Objects, allocated at the context level, are shared between devices
in the context. For example, both buffers and images created with regular
clCreateImage2D are effectively
shared by default, so no implicit copying/synchronization happens between
devices or between the devices and the host on Map/Unmap calls.
NOTE: Shared memory objects cannot be written concurrently by different command queues. Use explicit synchronization of the write access with OpenCL* synchronization objects, such as events.
You can also avoid implicit copying when you share data with the host, as explained in the “Mapping Memory Objects” section (see Related Documents).
You need to properly align the pointer to the memory referenced with
otherwise explicit copying might happen between host and devices on Map/Unmap.
The following table demonstrates the cases were the application can
benefit from zero-copy sharing with host using
True sharing when hostptr is aligned
to CPU device
|Always True Sharing||
when hostptr is aligned
|Image||True Sharing when
||No True Sharing||
Other resources that are shared automatically across all devices, include program and kernel objects.
Intel's compilers may or may not optimize to the same degree for non-Intel microprocessors for optimizations that are not unique to Intel microprocessors. These optimizations include SSE2, SSE3, and SSSE3 instruction sets and other optimizations. Intel does not guarantee the availability, functionality, or effectiveness of any optimization on microprocessors not manufactured by Intel. Microprocessor-dependent optimizations in this product are intended for use with Intel microprocessors. Certain optimizations not specific to Intel microarchitecture are reserved for Intel microprocessors. Please refer to the applicable product User and Reference Guides for more information regarding the specific instruction sets covered by this notice.
Notice revision #20110804