Abstract
The technology relates to a software framework providing for the simultaneous co-design of network topologies and collective communication algorithms in distributed computing environments. The software framework utilizes a graph-theory-based model to represent computing nodes and physical communication links, encoding hardware and software parameters into a solver-readable mathematical formulation. By applying combinatorial optimization solvers, the software framework identifies highly efficient physical cabling schemas and corresponding algorithmic routing schedules. To handle highly complex search spaces, the approach leverages graph automorphisms and symmetries in collective operations to significantly reduce the combinatorial search space. This enables the generation of tailored interconnect architectures and message-passing protocols for arbitrary or irregular topologies without relying on rigid, predefined network structures.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
N/A, "SYMMETRY-AWARE OPTIMIZATION FRAMEWORK FOR NETWORK TOPOLOGY AND COLLECTIVE ALGORITHM CO-DESIGN", Technical Disclosure Commons, (April 01, 2026)
https://www.tdcommons.org/dpubs_series/9693