The size and compute capacity of high performance computing (HPC) clusters continues to grow at a breakneck pace, with the realization of Exascale computing firmly within our grasp in the next few years. However, the obstacles to Exascale are daunting—CPU performance, power efficiency, reliability, memory, and storage are just a few.
With the Intel® OPA mechanism driver for Neutron , Intel® OPA hardware can be automatically reconfigured/managed to enable the high-performance fabric required for high-performance computing workloads. Furthermore, bare-metal provisioning can take advantage of this compatibility layer at Neutron for the automated attachment HPC nodes to the fabric.
This presentation will walk the audience through how OpenStack can be deployed in an HPC context thanks to the Intel® OPA mechanism driver for Neutron (via the networking-omnipath project). The following will be covered during this talk:
- introduction of Intel® OPA basics and hardware
- summary and architecture of the changes made in OpenStack
- demo
- future work
