SIMD Code Generation

You can generate SIMD code from certain algorithms in DSP System Toolbox such as the FIR filter and the LMS filter which are computationally intensive. The SIMD intrinsics significantly improve the performance of the code generated from these algorithms, in most cases meeting or exceeding the simulation performance. SIMD is a computing paradigm in which a single instruction processes multiple data. Many modern processors have SIMD instructions that, for example, perform several additions or multiplications at once.

You can generate SIMD code using the DSP Intel AVX2-FMA code replacement library (only for AVX2 platform) or by using the model configuration parameter Leverage target hardware instruction set extensions (for Apple processors, ARM^® Cortex^®-A processors, and all Intel^® platforms).

To generate SIMD code from MATLAB^® System objects, you need MATLAB Coder™ and Embedded Coder^®. To generate SIMD code from Simulink^® blocks using the Intel AVX2 code replacement library, you need Embedded Coder. To generate SIMD code from Simulink blocks using the model configuration parameter Leverage target hardware instruction set extensions, you need Simulink Coder or Embedded Coder.