cannot open source file "dynlink_nvcuvid.h" opencv_cudacodec

[Amin-Tgz]

System information (version)

• OpenCV => : 3.3:
• Operating System / Platform => : Windows 10 , 64 Bit:
• Compiler => :Visual Studio 2015:

Detailed description

i meet two bugs:

1.first enable preformance test cause error same as #6716

2.cannot open source file "dynlink_nvcuvid.h"

Severity Code Description Project File Line Suppression State
Error C1083
Cannot open include file: 'dynlink_nvcuvid.h': No such file or directory
opencv_cudacodec	opencv-master\modules\cudacodec\src\precomp.hpp	59	

cmake configure output

CUDA detected: 8.0
CUDA NVCC target flags: -gencode;arch=compute_50,code=sm_50;-gencode;arch=compute_52,code=sm_52;-D_FORCE_INLINES

General configuration for OpenCV 3.3.0-dev

Extra modules:
Location (extra): opencv_contrib-master/modules

 Platform:
   Timestamp:                   2017-09-02T15:37:49Z
   Host:                        Windows 10.0.15063 AMD64
   CMake:                       3.9.1
   CMake generator:             Visual Studio 14 2015 Win64
   CMake build tool:            C:/Program Files (x86)/MSBuild/14.0/bin/MSBuild.exe
   MSVC:                        1900

  CPU/HW features:
    Baseline:                    SSE SSE2 SSE3
      requested:                 SSE3
    Dispatched code generation:  SSE4_2
      SSE4_2 (3 files):          + SSSE3 SSE4_1 POPCNT SSE4_2

    Use Lapack:                  NO
    Use Eigen:                   NO
    Use Cuda:                    YES (ver 8.0)
    Use OpenCL:                  YES
    Use OpenVX:                  NO
    Use custom HAL:              NO

  NVIDIA CUDA
    Use CUFFT:                   YES
    Use CUBLAS:                  YES
    USE NVCUVID:                 YES
    NVIDIA GPU arch:             50 52
    NVIDIA PTX archs:
    Use fast math:               NO

  OpenCL:                        <Dynamic loading of OpenCL library>
    Include path:               opencv-master/3rdparty/include/opencl/1.2
    Use AMDFFT:                  NO
    Use AMDBLAS:                 NO


  Tests and samples:
    Tests:                       NO
    Performance tests:           NO
    C/C++ Examples:              NO

Configuring done
Generating done

[xsorifc28]

Also getting this error, same setup.

[jasjuang]

I also ran into the same problem

[lucaspbordignon]

Thanks @whizzzkid , the solution proposed by #9599 worked!

[whizzzkid]

@lucaspbordignon this is because cuda8 provides nvcuvid.h and cuda9 provides dynlink_nvcuvid.h so opencv builds for cuda8 are not compatible with cuda9

I am still working on the solution for this. Another workaround is to use -D WITH_NVCUVID=OFF flag when building.

[ghost]

Despite the merge, the issue persists with CUDA 10.

fatal error: dynlink_nvcuvid.h: No such file or directory

[achalshah20]

I have the same issue with CUDA 10.

Edit: I checked usr/local/cuda-10.0/include and there are no such files as dynlink_nvcuvid.h or nvcuvid.h!

[ghost]

Same problem with CUDA 10

precomp.hpp:60:18: fatal error: dynlink_nvcuvid.h: No such file or directory

[petrpulc]

NVCUVID is deprecated: https://docs.nvidia.com/cuda/video-decoder/index.html

A possible (and verified) hack is to download new NVDecode SDK: https://developer.nvidia.com/designworks/video_codec_sdk/downloads/v8.2-ga2

Which provides in Video_Codec_SDK_8.2.16/Samples/NvCodec/NvDecoder two header files (cuviddec.h and nvcuvid.h) that just need to be copied to your include folder (/usr/local/cuda/include/)

Changes to the header inclusion based on CUDA version need to be reverted accordingly, of course.

I would suggest moving these two header files to opencv directly as they only specify the interface to the decoder.

[ghost]

@petrpulc Solved thanks

[castleguarders]

This workaround uses CUDA 8 definitions on CUDA 10 for a deprecated API. What is the "longer" term permanent fix? In other words, It's unclear to me if the older nvcuvid.h from Cuda 8 and the one in Cuda 10 / NVDecode SDK are compatible? I don't have Cuda 8 headers to compare.