Cannot pass CUDA clang version check

[galaxy001]

/usr/local/cuda/bin/nvcc -ccbin /usr/local/bin/clang-omp test.cu
nvcc fatal   : The version ('30500') of the host compiler ('clang') is not supported

$ /usr/local/bin/clang-omp --version
clang version 3.5.0
Target: x86_64-apple-darwin14.4.0
Thread model: posix

$ gcc --version
Apple LLVM version 6.1.0 (clang-602.0.53) (based on LLVM 3.6.0svn)
Target: x86_64-apple-darwin14.4.0
Thread model: posix

[sfantao]

Hi galaxy001,

CUDA is being used to write the GPU runtime library, and it is known that
the CUDA libraries are not prepared to have clang as the host compiler.
There is a pull request by hanjo being reviewed to avoid that, and just
fail producing that library. We are just evaluating the impact of that
change in some other platforms. In the meantime you can try nvcc to use gcc
(if you have it) for the host compiler.

Thanks for reporting the issue!
Samuel

From: galaxy001 [email protected]
To: clang-omp/clang [email protected]
Date: 07/24/2015 04:56 AM
Subject: [clang] Cannot pass CUDA clang version check (#76)

/usr/local/cuda/bin/nvcc -ccbin /usr/local/bin/clang-omp test.cu
nvcc fatal : The version ('30500') of the host compiler ('clang') is not
supported

$ /usr/local/bin/clang-omp --version
clang version 3.5.0
Target: x86_64-apple-darwin14.4.0
Thread model: posix

$ gcc --version
Apple LLVM version 6.1.0 (clang-602.0.53) (based on LLVM 3.6.0svn)
Target: x86_64-apple-darwin14.4.0
Thread model: posix

—
Reply to this email directly or view it on GitHub.