-- Looking for a CUDA compiler
-- Looking for a CUDA compiler - NOTFOUND
-- CUDA_PATH: C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.6
-- CUDACXX: C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.6/bin/nvcc
CMake Error at CMakeLists.txt:153 (message):
CUDA not found, please build explicitly with -DENABLE_CUDA=OFF if you do
not want CUDA.
I have Cuda 11.6 and cudnn 8.3
Use CUDA version 11.1. Worked for me.
Related
Here is the error messsage:
Darknet_VERSION: 0.2.5.4
vcpkg not found, toolchain not defined, using integrated libs on win32
CUDA_PATH: C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.0
CUDACXX:
CMake Error at CMakeLists.txt:156 (message):
CUDA not found, please build explicitly with -DENABLE_CUDA=OFF if you do
not want CUDA.
Configuring incomplete, errors occurred!
See also "C:/Users/User/Desktop/darknet-master/darknet-master/CMakeFiles/CMakeOutput.log".
See also "C:/Users/User/Desktop/darknet-master/darknet-master/CMakeFiles/CMakeError.log".
Darknet_VERSION: 0.2.5.4
vcpkg not found, toolchain not defined, using integrated libs on win32
CUDA_PATH: C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.0
CUDACXX:
CMake Error at CMakeLists.txt:156 (message):
CUDA not found, please build explicitly with -DENABLE_CUDA=OFF if you do
not want CUDA.
Configuring incomplete, errors occurred!
See also "C:/Users/User/Desktop/darknet-master/darknet-master/CMakeFiles/CMakeOutput.log".
See also "C:/Users/User/Desktop/darknet-master/darknet-master/CMakeFiles/CMakeError.log".
Have a look at this GitHub actions run result on a machine running Windows 10:
nvcc_11.7 visual_studio_integration_11.7 curand_dev_11.7 nvrtc_dev_11.7 nvtx_11.7 cudart_11.7 visual_studio_integration_11.7 nsight_nvtx_11.7
Downloading CUDA Network Installer for 11.7.0 from: https://developer.download.nvidia.com/compute/cuda/11.7.0/network_installers/cuda_11.7.0_windows_network.exe
Downloading Complete
Installing CUDA 11.7.0. Subpackages nvcc_11.7 visual_studio_integration_11.7 curand_dev_11.7 nvrtc_dev_11.7 nvtx_11.7 cudart_11.7 visual_studio_integration_11.7 nsight_nvtx_11.7
CUDA_PATH C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.7
CUDA_PATH_VX_Y CUDA_PATH_V11_7
Adding CUDA to CUDA_PATH, CUDA_PATH_X_Y and PATH
1m 2s
Run cmake -B build -G "Visual Studio 16 2019" -A x64 -DBUILD_EXAMPLES=ON
-- The CUDA compiler identification is NVIDIA 11.7.64
-- The CXX compiler identification is MSVC 19.29.30145.0
-- Detecting CUDA compiler ABI info
-- Detecting CUDA compiler ABI info - done
-- Check for working CUDA compiler: C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v11.7/bin/nvcc.exe - skipped
-- Detecting CUDA compile features
-- Detecting CUDA compile features - done
-- Detecting CXX compiler ABI info
-- Detecting CXX compiler ABI info - done
-- Check for working CXX compiler: C:/Program Files (x86)/Microsoft Visual Studio/2019/Enterprise/VC/Tools/MSVC/14.29.30133/bin/Hostx64/x64/cl.exe - skipped
-- Detecting CXX compile features
-- Detecting CXX compile features - done
-- Unable to find cublas_v2.h in either "C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v11.7/include" or "C:/Program Files/NVIDIA GPU Computing Toolkit/math_libs/include"
-- Found CUDAToolkit: C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v11.7/include (found suitable version "11.7.64", minimum required is "9.0")
-- Looking for C++ include pthread.h
-- Looking for C++ include pthread.h - not found
-- Found Threads: TRUE
-- Automatic GPU detection failed. Building for common architectures.
-- Autodetected CUDA architecture(s): 5.3;6.0;6.1;7.0;7.5;8.0;8.6;8.6+PTX
-- Configuring done
CMake Error at CMakeLists.txt:58 (target_link_libraries):
The link interface of target "nvtx" contains:
-- Generating done
CUDA::nvToolsExt
but the target was not found. Possible reasons include:
* There is a typo in the target name.
* A find_package call is missing for an IMPORTED target.
* An ALIAS target is missing.
CMake Generate step failed. Build files cannot be regenerated correctly.
Error: Process completed with exit code 1.
It seems the installer is installing everything it needs, including NVTX - but somehow, CMake is not then able to locate the installed NVTX library.
Why would this happen? What can I do about it?
This is an open bug against CMake:
Bug 22893: FindCUDAToolkit: Fails to locate nvToolsExt, no error nor warning
Please let Kitware know it's important to you, because they don't seem very keen on fixing it for some reason.
I have been trying to build BlockFactory and I keep getting following error when trying to build it.
PS F:\simconnect-monitor\blockfactory> cmake -S . -B build
-- Selecting Windows SDK version 10.0.17763.0 to target Windows 10.0.19044.
