From ATOP to ZCube: Automated Topology Optimization Pipeline and A Highly Cost-Effective Network Topology for Large Model Training
Open Access
- 27 August 2025
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 861-881
Abstract
No abstract availableThis publication has 26 references indexed in Scilit:
- Rail-only: A Low-Cost High-Performance Network for Training LLMs with Trillion Parameters2024
- xTrimoPGLM: Unified 100B-Scale Pre-trained Transformer for Deciphering the Language of ProteinbioRxiv, 2023
- A survey on multi-objective hyperparameter optimization algorithms for machine learningArtificial Intelligence Review, 2022
- Systematic Topology Design for Large-Scale Networks: A Unified Framework2020
- Versatile, scalable, and accurate simulation of distributed applications and platformsJournal of Parallel and Distributed Computing, 2014
- On the impact of packet spraying in data center networks2013
- REWIRE: An optimization-based framework for unstructured data center network design2012
- Multiobjective optimization using evolutionary algorithms — A comparative case study1998
- Pareto optimality in multiobjective problemsApplied Mathematics & Optimization, 1977
- A Study of Non-Blocking Switching NetworksBell System Technical Journal, 1953