Migrating x86_64 workloads to aarch64
Introduction
Porting methodology
Porting analysis
Development environment
Application porting
Run and evaluate
Evaluating performance on Arm hardware (Optional)
Review
Next Steps

  Log an issue

  Fork and edit

  Discuss on Discord

Application porting

Based on the porting analysis you can start making changes to the source code and build options. This might be an iterative process. If the compilation fails, you can make modifications and then compile again.

The steps below assume that you are running all commands inside the aarch64 GCC development container.

Sobel filter

Start by cloning the Sobel filter repository.

    

        
        
            git clone https://github.com/m3y54m/sobel-simd-opencv.git
cd sobel-simd-opencv

x86 intrinsics porting

To port the AVX intrinsics, you can use SIMD Everywhere ( SIMDe ). By using SIMDe you can keep the AVX intrinsics in the source code and the intrinsics will be replaced by NEON instructions.

Start by cloning the SIMDe repository:

    

        
        
            git clone https://github.com/simd-everywhere/simde.git

Here are the changes required in CMakeLists.txt:

    

        
        # Add SIMDe options
include_directories(../simde)
set(CMAKE_CXX_STANDARD 14)
set(CMAKE_CXX_STANDARD_REQUIRED ON)
set(CMAKE_CXX_EXTENSIONS OFF)

Run the command below to make the changes:

    

        
        
            sed -i "28i # Add SIMDe options\ninclude_directories(../simde)\nset(CMAKE_CXX_STANDARD 14)\nset(CMAKE_CXX_STANDARD_REQUIRED ON)\nset(CMAKE_CXX_EXTENSIONS OFF)\n" src/CMakeLists.txt

Finally, you need to include SIMDe AVX headers in main.cpp like this:

    

        
        #define SIMDE_ENABLE_NATIVE_ALIASES
#ifdef __aarch64__
#include "simde/x86/avx.h"
#else
#warning AVX support is not available. Code will not compile
#endif

The changes can be made by running:

    

        
        
            sed -i "40i #define SIMDE_ENABLE_NATIVE_ALIASES\n#ifdef __aarch64__\n#include \"simde/x86/avx.h\"\n#else\n#warning AVX support is not available. Code will not compile\n#endif" src/main.cpp

Compiler options porting

The -mavx compiler option needs to be removed. You can add the optimization flag -O2 as it’s recommended when transitioning to Arm .

Below you can see the changes to make in CMakeLists.txt:

    

        
        # Enable SIMD instructions for Intel Intrinsics
# https://software.intel.com/sites/landingpage/IntrinsicsGuide/
if(NOT WIN32)
    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -O2")
endif()

Make the changes by running the following command:

    

        
        
            sed -i "s/-mavx/-O2/g" src/CMakeLists.txt
Note

Compiler options should be tuned and optimized to achieve higher performance, but you can keep it simple for now as performance optimization comes at a later phase of the project.

Application porting is now complete. Go to the next section to compile and run the ported application.

Back
Next