Build Steps for CMSIS-NN Library

To generate and run C code for deep neural networks on Cortex®-M hardware, you must have the CMSIS-NN (Cortex Microcontroller Software Interface Standard - Neural Networks) library. The following describes the build steps for CMSIS-NN on Windows® and Linux® platforms that use a cross-compiler toolchain.

Requirements

Makefile

To build the CMSIS-NN static library, you must first create a Makefile. Create a copy of the Makefile.mk file in this repository and save it to your computer. To avoid errors during code generation, you must supply the appropriate CFLAGS variable for your target hardware. The Makefile shared here includes several CFLAGS variables pre-validated for popular Cortex-M development boards supported by MATLAB®.

For example, for an STM32F746G-Discovery board, define CFLAGS in the Makefile as:

CFLAGS = -fPIC -c -mcpu=cortex-m7 -Ofast -DNDEBUG -mfloat-abi=hard -mfpu=fpv5-sp-d16

If you do not see your hardware target listed, you must create a custom CFLAGS variable with the necessary -mcpu, -mfloat-abi, and -mfpu flags.

Toolchain

We recommend using the same toolchain for building your MATLAB-based application and the CMSIS-NN library. We have validated the CMSIS-NN library build processes below using the following toolchains:

• Linux platforms: GNU ARM® Embedded Toolchain version 8.3.0
• Windows platforms: GNU ARM® Embedded Toolchain version 10.3.1

Linux Install Steps

1. Install the open-source, AArch32 bare-metal target (arm-eabi) GNU Arm Embedded Toolchain (v8.3.0) provided by ARM (https://developer.arm.com/tools-and-software/open-source-software/developer-tools/gnu-toolchain/gnu-a/downloads/8-3-2019-03).
2. Download the source code for CMSIS version 5.7.0 (available at https://github.com/ARM-software/CMSIS_5/archive/refs/tags/5.7.0.zip).
3. Unzip the CMSIS source code to a folder and follow these steps to build and generate the static library:
   • Open a Linux terminal.
   • Change directory to the CMSIS-NN source folder by running the below command. Here, <CMSIS Root folder> refers to the extracted CMSIS folder.

   cd <CMSIS Root folder>/CMSIS/NN
   

   • Copy your Makefile.mk file to the current directory (<CMSIS Root folder>/CMSIS/NN). Ensure the CFLAGS variable is correctly defined for your hardware target in the Makefile.
   • At the terminal, run the makefile by using the make command:

   make -f Makefile.mk
   

   • Running the makefile creates the static library libcmsisnn.a in the <CMSIS Root folder>/CMSIS/NN/lib folder.
4. Configure the MATLAB environment to generate code that uses the CMSIS-NN library:
   • At /usr/local/, create a folder named cmsisnn.
   • Copy the header files located at <CMSIS Root folder>/CMSIS/DSP/Include and <CMSIS Root folder>/CMSIS/NN/Include to the location /usr/local/cmsisnn/include.
   • Copy the generated static library located at <CMSIS Root folder>/CMSIS/NN/Include to the location /usr/local/cmsisnn/lib.
   • Open a Linux terminal and use the below command to create a CMSISNN_PATH environment variable:

   export CMSISNN_PATH=/usr/local/cmsisnn
   

Windows Install Steps

1. Install the open-source GNU Arm Embedded Toolchain provided by ARM (available at https://developer.arm.com/-/media/Files/downloads/gnu-rm/10.3-2021.10/gcc-arm-none-eabi-10.3-2021.10-win32.exe).
   • Add C:\Program Files (x86)\GNU Arm Embedded Toolchain\10 2021.10\bin to the PATH system variable in the Windows evironment variable editor.
2. Download and install the make tool. We have tested this step with Cygwin (available at https://www.cygwin.com/install.html). Within the Cygwin installer, during package selection, find the make package under "Devel". Open the dropdown list and replace the default selection "Skip" with the version number of make to install. Finish the installation process.
   • Add C:\cygwin64\bin to the PATH system variable in the Windows evironment variable editor.
3. Download the source code for CMSIS version 5.7.0 (available at https://github.com/ARM-software/CMSIS_5/archive/refs/tags/5.7.0.zip).
4. Unzip the source code to a folder, and follow these steps to build and generate the static library:
   • Open a Windows command prompt.
   • Change directory to the CMSIS-NN source folder by running the below command. Here, <CMSIS Root folder> refers to the location where the CMSIS folder was extacted.

   cd <CMSIS Root folder>\CMSIS\NN 
   

   • Copy your Makefile.mk file to the current directory (<CMSIS Root folder>\CMSIS\NN). Ensure the CFLAGS variable is correctly defined for your hardware target in the Makefile.
   • At the command prompt, run the Makefile with the make command:

   make -f Makefile.mk
   

   • Running the Makefile creates the static library libcmsisnn.a in the <CMSIS Root folder>\CMSIS\NN\lib folder.
5. Configure the MATLAB environment to generate code that uses the CMSIS-NN library:
   • Create a folder named cmsisnn in an arbitrary location. Ensure the file path does not include any space. For example, we validated the following steps with the file path C:\cmsisnn.
   • Create two subfolders within cmsisnn called include and lib.
   • Copy the header files located at <CMSIS Root folder>\CMSIS\DSP\Include and <CMSIS Root folder>\CMSIS\NN\Include to the location ..\cmsisnn\include in the new folder you created in the previous step.
   • Copy the generated static library located at <CMSIS Root folder>\CMSIS\NN\lib to the location ..\cmsisnn\lib.
   • Create a Windows system environment variable named CMSISNN_PATH with the value <filepath>\cmsisnn:
     • For Windows 10, right-click the start menu and select System.
     • Click System Info.
     • Click Advanced System Settings.
     • In the Advanced tab, click Environment Variables.
     • Click New.
     • You will be presented with a prompt to enter a new environment variable. In the Variable name field, enter CMSISNN_PATH. In the Variable value field, enter the file path to the cmsisnn folder you created.