ompi

ompi copied to clipboard

This is the third chunk of #12318, which makes datatype copying stream-aware. This will be needed in the coll components to order data movement with kernel invocations.

Adds ompi_datatype_copy_content_same_ddt_stream (and opal equivalent), which takes a stream that is used to submit a series of stream-aware memcpy/memmove operations. If no stream is provided, the default stream is used and synchronized at the end.

Requires #12356, will rebase once that's merged.