DD-$\alpha$AMG on QPACE 3

TL;DR

This paper details the porting and performance evaluation of the DD-$\alpha$AMG solver from QPACE 2 to QPACE 3, highlighting code adaptation, communication library changes, and scalability results.

Contribution

It provides a comprehensive account of adapting the DD-$\alpha$AMG solver to new hardware and communication infrastructure, demonstrating near-ideal scalability and performance.

Findings

Speedup close to theoretical expectations

Successful porting from Knights Corner to Knights Landing

Effective scaling on multiple nodes

Abstract

We describe our experience porting the Regensburg implementation of the DD-$\alpha$AMG solver from QPACE 2 to QPACE 3. We first review how the code was ported from the first generation Intel Xeon Phi processor (Knights Corner) to its successor (Knights Landing). We then describe the modifications in the communication library necessitated by the switch from InfiniBand to Omni-Path. Finally, we present the performance of the code on a single processor as well as the scaling on many nodes, where in both cases the speedup factor is close to the theoretical expectations.