I use Ifort ( Intel Compilers 17 (a.k.a Intel® Parallel Studio XE 2017 (Update 4)) and Intel MKL version 11.1.0.080 ) to build old Fortran code using makefile. I work on claster NSC Thriolith. When I try to compile I get mistake.
ipo: warning #11021: unresolved daxpy_
Referenced in /tmp/ipo_ifort1Ub0el.o
It is clear to me because I loaded Intel MKL version 11.1.0.080, but it looks like compiler does not see MKL.
Related
I just tried to compile a C++ program with boost::multiprecision::float128 support in Visual Studio 2017 using Intel C++ compiler 18.0 update 3.
I add /Qoption,cpp,--extended_float_type to project C/C++ additional options.
However, ICC emit E0020: identifier "_Quad" is undefined error message.
I searched the latest ICC 18 Developer Guide and Reference relating to quadruple precision float and _Quad topics and couldn't find anything.
Is _Quad keyword still supported in ICC 18, or I made mistakes in compiler options?
Thanks.
OS: Windows 10 x64 1803
Visual studio 2017 15.7.3
Intel C++ compiler 18.0 update 3
Has anyone managed to compile the latest release of FFTW 3.3.4 using the Visual C++ compiler? I'm aware that the project maintainers offer pre-compiled dll's that were built with MinGW, but I'm looking to compile them myself.
So far, I've downloaded the source (of course), and I am trying to use one of the pre-made VS solutions that was reported to work for Visual Studio 2010 and FFTW 3.3, but when I'm building, I'm getting a few "unresolved external" errors like the following:
Error 1259 error LNK2019: unresolved external symbol _fftwf_mkprinter_str referenced in function _fftwf_export_wisdom_to_string C:\Users\bryan\Downloads\fftw-3.3.4\fftw-3.3.4\fftw-3.3-libs\libfftwf-3.3\export-wisdom-to-string.obj libfftwf-3.3
I know this is a bit of a stretch, but has anyone built FFTW 3.3.4 using Visual Studion 2013?
I have had the same problem.
Take a look on
http://wiki.panotools.org/Hugin_SDK_%28MSVC_2013%29
There is explained how you can build libfftw-3.3
In the same way you can build libfftwf-3.3
To build bench and benchf you have to remove '....\libbench2\aligned-main.c'
That is explained here (commentar 6)
Compiling FFTW source in Visual studio
I followed the instructions here to compile cuda 5 with visual studio 2012 but I'm still receiving this:
nvcc : fatal error : nvcc cannot find a supported cl version. Only MSVC 9.0 and MSVC 10.0 are supported
What could cause this problem?
Seems that if your code is using dynamic parallelism you need to use msvc2010
At lest for me, the Corvella's answer worked, I built OpenCV 2.4.5(from OpenCV official site) and then Emgu(from Emgu git with OpenCV 2.4.9) with VS2012 and CUDA support, both were successful. I used Windows 7 64 and Windows 8 64. The error occurs when you missed "Build Customization" for some .cu files. Even though, it is just the beginning, you will meet many other errors. Good luck!
I have a C++ program that compiled and ran fine on Linux that I am now trying to compile on a Windows machine in Visual Studio. The main problem I'm facing is the following error message:
error LNK2019: unresolved external symbol _gzread referenced in function
I have downloaded the zlib library and attempted to link it to my project, but apparently unsuccessfully. I'd appreciate clear instructions on how to link this to my project. The less you assume in your instructions, the better, as I'm relatively new to Visual Studio.
Found the solution. Even though I'm working on a 64-bit machine, VS was compiling in 32-bit mode. I just used the 32-bit version of zlibwapi instead of the 64-bit version and it worked.
For work, I am converting the Image Denoising program that comes with the CUDA SDK into a MATLAB program. As far as I know, I have made all the necessary changes required by MATLAB, but when I try to call mex on it, MATLAB returns a bunch of linkage errors that I have no idea how to fix. If anyone has any suggestions on what I might be doing wrong, I would greatly appreciate it.
The command I am giving MATLAB is:
mex imageDenoisingGL.cpp -I..\..\common\inc -IC:\CUDA\include -L..\..\common\lib -lglut32
And the output from MATLAB is a bunch of these:
imageDenoisingGL.obj : error LNK2019: unresolved external symbol __imp__cutCheckCmdLineFlag#12 referenced in function "void __cdecl __cutilExit(int,char * *)" (?__cutilExit##YAXHPAPAD#Z)
I am running:
Windows XP x32
Visual Studio 2005
MATLAB 2007a
You need to link the CUDA libraries to your MEX file. It looks like you're also using some of the "cutil.h" stuff from the CUDA SDK (such as cutCheckCmdLineFlag), so you'll need to link against not only the cudart library, but also cutil. I.e. you probably need to add something like
-Lc:\CUDA\lib -lcudart -lcuda -L<path-to-cutil.lib> -lcutil
to your MEX command-line.
If you are converting from CUDA to MATLAB, then why are you still calling the CUDA functions?
unresolved external symbol
__imp__cutCheckCmdLineFlag#12
High Performance Mark is suggesting, in his comment, to compile mexfiles using CUDA directly under Visual Studio. At the page
Compiling CUDA mex files with Visual Studio
it is described how the compile mexfiles using CUDA under Visual Studio. There is also a downloadable Visual Studio sample project.
The procedure has been tested for CUDA 5.0, Visual Studio 2010 and Matlab 2010a/2012b, but perhaps it could be of interest also to people using other versions of the above products.