My Platform ;
-Vista 64bit Ultimate
-Intel QuadCore 2.6Ghz
-4gb 1200mhz Kingston DDR3 Ram
-Visual Studio 8 ,and 9 (installed)
My Problem ;
At first when I typed “nvcc test.cu” it couldn’t find cl.exe so I edited nvcc.profile and added the path there.I couldn’t get it working so I event copied whole content of C:\Program Files (x86)\Microsoft Visual Studio 9.0\VC\bin to nvcc’s existing folder but no luck again.
nvcc fatal : nvcc cannot find a supported cl version. Only MSVC 8.0 and MSVC 9
.0 are supported
After that I tried to pass ;
C:\CUDA\bin>nvcc --compiler-bindir “C:\Program Files (x86)\Microsoft Visual Stud
io 9.0\VC\bin” -c -D_DEBUG -DWIN64 -D_CONSOLE -D_MBCS -Xcompiler /EHsc,/W3,/nolo
go,/Wp64,/Od,/Zi,/MTd-I “C:\Users\shep\AppData\local\NVIDIA Corporation\NVIDIA C
UDA SDK\common\inc” -I./ -o BlackScholes.exe BlackScholes.cu
resulting in this ;
nvcc fatal : A single input file is required for a non-link phase when an outp
utfile is specified
Ok, so for nothing special it’s all disgussed here (however I can’t still get it working ).
But the thing is when I open NVIDIA GPU Computing SDK Browser and run examples it works perfectly.I modify the codes and run through SDK Browser without a problem.Finally I concluded that there is nothing wrong with the system and tools installed and there is nothing wrong with the nvcc.I read posts about running from Windows SDK’s 64-bit cmd prompt but again no luck with that.
If the SDK browser can run examples, I must be able to run them too!
Any suggestions ?