llama.cpp

opencl: fix `get_tensor` for q5_K adreno gemm_nonshuffle kernel

#29555

Merged