Most motherboards disable the on-chip graphics when an add-in graphics solution is detected. If you have an NVIDIA card installed in your system, the on-chip graphics will be disabled. You can enable it, but understand that you cannot use both solutions at the same time without special software (or driver modifications) to support switching between them. The generic drivers provided by Intel, NVIDIA and AMD do not have support for switching between solutions. Whether any such special software is available for your system is unknown. In the case of laptops, the laptop vendors will provide versions of the Intel and NVIDIA/AMD drivers that have been modified to support switching. It is not normal to see such a capability in desktop designs, however.
Thank you for your response, but I do not want to switch between solutions. I want to use the motherboard VGA for all graphics and I want to use the NVIDIA GPU for signal processing algorithms (no display). When I set the motherboard jumper to enable the onboard VGA the GPU is offline as shown in the Detailed System Report, and CUDA GetDeviceCount API returns 0.
This used to work until I updated the motherboard and CPU
None of the desktop boards I work with support this capability. If the server board you are working with is supposed to support this, then I suggest you contact the vendor for instructions on how to enable it.