Neon Support

The NDK supports ARM Advanced SIMD, commonly known as Neon, an optional instruction set extension for ARMv7 and ARMv8. Neon provides scalar/vector instructions and registers (shared with the FPU) comparable to MMX/SSE/3DNow! in the x86 world.

Not all ARMv7-based Android devices support Neon, but devices that do may benefit significantly from its support for scalar/vector instructions.

All ARMv8-based devices support Neon.

The NDK supports the compilation of modules or even specific source files with support for Neon.

Building

Enabling Neon globally

ndk-build

ndk-build does not support enabling Neon globally. To enable Neon an entire ndk-build application, apply the per-module steps to every module in your application.

CMake

Pass -DANDROID_ARM_NEON=ON when invoking CMake. If building with Android Studio/Gradle, set the following option in your build.gradle:

android {
    defaultConfig {
        externalNativeBuild {
            cmake {
                arguments "-DANDROID_ARM_NEON=ON"
            }
        }
    }
}

Enabling Neon per module

ndk-build

To build all the source files in an ndk-build module with NEON, add the following to the module definition in your Android.mk:

LOCAL_ARM_NEON := true

CMake

To build all the source files in a CMake target with NEON, add the following to your CMakeLists.txt:

if(ANDROID_ABI STREQ armeabi-v7a)
    set_target_properties(${TARGET} PROPERTIES COMPILE_FLAGS -mfpu=neon)
endif()

Where ${TARGET} is replaced with the name of your library.

It can be especially useful to build all source files with Neon support if you want to build a static or shared library that specifically contains Neon-only code.

Enabling Neon per source file

ndk-build

When listing source files for your LOCAL_SRC_FILES variable, you have the option of using the .neon suffix to indicate that you want to build individual files with Neon support. For example, the following builds one file (foo.c) with Neon support and another (bar.c) without:

LOCAL_SRC_FILES := foo.c.neon bar.c

You can combine the .neon suffix with the .arm suffix, which specifies the 32-bit ARM instruction set (instead of Thumb2) for non-Neon instructions. In such a case, .arm must come before .neon. For example: foo.c.arm.neon works, but foo.c.neon.arm does not.

CMake

To build a specific source file with Neon, add the following to your CMakeLists.txt:

if(ANDROID_ABI STREQ armeabi-v7a)
    set_source_files_properties(foo.cpp PROPERTIES COMPILE_FLAGS -mfpu=neon)
endif()

Runtime Detection

For maximum compatibility with old devices that may not have Neon support, your app should perform runtime detection to confirm that Neon code can be run on the target device. This is because not all ARMv7-based Android devices support Neon. The app can perform this check using any of the options mentioned in Dealing with CPU features.

You should check that android_getCpuFamily() returns ANDROID_CPU_FAMILY_ARM and that android_getCpuFeatures() returns a value with the ANDROID_CPU_ARM_FEATURE_NEON flag set. For example:

#include <cpu-features.h>

...

if (android_getCpuFamily() == ANDROID_CPU_FAMILY_ARM &&
    (android_getCpuFeatures() & ANDROID_CPU_ARM_FEATURE_NEON) != 0) {
    // Use Neon-optimized routines
    ...
} else {
    // Use non-Neon fallback routines instead
    ...
}

...

Cross-Platform Support for x86

NDK supports cross-platform compilation of your existing ARM SIMD (Neon) instrinsic functions into x86 SSE code, through the use of the third-party NEON_2_SSE.h. For more information on this topic, see From ARM NEON to Intel SSE-the automatic porting solution, tips and tricks.

Sample Code

The hello-neon sample provides an example of how to use the cpufeatures library and Neon intrinsics at the same time. This sample implements a tiny benchmark for a FIR filter loop with a C version and a Neon-optimized version for devices that support Neon.