Memory access error in optixTutorial (Linux)

Hi,

I just wanted to get started with OptiX by compiling and having a look at the provided Example “optixTutorial”.
When I execute the generated executable I get a message telling me that there was an error while accessing memory (no further information). By including some fprintf’s in optixTutorial.cpp I found out that this error occurs in the function “createOutputBuffer”. I have no idea where this error comes from and how I can get the example running on my system.

Here some information about my system settings:
Linux system openSUSE 12.2 (x86_64)
Graphic card GeForce GTX 680
Graphics driver: NVIDIA 310.19
OptiX SDK: 4.0.2
CUDA: 5.5

Any ideas?

I guess that there is a problem in my system settings and the combination of driver, CUDA version, OptiX version etc. But I don’t know what would be a working setting for my openSUSE system.

Correct and that’s listed on the OptiX 4.0.2 download site under “System Requirements” near the bottom of the page:

[i]System Requirements (for running binaries referencing OptiX)
Graphics Hardware: CUDA capable devices of Compute Capability 3.0 (“Kepler”) or higher are supported on GeForce, Quadro, or Tesla class NVIDIA products. Fermi class GPUs are no longer supported
Graphics Driver:
Windows 7/8: driver version 368.86 or later is required.
Windows 10: driver version 369.09 or later is required.
Linux: driver version 367.35 or later is required.
Operating System: Windows 7/8.1/10 64-bit; Linux RHEL 4.8+ or Ubuntu 10.10+ 64-bit; Mac OS 10.9 or higher
CUDA Toolkit: It is not required to have any CUDA toolkit installed in order to run OptiX-based applications.

Development Environment Requirements (for compiling with OptiX)
CUDA Toolkit 6.5, 7.0, or 7.5: OptiX 4.0 has been built with CUDA 7.5, but any specified toolkit should work when compiling PTX for OptiX. If an application links against both the OptiX library and the CUDA runtime on Linux, it is recommended to use the same version of CUDA that was used to build OptiX. OptiX supports running with NVIDIA Parallel Nsight but does not currently support kernel debugging. In addition, it is not recommended to compile PTX code using any -G (debug) flags to nvcc.
C/C++ Compiler: A compiler compatible with the used CUDA Toolkit is required. Please see the CUDA Toolkit documentation for more information on supported compilers.
[/i]
Means your GTX680 is a Kepler board which is supported by OptiX 4.0.2.
The display driver is much too old, it should be one supporting CUDA 7.5 and that would also be the recommended CUDA toolkit to compile your shader programs to PTX code.

Thank you very much for your answer. Since the root password was unknown I decided to start from scratch and use Ubuntu as operating system. After facing many many other problems and trying two different Ubuntu releases I finally managed to set up a running system!

Operating system: Ubuntu 15.04 64-bit
Graphic card: GeForce GTX 680
Graphics driver: 367.35
OptiX SDK: 4.0.2
CUDA Toolkit: 7.5
CMake 3.0.2
gcc: 4.4

Thanks again for your help!