I try to build a Mex-File where I just included the cuFFTDx library. The programm works without the library but when I build it with the library included I get the follwing errrors out of the library:
C:\Users\sonan\OneDrive - ETH Zurich\Desktop\Bachelor Michael\VolumeStitchingMichael\VolumeStitching_Michael\VolumeStitching_Michael\nvidia\mathdx\22.11\include\cufftdx\include\operators/block_operators.hpp(26): error: expression must have a constant value
C:\Users\sonan\OneDrive - ETH Zurich\Desktop\Bachelor Michael\VolumeStitchingMichael\VolumeStitching_Michael\VolumeStitching_Michael\nvidia\mathdx\22.11\include\cufftdx\include\operators/block_operators.hpp(26): error: a constexpr variable must have a literal type or a reference type
C:\Users\sonan\OneDrive - ETH Zurich\Desktop\Bachelor Michael\VolumeStitchingMichael\VolumeStitching_Michael\VolumeStitching_Michael\nvidia\mathdx\22.11\include\cufftdx\include\detail../traits/detail/check_and_get_trait.hpp(287): error: a constexpr variable must have a literal type or a reference
type
C:\Users\sonan\OneDrive - ETH Zurich\Desktop\Bachelor Michael\VolumeStitchingMichael\VolumeStitching_Michael\VolumeStitching_Michael\nvidia\mathdx\22.11\include\cufftdx\include\detail/fft_execution.hpp(973): error: a constexpr variable must have a literal type or a reference type
C:\Users\sonan\OneDrive - ETH Zurich\Desktop\Bachelor Michael\VolumeStitchingMichael\VolumeStitching_Michael\VolumeStitching_Michael\nvidia\mathdx\22.11\include\cufftdx\include\detail/fft_execution.hpp(995): error: member “cufftdx::detail::fft_block_execution<Operators…>::block_dim” has already
been defined
C:\Users\sonan\OneDrive - ETH Zurich\Desktop\Bachelor Michael\VolumeStitchingMichael\VolumeStitching_Michael\VolumeStitching_Michael\nvidia\mathdx\22.11\include\cufftdx\include\detail/fft_execution.hpp(995): error: “constexpr” is not valid here
6 errors detected in the compilation of “C:/Users/sonan/OneDrive - ETH Zurich/Desktop/Bachelor Michael/VolumeStitchingMichael/FFT_test/FFT_test_cuda_file/FFT_test_cuda_file/fft_michael.cu”.
fft_michael.cu
I start the build with the following command:
mexcuda ‘-LC:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.6\lib\x86’ -I<path_to_cuFFTDx_location> -lcufft -v COMPFLAGS=‘$COMPFLAGS -std=c++17’ -v ‘NVCCFLAGS=-gencode=arch=compute_70,code=sm_70’ fft_michael.cu