[osg-users] [osgPPU] 64-bit Windows build
odedkow at gmx.de
Mon Mar 23 12:27:30 PDT 2009
Thank you for your reply. I didn't make any changes to the CMake files, since I'm using OSG (and now osgPPU :-)) in my spare time and I didn't have much of it in the last few days...
I've checked out the latest version, it compiles without any problems! There are still some config problems, but they could be easily improved, I hope :-) :
1) You have to add the following value
-ccbin "$(VCInstallDir)bin" -DWIN32
to the CUDA_NVCC_FLAGS variable. I'm sure, this string doesn't make
any sense on Linux, so maybe we can add a new CMake varibale, e.g.
CUDA_WIN_NVCC_FLAGS, and set the default value to the
above-mentioned string. It would be then used as additional "nvcc"
parameter when compiling the "kernel.cu" file on Windows.
2) The auto-generated "cudakernel" project does not contain the
"Export.h" file. However, this file could be found on the include path
and used during compiling. That is not a real error :-)
3) The "cudakernel" project does not contain references to the following
"osg.lib", "OpenThreads.lib" and "osgPPU.lib"
(Debug: "osgd.lib", "OpenThreadsd.lib" and "osgPPUd.lib")
I've changed the "Export.h" file (please, see attachment, line 35) to use the already defined "cudakernel_EXPORTS" preprocessor makro.
Read this topic online here:
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
More information about the osg-users