NOTES:
ffmpeg -hwaccels
ffmpeg -codecs | grep videotoolbox
# if you run this you'll see the gpu being used.
mkdir -p ./output
ffmpeg -f lavfi -i testsrc=size=1920x1080 -t 00:02:00 -c:v h264_videotoolbox -pix_fmt yuv420p ./output/test_gpu.mp4 -y
cmd+4 in activity monitor
powermetrics -h
sudo powermetrics --samplers gpu_power -i500 -n1
vlc ./output/test_gpu.mp4
> videotoolbox decoder: vt cvpx chroma: 420v
- How to Use GPU Acceleration for Video Processing with FFmpeg here
- NVIDIA CUDA Toolkit - Developer Tools for macOS - CUDA Toolkit 11.7 [here(https://developer.nvidia.com/nvidia-cuda-toolkit-11_7_0-developer-tools-mac-hosts)
- How to Install CUDA on Ubuntu 20.04 LTS here
- Officially official: NVIDIA drops CUDA support for macOS here
- ffmpeg HWAccelIntro here
- Video Toolbox Work directly with hardware-accelerated video encoding and decoding capabilities. here
- Following a Vulkan Tutorial on Your MacBook here
- Accelerated PyTorch training on Mac here
- Get started with tensorflow-metal here