The first curriculum design of Digital Signal Processing, FFT

• Build library

  make build [options]

  The options can be:

  • COMPILER: Specify the compiler you want to build with. The value can be g++ or clang++. g++ by default.
  • CPP_STANDARD: Specify the C++ language standard. The value can be -std=c++11, -std=c++14, -std=c++17, -std=gnu++11, -std=gnu++14, -std=gnu++17, etc. The default value is -std=c++11.
  • OPTIMIZATION: Specify the level of optimization. The value can be -O0, -O1, -Os, etc. -O2 by default.
  • WARNING_LEVEL: Specify the warning level. The default value is -Wall -Wpedantic -Wextra
  • PREDEFINED_MACRO: Specify predefined macros, such as -DUNICODE. Empty by default.

  Example:

  make build CPP_STANDARD=-std=c++17 COMPILER=clang++

  Then the library will be in the build directory. The binary library will be build/bin/libdspfft.a and its header file will be in the build/include directory.

• Build test program

  make test [options]

  The options is the same as that in 'Building library'. And when PREDEFINED_MACRO contains -DTRANSFORM_TO_CRLF_NEWLINE , the program will generate MATLAB code with CRLF newline character (which is essential for MATLAB for Windows to run). Otherwise, the newline character will be LF.

  The executable file will be build/test/main.out.

• Build all

  make build_and_test [options]

  This command is equivalent to:

  make build [options] && make test [options]

• Clean

  make clean

  This will clean up all files built.

Enter the directory src, open DSPFFT.sln with Visual Studio 2019 or later versions, choose a target platfrom (Release | x64 recommended) and then build the solution. Then run publish.cmd to publish libraries and test programs. The binary library will be publish\bin\dspfft.lib, whose header files will be in publish\include, and the executable file will be publish\test\homework.exe. Notice that publish.cmd is used for Release | x64 platfrom by default. To change platform, please edit it and change the value of the variable BINARY_DIR.

To clean up, you can run clean_publish.cmd.

The executable file is build/test/main.out on Linux and homework.exe on Windows. When running this program, you can pass parameters as follows:

• --enable-time-comparing: Compare time consumed using FFT and that using DFT directly.
• --writing-matlab-code-to-file: Generate MATLAB Code and write it into result.m, which can be run by MATLAB.

You should include the header file dspfft.hpp in C++ and include dspfft.h in C. And link the library libdspfft.a on Linux and dspfft.lib on Windows.

The APIs are as follows:

Declared in <dspfft.hpp>. All APIs are in the namespace dspfft.

• template <typename floating_type> ::std::vector<::std::complex<floating_type>> base_2_fft(const ::std::vector<::std::complex<floating_type>>& x)
• template <typename floating_type> ::std::vector<::std::complex<floating_type>> dft(const ::std::vector<::std::complex<floating_type>>& x)

Declared in <dspfft.h>. When included by C++ files, all APIs are in the namespace dspfft and all functions are declared with C language linkage (extern "C").

• complexf, complexl, complexll: Structures of complex number for float, double and long double.
• complexf *base_2_fftf(complexf *, size_t), complexl *base_2_fftl(complexl *, size_t), complexll *base_2_fftll(complexll *, size_t)
• complexf *dftf(complexf *, size_t), complexl *dftl(complexl *, size_t), complexll *dftll(complexll *, size_t)

• When using C language to call the library dspfft, you should link C++ runtime manually for this library is written in C++.
• If you want to run the test program on Linux and run the MATLAB Code generated with MATLAB for Windows, be sure to define TRANSFORM_TO_CRLF_NEWLINE macro when building to generate code with CRLF newline character.

语言：

English

中文（简体）