https://community.intel.com/t5/OpenCL-for-CPU/Map-Unmap-operations-with-Zero-copy-buffer/m-p/1177832#M6667 <P>Assuming by CPU you meant host, then map/unmap is needed still because it is UB by the spec (implementation may cache the data). So automatic propagation is not guaranteed by spec and is implementation defined.</P><P>Also good reading on host pointer topic:&nbsp;<A href="https://community.khronos.org/t/clarify-cl-mem-use-host-ptr/2333" style="color:#0563c1; text-decoration:underline">https://community.khronos.org/t/clarify-cl-mem-use-host-ptr/2333</A></P> Fri, 18 Oct 2019 11:02:24 GMT Evgeniy_T_Intel 2019-10-18T11:02:24Z https://community.intel.com/t5/OpenCL-for-CPU/Map-Unmap-operations-with-Zero-copy-buffer/m-p/1177831#M6666 <P>When we create physical zero copy buffers using USE_HOST_PTR, do we really need to perform map/unmap operations everytime we make changes to the buffer from the CPU side. Since, both GPU and CPU access the same memory, will not the changes propagate themselves automatically?</P> Fri, 11 Oct 2019 05:48:57 GMT https://community.intel.com/t5/OpenCL-for-CPU/Map-Unmap-operations-with-Zero-copy-buffer/m-p/1177831#M6666 verma__abhi1 2019-10-11T05:48:57Z https://community.intel.com/t5/OpenCL-for-CPU/Map-Unmap-operations-with-Zero-copy-buffer/m-p/1177832#M6667 <P>Assuming by CPU you meant host, then map/unmap is needed still because it is UB by the spec (implementation may cache the data). So automatic propagation is not guaranteed by spec and is implementation defined.</P><P>Also good reading on host pointer topic:&nbsp;<A href="https://community.khronos.org/t/clarify-cl-mem-use-host-ptr/2333" style="color:#0563c1; text-decoration:underline">https://community.khronos.org/t/clarify-cl-mem-use-host-ptr/2333</A></P> Fri, 18 Oct 2019 11:02:24 GMT https://community.intel.com/t5/OpenCL-for-CPU/Map-Unmap-operations-with-Zero-copy-buffer/m-p/1177832#M6667 Evgeniy_T_Intel 2019-10-18T11:02:24Z