Skip to main content
Skip table of contents

How to install Nvidia CUDA Toolkit

Nvidia CUDA Toolkit is essential part of Mirasys VMS when you want to use VCA Deep Learning features.

Currently this component is needed to install separately to that server where you want to use VCA Deep Leaning features with Mirasys VMS and Nvidia GPU.

Package Download

  • Package can be downloaded from

    • Mirasys Extranet

    • Or Documentation site Software Download section

  • When you have downloaded package, extract this to wanted location example C:\temp.


  • Double click setup file; cuda_XX.X.X_XXX.XX_windows.exe.

  • After this Windows will ask permission to run this setup file.

  • Click Yes to continue.

  • Installer is asking now where to extract package details.

    • If there is need, change this but if not, then click OK to continue.

  • Extraction takes some time until is done.

    • This depends server specifications how long this will take.

  • After this you should see NVIDIA Installer which check system.

  • Accept NVIDIA software license agreement clicking Agree and Continue.

  • Select Custom installation type and click Next.

  • Select CUDA Component and under that select Development and Runtime components. You can untick Other components and Driver components.

    • If you select these components, this will install package included drivers and other components.

    • Some cases those drivers can be too old for normal usage.

  • Click Next to continue.

  • Use default installation location and click Next to continue.

  • If you have selected more than libraries, you can get the below window.

  • Click I understand tick box and click Next to continue.

  • Now installer start installing CUDA Components to this server.

  • Please wait that installation is done.

  • After installation you get summary of installation if there was selected more than needed components. Click Next to continue.

  • Click Close to close installer.

  • Now NVIDIA CUDA Toolkit is installed to server and you are ready to use VCA Deep Learning features.

    • If VMS Server not recognize CUDA and can’t initialize VCA Deep Learning trackers, please restart VMS Server.

JavaScript errors detected

Please note, these errors can depend on your browser setup.

If this problem persists, please contact our support.