Skip to content

Latest commit

 

History

History
48 lines (31 loc) · 2.16 KB

GPU_SUPPORT.md

File metadata and controls

48 lines (31 loc) · 2.16 KB
Raw

GPU Support

MegaDetector comes with GPU support which enables much faster image processing.

The following instructions are somewhat technical and have not been tested extensively (meaning there is a chance that you might run into some issues)

1. Check if have a compatible CUDA GPU.

If you don't, then you cannot proceed. You can find the list of compatible GPUs here. It may or may not be up to date, so do some research if you believe you have a compatible GPU that isn't listed.

2. Update your GPU drivers to the latest version

You can find the latest drivers for your model here

3. Install CUDA Toolkit 10.1

It's important to ensure you download the exact versions specified. The app was built with them in mind and I cannot guarantee it will work with other versions.

3.1 Download the Toolkit



3.2 Download cuDNN v7.6.5 for Windows 10 (you will need to create an NVIDIA account)

  • Extract the contents of the zip file into a temporary folder
  • Copy <tempfolder>/cuda/bin/cudnn64_7.dll to C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.1\bin\
  • Copy <tempfolder>/cuda/include/cudnn.h to C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.1\include\
  • Copy <tempfolder>/cuda/lib/x64/cudnn.lib to C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.1\lib\x64\

4. Add the CUDA environment variables to your PATH

You might find that they are already there in which case it's fine. Here are some screenshots to guide you: