OpenCL, direct access to host memory from GPU kernels

Is there any way to allocate memory on host, that is accessible directly from GPU, without copying?

Like cudaHostGetDevicePointer in CUDA开发者_如何学JAVA.

Yes, use clCreateBuffer with flags containing one of:

• CL_MEM_USE_HOST_PTR
• CL_MEM_ALLOC_HOST_PTR

Which does what you want. For more information visit the man page of clCreateBuffer.