This is not the primary "readme" file! Please start with README.md. This file contains an optional subset of the instructions which is specific to AMD GPUs.
If you have a modern GPU made by AMD which is supported by ROCm and HIP, then Darknet can be built to take advantage of the GPU to process images and video frames. This will make Darknet/YOLO run much faster.
The Darknet build process does not yet support auto-detecting the AMD GPU. You must set CMAKE_HIP_ARCHITECTURES correctly. See CM_dependencies.cmake or set it on the command-line like this:
cmake -DCMAKE_BUILD_TYPE=Release -DCMAKE_HIP_ARCHITECTURES=gfx1101 ..If you'd like to prevent the Darknet build process from attempting to detect AMD, ROCm, and HIP, you can define DARKNET_TRY_ROCM=OFF like this when running CMake:
cmake -DCMAKE_BUILD_TYPE=Release -DDARKNET_TRY_CUDA=OFF ..The AMD ROCm and HIP software must be installed prior to running cmake for Darknet/YOLO. This is because as part of Darknet's CMake process, it will attempt to identify your GPU and the necessary ROCm and HIP files.
- If you install ROCm and HIP after having already built Darknet/YOLO, you'll need to delete your
src/darknet/build/CMakeCache.txtfile to force CMake to re-detect ROCm, HIP, your GPU, and the necessary files. - Visit https://rocm.docs.amd.com/projects/install-on-linux/en/latest/install/quick-start.html to download and install ROCm and HIP.