Optimizing Allreduce Operations for Modern Heterogeneous Architectures with Multiple Processes per GPU

[2508.13397] Optimizing Allreduce Operations for Modern Heterogeneous Architectures with Multiple Processes per GPU Skip to main content Search Submit Donate Log in Computer Science > Distributed, Parallel, and Cluster Computing arXiv:2508.13397 (cs) [Submitted on 18 Aug 2025 ( v1 ), last revised 24 Aug 2026 (this version, v3)] Title: Optimizing Allreduce Operations for Modern Heterogeneous Architectures with Multiple Processes per GPU Authors: Michael Adams , Amanda Bienz View PDF HTML (experimental) Abstract: Large inter-GPU all-reduce operations, prevalent throughout deep learning, are bottlenecked by communication costs. Emerging heterogeneous architectures are comprised of complex nodes, often containing $4$ GPUs and dozens to hundreds of CPU cores per node. Parallel applications are typically accelerated on the available GPUs, using only a single CPU core per GPU while the remaining cores sit idle. This paper presents novel optimizations to large GPU-aware all-reduce operations by extending the lane-aware algorithm to heterogeneous architectures and notably using multiple CPU cores per GPU to accelerate these operations. Using GPUDirect RDMA and host copy communications respectively, these multi-CPU-accelerated GPU-aware all-reduces yield speedups over system MPI of up to $3$x on LLNL's Tuolumne supercomputer and up to $2.45$x for large MPI all-reduces across the NVIDIA A100 GPUs of NCSA's Delta supercomputer. Subjects: Distributed, Parallel, and Cluster Computing (cs.DC) Cite as: arXiv:2508.13397 [cs.DC] (or arXiv:2508.13397v3 [cs.DC] for this version) https://doi.org/10.48550/arXiv.2508.13397 Focus to learn more arXiv-issued DOI via DataCite Journal reference: M. Adams and A. Bienz, "Optimizing Allreduce Operations for Modern Heterogeneous Architectures with Multiple Processes per GPU," 2026 IEEE International Parallel and Distributed Processing Symposium Workshops (IPDPSW), pp. 135-144 Related DOI : https://doi.org/10.1109/IPDPSW71298.2026.00023 Focus to learn more DOI(s) linking to related resources Submission history From: Michael Adams [ view email ] [v1] Mon, 18 Aug 2025 23:12:04 UTC (494 KB) [v2] Tue, 24 Feb 2026 19:49:23 UTC (1,533 KB) [v3] Mon, 24 Aug 2026 19:21:57 UTC (1,684 KB) Full-text links: Access Paper: View PDF HTML (experimental) TeX Source view license Additional Features Audio Summary Current browse context: cs.DC < prev | next > new | recent | 2025-08 Change to browse by: cs References & Citations SciX Google Scholar Semantic Scholar export BibTeX citation Bookmark Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer ( What is the Explorer? ) Connected Papers Toggle Connected Papers ( What is Connected Papers? ) Litmaps Toggle Litmaps ( What is Litmaps? ) scite.ai Toggle scite Smart Citations ( What are Smart Citations? ) Code, Data, Media Code, Data and Media Associated with this Article alphaXiv Toggle alphaXiv ( What is alphaXiv? ) Links to Code Toggle CatalyzeX Code Finder for Papers ( What is CatalyzeX? ) DagsHub Toggle DagsHub ( What is DagsHub? ) GotitPub Toggle Gotit.pub ( What is GotitPub? ) Huggingface Toggle Hugging Face ( What is Huggingface? ) ScienceCast Toggle ScienceCast ( What is ScienceCast? ) Demos Demos Replicate Toggle Replicate ( What is Replicate? ) Spaces Toggle Hugging Face Spaces ( What is Spaces? ) Spaces Toggle TXYZ.AI ( What is TXYZ.AI? ) Related Papers Recommenders and Search Tools Link to Influence Flower Influence Flower ( What are Influence Flowers? ) Core recommender toggle CORE Recommender ( What is CORE? ) Author Venue Institution Topic About arXivLabs arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs . Which authors of this paper are endorsers? | Disable MathJax ( What is MathJax? ) We gratefully acknowledge support from our major funders , member institutions , and all contributors. About · Help · Contact · Subscribe · Copyright · Privacy · Accessibility · Operational Status (opens in new tab)

Related documents

• Optimizing Allreduce Operations for Modern Heterogeneous Architectures with Multiple Processes per GPU#670369
• FliX: Flipped-Indexing for Scalable GPU Queries and Updates#120487
• What Irregularity Costs: CUDA C++, Rust, and Triton on a Hash-Blocked GPU Workload#441766
• Graph Traversal on Tensor Cores: A BFS Framework for Modern GPUs#259426
• KernelArc: A Multi-Agent Framework for GPU Kernel Optimization#658877