[ofiwg] Proposal for enhancement to support additional Persistent Memory use cases (ofiwg/libfabric#5874)
sean.hefty at intel.com
Fri May 1 08:45:06 PDT 2020
> > Commit is not a valid term.
> What 'commit' is named doesn't matter much to me. As long as we agree on the
> functionality of the feature, I'm not particularly picky about names. I think from the
> slides I've seen, 'flush' in the slides is equivalent to the 'commit' I've suggested
Agreed - commit is consistent and aligns with the libfabric completion levels. The API definition will eventually be named based on the semantic that is exposed to the application, and what types of transfers it applies to.
> > Is this already in the wild? If not we shouldn’t release it without further
> This is a proposal for a modification to libfabric. This is not in master, this is
> simply a proposal.
James email *is* the further discussion and consideration, aimed at expanding the existing libfabric solution.
More information about the ofiwg