nvEncodeAPICreateInstance() fails with NV_ENC_ERR_INVALID_VERSION

I am doing encoding using NVIDIA encoder.My code runs successfully on GTX780 card, windows 10(64bit).
But nvEncodeAPICreateInstance() fails with NV_ENC_ERR_INVALID_VERSION when i run the same code on Tesla k80 card, windows server 2016 .

What might be the issue??

Which driver and SDK versions are you using on each system?

NV_ENC_ERR_INVALID_VERSION normally indicates a driver/SDK version mismatch. For each system, please ensure that you have installed a driver with version greater than or equal to the minimum required driver version for the SDK. You can find the minimum required driver version in Readme.txt file included in the SDK package.

I have got the same problem"nvEncodeAPICreateInstance() fails with NV_ENC_ERR_INVALID_VERSION" when I run the sample “Video_Codec_SDK_8.0.14\Samples\NvEncoder.cpp and NvEncoderCudaInterop.cpp”.

And I set the enviroment as:

  • cuda 7.5 toolkit
  • GeForce 920M with driver 385.41
  • Win10(64bit)

The Readme.txt shows:

[System Requirements]

  • NVIDIA Kepler/Maxwell/Pascal GPU with hardware video accelerators - Refer to the NVIDIA Video SDK developer zone web page (https://developer.nvidia.com/nvidia-video-codec-sdk) for GPUs which
    support encoding and decoding acceleration.
  • Windows: Driver version 378.66 or higher
  • Linux: Driver version 378.13 or higher
  • CUDA 7.5 Toolkit (optional)

[Windows Configuration Requirements]

  • DirectX SDK is needed. You can download the latest SDK from Microsoft’s DirectX website
  • The CUDA 7.5 Toolkit is optional to install (see below on how to get it)
  • CUDA toolkit is used for building CUDA kernels that can interop with NVENC.