feature: Add multirail support and optional usage of hwloc for optimal NIC assignment

Sandia-OpenSHMEM / SOS

Sandia OpenSHMEM is an implementation of the OpenSHMEM specification over multiple Networking APIs, including Portals 4, the Open Fabric Interface (OFI), and UCX. Please click on the Wiki tab for help with building and using SOS.

Other

61 stars 53 forks source link

feature: Add multirail support and optional usage of hwloc for optimal NIC assignment #1107

Closed philipmarshall21 closed 4 months ago

philipmarshall21 commented 9 months ago

This PR adds support for utilizing multiple NICs to increase peak RMA bandwidth. Additionally, it adds support for use of hwloc to optimally assign processes to a NIC based on each process' CPU affinity.

philipmarshall21 commented 5 months ago

@wrrobin @markbrown314 @parkerha1 Just wanted to give a little reminder that this branch is ready for review when you have the cycles to give it a look :)