OpenCL VectorAdd demo buffers confusion

Note sure if this can be posted publically, but there is not much information disclosed in this report:

The OpenCL SDK includes a basic sample in OpenCL\src\oclVectorAdd.
It initializes 6 buffers cmMemObjs[6], but only the last 3 are used (basically cmMemObjs[5] = cmMemObjs[3]+cmMemObjs[4]).
Why do the first 3 ‘host’ buffers of this cmMemObjs array exist (line 141-149)? They seem to be unused. Can they be removed, to make the demo more clear?

Apart from that, the VectorAdd.cl kernel uses float8 type, instead of cl_float8, is that on purpose?
Thanks,
Erwin

by the way, the [url=“http://developer.download.nvidia.com/OpenCL/NVIDIA_OpenCL_JumpStart_Guide.pdf”]http://developer.download.nvidia.com/OpenC...Start_Guide.pdf[/url] Jumpstart Guide, which is similar to VectorAdd, only uses 3 buffers.