I think there’s a minor mistake in the setup for Visual Studio integration.
It seems that the “Project Property → CUDA C/C++ → Host → Additional Compiler Options” property
is not applied to the compiler, cl.
I’m using;
- Visual Studio 2010 Pro. (Japanese)
- Windows 7 Pro. x64 (Japanese)
- CUDA 5 SDK
Hacking around, I applied a patch like the following to “CUDA 5.0.targets” file.
This worked for me, at least.
Would there be anyone who’d like to verify?
--- C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v5.0/extras/visual_studio_integration/MSBuildExtensions/CUDA 5.0.targets 2012-09-25 23:46:58.000000000 +0900
+++ C:/Program Files (x86)/MSBuild/Microsoft.Cpp/v4.0/BuildCustomizations/CUDA 5.0.targets 2013-02-27 16:59:02.000000000 +0900
@@ -612,6 +612,7 @@
MaxRegCount="%(CudaCompile.MaxRegCount)"
PtxAsOptionV="%(CudaCompile.PtxAsOptionV)"
+ AdditionalCompilerOptions="%(CudaCompile.AdditionalCompilerOptions)"
Defines="%(CudaCompile.Defines)"
Emulation="%(CudaCompile.Emulation)"
HostDebugInfo="%(CudaCompile.HostDebugInfo)"