Erdenay Alpay
2015-10-20 18:31:00 UTC
Hi,
On Linux, I create a texture in OpenGL, and then try to make connection to
it from pyCuda.
What I do is the following:
1-> Do the initialization for OpenGL and CUDA
2-> Create OpenGL texture
3-> Then register the texture to CUDA by RegisteredImage function.
4-> Then get mapped object.
5-> Then try to get the device pointer for texture mapped object
by device_ptr_and_size() function.
But at this point, I get segmentation fault in libcuda.so. As far as I can
see from debugger, cuGraphicsResourceGetMappedPointer_v2 is called, and and
then libcuda.so craches. I can use all other functions of mapped object and
registered image in pyCuda. But device_ptr_and_size() function crashes.
Do you have any suggestions or opinions about what would go wrong?
Thanks,
On Linux, I create a texture in OpenGL, and then try to make connection to
it from pyCuda.
What I do is the following:
1-> Do the initialization for OpenGL and CUDA
2-> Create OpenGL texture
3-> Then register the texture to CUDA by RegisteredImage function.
4-> Then get mapped object.
5-> Then try to get the device pointer for texture mapped object
by device_ptr_and_size() function.
But at this point, I get segmentation fault in libcuda.so. As far as I can
see from debugger, cuGraphicsResourceGetMappedPointer_v2 is called, and and
then libcuda.so craches. I can use all other functions of mapped object and
registered image in pyCuda. But device_ptr_and_size() function crashes.
Do you have any suggestions or opinions about what would go wrong?
Thanks,