APECAST

Innovative methodology for the acceleration of collective data transfer operations in the context of High Performance Computing in general and parallel computing in particular. The purpose of this project is twofold and involves: the implementation and finalization of some details of the communication mechanisms described in the patent;

the prototype update on the latest FPGA platforms.

The proposed system leverages the advantages of a dedicated high-performance network suitable for HPC applications, to accelerate a specific type of collective communications (multicast, broadcast, all-to-all in all its variants) in a manner closely linked to the use of the RDMA protocol, a feature absent or differently implemented in communication technologies aligned with the current state of the art.