OpenGL texture interop in 280.26

Accelerated Computing CUDA CUDA Programming and Performance
1

Hello,

Does any one of the OpenCL code sample illustrate OpenGL texture interoperability. i.e. a program that calls clCreateFromGLTexture2D and then at some point clEnqueueAcquireGLObjects/clEnqueueRelsaeGLObjects using the corresponding cl_mem object. I just can’t get this to work anymore since driver 280.26. clEnqueueReleaseGLObjects returns with CL_OUT_OF_RESOURCES. It works fine with buffer objects but not with textures.

Can anyone confirm that this actually still works, and if so what has changed in the specs that it would not work from the previous driver version. A demo program with clCreateFromGLTexture2D would be nice.

2

I have the same problem, even with buffer objects. did you solve it?

3

No I haven’t solved it yet. I have given up on the 280.26 driver for now and went back to the one before. If this happens for you with buffer

objects try to run the OpenGL interop source code examples. I am pretty sure they use buffer interop, and they do work with the 280.26 driver.