Jan 13, 2022 · We study the impact of different routing mechanisms on the All-reduce algorithm, and their sensitivity to topology size and message size. Our ...

In this paper, we propose g-PAARD, a general proximity-aware All-reduce communication on the Dragonfly network. We study the impact of different routing ...

We evaluate their performance through event-driven simulation for small-and large-sized broadcasts (in terms of both data size and number of processes). We ...

Abstract—Two-tiered direct network topologies such as Drag- onflies have been proposed for future post-petascale and exascale.

Sep 12, 2016 · We study the effect of different routing mechanisms on the topology-aware collective algorithms, as well as their sensitivity to network ...

To this end, this paper introduces. TACOS, an automated synthesizer that generates topology-aware collective algorithms for common distributed machine learning.

Nov 21, 2016 · Design and Evaluation of Topology-aware Scatter and All-. Gather Algorithms for Dragonfly Networks. Supercomputing 2016, Nov 2016, Salt Lake ...

cution of all algorithms on a Dragonfly network using CODES [8], a network simulation framework that has already demonstrated its accuracy [6, 7]. We ...

This paper offers a first attempt to design an efficient all- reduce algorithm on dragonfly networks, based on previous algorithms which have been found to be ...

We evaluate each optimization through CODES network simulation framework with minimal, non-minimal, and adaptive routing. The simulation results demonstrate ...