-- The CXX compiler identification is MSVC 19.16.27045.0
-- Detecting CXX compiler ABI info
-- Detecting CXX compiler ABI info - done
-- Check for working CXX compiler: C:/Program Files (x86)/Microsoft Visual Studio/2017/BuildTools/VC/Tools/MSVC/14.16.27023/bin/Hostx86/x86/cl.exe - skipped
-- Detecting CXX compile features
-- Detecting CXX compile features - done
-- Could NOT find Matlab (missing: Matlab_MEX_LIBRARY Matlab_MX_LIBRARY ENG_LIBRARY) (found version "9.10")
CMake Error at C:/Program Files/CMake/share/cmake-3.18/Modules/FindPackageHandleStandardArgs.cmake:165 (message):
Could NOT find Matlab (missing: Matlab_MEX_LIBRARY Matlab_MX_LIBRARY MX_LIBRARY) (found version "9.10")
Call Stack (most recent call first):
C:/Program Files/CMake/share/cmake-3.18/Modules/FindPackageHandleStandardArgs.cmake:458 (_FPHSA_FAILURE_MESSAGE)
cmake/FindMatlab.cmake:1873 (find_package_handle_standard_args)
deps/mxpp/CMakeLists.txt:45 (find_package)
As you can see, MatLab was detected but the correct library cannot be found. I have tried with CMake 3.18.14 and 3.22.3 with no difference. I am really loss right now honestly. I have posted an issue in the BlockFactory GitHub but so far, it is still unresolved.
So, I found the solution. I am using VS2017. By default for VS2017, cmake makes 32-bit project. You have to pass cmake -A x64 to force it to a 64-bit project. In addition to that, though, I have to force finfmatlab.cmake to only look for 64-bit library by basically setting _matlab_64build to always be true. Make sure you have 64-bit MatLab if you do this.
Edit: Alternatively, you can install VS2019 instead since it defaulted to 64-bit project.
I have the follow configuration:
Windows 10 64bit
Matlab r2016b, NVidia GeForce 920MX, compute capability 5.0, toolkit version 7.5
Visual Studio 2015 for community
MatConvNet beta 23
Cuda 8.0 (ga1 02/2017)
cuDNN 6.0 (04/2017)
When I compile with vl_compilenn (only cpu) the compile is correct.
But when I compile with:
vl_compilenn ('enableGpu', true)
or more specific with:
vl_compilenn('enableGpu', true, 'cudaMethod', 'nvcc', 'cudaRoot', 'C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v8.0', 'enableCudnn', true, 'cudnnRoot', 'C:\..\..\..\..\..\MatConvNet\cuda')
I have the follow issue:
vl_compilenn: CUDA: MEX config file: 'C:\Program
Files\MATLAB\R2016b\toolbox\distcomp\gpu\extern\src\mex\win64\mex_CUDA_win64.xml'
C:/Program Files (x86)/Microsoft Visual Studio
14.0/VC/../VC/bin/../../VC/INCLUDE\crtdefs.h(10): fatal error C1083: Non Þ possibile aprire il file inclusione: 'corecrt.h': No such file
or directory data.cu Error using vl_compilenn>nvcc_compile (line 540)
Command "C:\Program Files\NVIDIA GPU Computing
Toolkit\CUDA\v8.0\bin\nvcc" -c
"C:.........\MatConvNet\matconvnet-1.0-beta23\matlab\src\bits\data.cu"
-DNDEBUG -DENABLE_GPU -DENABLE_CUDNN -I"C:..........\MatConvNet\cuda\include" -DENABLE_DOUBLE -D__SSSE3__ -gencode=arch=compute_50,code=\"sm_50,compute_50\" -I"C:\Program Files\MATLAB\R2016b\extern\include" -I"C:\Program Files\MATLAB\R2016b\toolbox\distcomp\gpu\extern\include"
-gencode=arch=compute_50,code=\"sm_50,compute_50\" -O3
-Xcompiler /MD --compiler-bindir "C:\Program Files (x86)\Microsoft Visual Studio 14.0\VC..\VC\bin" -o
"C:..........\MatConvNet\matconvnet-1.0-beta23\matlab\mex.build\bits\data.obj"
failed.
Error in vl_compilenn (line 485) nvcc_compile(opts, srcs{i}, objfile,
flags.nvcc) ;
Also, i have copied the cudnn64_6.dll into the specific directory matlab/mex
If you have an idea, i would be gratefull, since I have made several attempt with another version of matconvnet (example b24), but the issue is similar.
Thank you very much.
I have solved used the follows libraries and software:
Matlab r2017a
Visual studio 2015
MatConvNet beta 24
Cuda 8.0.66
Regards
Probably there is no bin folder so Create 'C:\Program Files (x86)\Microsoft Visual Studio 14.0\VC..\VC\bin' folder. It will stay empty but compiling will finished.
I am trying to use GPUmat(MATLAB) under ubuntu.
For my system, GPUstart works well without any error message like follows :
Starting GPU
- GPUmat version: 0.280
- Required CUDA version: 4.2
There is 1 device supporting CUDA
CUDA Driver Version: 4.20
CUDA Runtime Version: 3.0
Device 0: "GeForce GT 520"
CUDA Capability Major revision number: 2
CUDA Capability Minor revision number: 1
Total amount of global memory: 1073283072 bytes
- CUDA compute capability 2.1
...done
- Loading module EXAMPLES_CODEOPT
- Loading module EXAMPLES_NUMERICS
-> numerics21.cubin
- Loading module NUMERICS
-> numerics21.cubin
- Loading module RAND
But when I try to create variable like ' a = GPUdouble(rand(2)); '
the following error message appears
Error using mxNumericArrayToGPUtypePtr
Unable to allocate memory using cudaMalloc
Error in GPUdouble (line 52)
p.slot = mxNumericArrayToGPUtypePtr(p,
double(A));
I can't guess any of reason why this is hapenning. Can you give me some advice to solve this? I really appreciate for your help.
p.s) At the first time, GPUstart does not work due to the library problems. So I moved all the libraries of CUDA 4.2 to matlab library folders according to GPUmat developer's advice.
Thank you !
You have an incompatible version of the CUDA runtime installed. GPUStart tells you "Required CUDA version: 4.2" but you have the CUDA 3.0 toolkit installed.
You will need to update your CUDA toolkit to a supported version.