Cannot use compute capability 5.2


I’ve recently switch to GeForce GTX 960 which is said to support compute capability 5.2. And deviceQuery utility from CUDA Samples also prints that cc 5.2 is supported. I am using CUDA Toolkit 7.0 and the latest driver.
However, when i set ‘Code Generation’ property (in MS visual Studio) to “compute_52,sm_52” i get compile-time error:

Error 1 error MSB3721: The command ““C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v6.5\bin\nvcc.exe” -gencode=arch=compute_52,code=“sm_52,compute_52” --use-local-env --cl-version 2013 -ccbin “C:\Program Files (x86)\Microsoft Visual Studio 12.0\VC\bin\x86_amd64” -I"C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v6.5\include” -I"C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v6.5\include" --keep-dir x64\Release -maxrregcount=0 --machine 64 --compile -cudart static -DWIN32 -DWIN64 -DNDEBUG -D_CONSOLE -D_MBCS -Xcompiler “/EHsc /W3 /nologo /O2 /Zi /MD " -o x64\Release\ “C:\Users\dmsuvorov\Documents\Visual Studio 2013\Projects\cuda_test\cuda_test\”” exited with code 1. C:\Program Files (x86)\MSBuild\Microsoft.Cpp\v4.0\V120\BuildCustomizations\CUDA 6.5.targets 593 9 cuda_test

And with ‘Code Generation’ set to “compute_50,sm_50” everything works just fine. What am I doing wrong?

Thanks in advance.

Well, you say you are using CUDA Toolkit 7.0, but your toolchain being invoked by MSVS is actually the CUDA 6.5 toolchain:

"C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v6.5\bin\nvcc.exe" -...

There were multiple versions of the CUDA 6.5 toolkit, some did not support cc5.2. The 6.5.19 update or CUDA 7 should support cc5.2

Pardon my silliness. It appears that I haven’t confiigured VS to use CUDA 7.0 Toolkit.

Yeah, I created that project before SDK update, so it was configured to use the old version. Thanks a lot for help.