VVdeC2_ARM_Neon

Introduction

“VVdeC2_ARM_Neon” is a SIMD Neon-based VVC decoder implementation. It is an ARM-based configured version of Fraunhofer Versatile Video Decoder (VVdeC) version 0.2.0.0. VVdec has been configured for a heterogeneous NVIDIA Xavier System on Chip embedded platform based on ARM + GPUs. Here, VVdeC has also been configured for CUDA compatibility with GPUs. Moreover, SIMD Neon-based optimization has been performed by adapting the original SIMD instructions with the Neon intrinsics for ARM processors over the embedded platform.

Hardware

The following architectures are tested:

• Aarch64
• ARMv8
• NVIDIA GPU

How to build?

Cmake is used to build the software over Linux (gcc).

Create a build directory in the root directory:

mkdir build

Change to build directory:

cd build

Therefore, use one of the following cmake commands that suited to you. Linux Release Makefile sample:

cmake .. -DCMAKE_BUILD_TYPE=Release

Linux Debug Makefile sample:

cmake .. -DCMAKE_BUILD_TYPE=Debug

Then build using plain make

make

Adoptions

The following open source projects have been adopted during the development of VVdeC2_ARM_Neon.

• Vvdec-0.2.0.0 : https://github.com/fraunhoferhhi/vvdec/releases/tag/v0.2.0.0 -provides a fast VVC x86 software decoder implementation.
• SIMD Everywhere simde: https://github.com/simd-everywhere/simde -provides fast, portable implementations of SIMD intrinsics SSE functions on ARM.
• sse2neon: https://github.com/DLTcollab/sse2neon -provides translation of Intel SSE (Streaming SIMD Extensions) intrinsics to Arm NEON.

Notes

VVdeC2_ARM_Neon is CUDA compatible. VVdeC2_ARM_Neon has been configured focusing on embedded platfrom with NVIDIA GPU.