that users can still modify the internal circuit structure repeatedly after manufacturing, combining the performance of customized chips with the flexibility of general devices.
2. What are the Core Architecture and Technical Features of FPGAS?
1) Programmable Logic Unit
The basic unit is a lookup table (LUT) + register, which realizes any digital logic function through combination.
Modern FPGA integrates special modules (such as DSP units and high-speed transceivers) to support high-performance computing.
2) Interconnection Resources
The programmable metal wiring network connects logic units, supports dynamic reconstruction of signal paths, and occupies more than 60% of the chip area.
3) Configuration Storage Technology
SRAM Type (mainstream): requires an external configuration chip, and supports unlimited reprogramming.
Flash Type: non-volatile storage, no configuration chip required, suitable for low-power scenarios.
Antifuse Type: one-time programming, radiation-resistant, used in the aerospace field.
3. What are the Core Advantages of FPGAs (Field Programmable Gate Array)?
Reconfigurability: Supports real-time algorithm updates (such as communication protocol upgrades) without hardware replacement.
Parallel Processing Capabilities: Hardware-level concurrent execution (for example, AI reasoning speed is 10 times faster than CPU).
Short Development Cycle: Eliminates ASIC tape-out and development time is shortened by an average of 55%.
Flexibility: Single-chip adapts to multiple scenarios (communication acceleration, image processing, industrial control).
4. What are FPGAs (Field Programmable Gate Array) Used for?
Communication System: 5G base station signal processing (JESD204B interface protocol).
Artificial Intelligence: Neural network reasoning acceleration (such as ResNet-50 hardware optimization).
Industrial Control: Real-time reconstruction of test circuits, adapting to multi-device detection.
Aerospace: Radiation-resistant design ensures reliability in extreme environments.
5. Industry Evolution of FPGAs
Since the first commercial FPGA (XC2064) was launched in 1985, the global cumulative shipments have exceeded 3 billion units, and the market size has exceeded 10 billion US dollars.
6. Mainstream Product Series of FPGAs
Xilinx (AMD): Low-end Spartan → mid-range Artix/Kintex → high-end Virtex.
Intel: Cyclone (low-cost), Arria (mid-range), Stratix (high-performance).
7. FPGAs (Field Programmable Gate Array) FAQs
1) What is the core architecture of FPGA?
FPGAs are composed of configurable logic blocks (CLBs), input/output units (IOBs), and wiring resources. CLB contains a lookup table (LUT) and registers. LUT implements combinational logic (such as 4-6 input logic operations), and registers are used for timing control; IOB supports multi-level standards (such as LVDS, LVCMOS); wiring resources connect modules through programming switches.
2) What are the configuration methods of FPGA?
SRAM Type: mainstream solution, supports repeated programming, and requires an external configuration chip (such as Xilinx Zynq).
Antifuse Type: one-time programming, strong radiation resistance, suitable for aerospace fields.
Flash Type: non-volatile, suitable for low-power scenarios.
3) What is the setup time and hold time?
Setup Time: the minimum time that data needs to be stable before the rising edge of the clock. If it is not met, the data may not be sampled.
Hold Time: the minimum time that data needs to remain stable after the rising edge of the clock. Violation will cause metastability.
4) How to solve the metastability problem?
Use synchronous reset, Gray code counter, asynchronous reset synchronous release circuit (as shown in the figure), or optimize path delay through timing report.
5) What is the role of the global clock network?
Provide low-skew clock distribution to ensure synchronous operation of the entire chip logic and reduce timing deviation.
6) What are the key resource indicators of FPGA?
Number of logic units, number of LUTs, number of block RAM (BRAM), number of DSP modules, number of phase-locked loops (PLL), and maximum user I/O.
7) What are the core application scenarios of FIFO?
Data buffering (balancing the processing rate differences between modules).
Cross-clock domain transmission (asynchronous FIFO is the mainstream solution).
8) How do SPI and I²C achieve multi-machine communication?
SPI: The host enables a specific slave through the chip select signal (CS/SS).
I²C: Using 7-bit address addressing, a single host can control multiple slaves.
9) What are the possible reasons for the failure of board loading but the simulation passing?
Simulation cannot fully simulate the real hardware environment, and it is necessary to check timing constraints, clock stability, power supply noise, or cross-clock domain processing.
10) How to evaluate FPGA performance?
Maximum Frequency (Fmax): measured by WNS (worst negative timing margin) reported by timing.
Latency: the number of clock cycles from input to output, and the pipeline and frequency need to be balanced.
11) What are the channels of the AXI bus?
Including read/write address channels, read/write data channels, and write response channels (no independent read response channels).
12) What dedicated modules are integrated into modern FPGAs?
Embedded block RAM (18Kb units), high-speed serial interfaces (such as JESD204B), floating-point units, and AI acceleration hard cores.
>