Sorry for the confusion. I’ve updated the instructions, as these were pretty out of date. Here’s the new text:
In order to compile the samples in the SDK you need the following.
Visual Studio. Supported versions are listed in the release notes.
CUDA Toolkit. Supported versions are listed in the release notes.
CMake 3.0 minimum (http://www.cmake.org/cmake/resources/software.html).
I suggest the executable installer.
Instructions for building.
Start up cmake-gui from the Start Menu.
Select the C:\ProgramData\NVIDIA Corporation\OptiX SDK \SDK directory
from the installation for the source file location.
Create a build directory that isn’t the same as the source directory. For
example, C:\ProgramData\NVIDIA Corporation\OptiX SDK \SDK\build.
If you don’t have permissions to write into the this directory (writing into
the “C:/Program Files” directory can be restricted in some cases), pick a different
directory where you do have write permissions. If you type in the directory
(instead of using the “Browse Build…” button), CMake will ask you at the
next step to create the directory for you if it doesn’t already exist.
Press “Configure” button and select the version of Visual Studio you wish to
use. Note that the 64-bit compiles are separate from the 32-bit compiles
(e.g. look for “Visual Studio 12 2013 Win64”). Leave all other options on
their default. Press “OK”. This can take a while while source level
dependencies for CUDA files are computed.
Press “Configure” again. Followed by “Generate”.
Open the OptiX-Samples.sln solution file in the build directory you created.
Select “Build Solution” from the IDE.
Right click on one of the sample program targets in the solution explorer and
select “Set as start up project”.
Run the sample. “q” or “Esc” will close the window.
Note that due to the way dependencies are automatically handled for CUDA
compilation in Visual Studio, if you build again Visual Studio will likely ask
you to reload your projects. Please do so. Subsequent compiles should not
result in reloading unless you change the files that are included by a CUDA
Further instructions regarding the build system can be found in comments in the
SDK’s CMakeLists.txt file.