aiqbench analysis
OctoPipe addresses the efficiency bottleneck in training heterogeneous large language models by jointly optimizing partitioning, placement, and scheduling to minimize pipeline bubbles. By utilizing a graph-based simulator and an iterative tuner, it achieves significant throughput improvements over state-of-the-art methods that optimize these phases in isolation. This system is particularly relevant for researchers tackling complex model architectures where irregular execution orders and communication-computation overlap are critical constraints.
OctoPipe: Reducing Pipeline Bubbles for Heterogeneous Models via Co-Optimizing Partitioning, Placement, and Scheduling
arXiv:2509.23722v3 Announce Type: replace-cross Abstract: Pipeline parallelism is widely used to train large language models (LLMs). However, increasing heterogeneity in model architectures exacerbates pipeline bubbles, thereby reducing training efficiency. Prior approaches typically optimize a single phase of the pipeline schedule (i.e., partitioning, placement, or scheduling), leaving substantial pipeline bubbles. While promising, co-optimization poses three key challenges: (1) complex performance modeling, (2) a combinatorial search space, and (3) irregular execution orders. To address these challenges, we propose OctoPipe, a pipeline parallelism system to jointly optimize partitioning, placement, and scheduling. First, we build a graph-based pipeline simulator to model heterogeneous pipeline execution for co-optimization. Second, on top of the simulator, we develop an iterative bubble-aware tuner to efficiently explore the combinatorial search space. Third, we implement a unified pipeline executor that dynamically orchestrates computation and communication to support irregular execution orders without deadlocks while maximizing communication-computation overlap. Experiments show that OctoPipe achieves 1.09--1.49$\times$ throughput improvement over the state-of-the-art pipeline parallelism approaches across various heterogeneous model configurations and GPU cluster scales.
Abstract:Pipeline parallelism is widely used to train large language models (LLMs). However, increasing heterogeneity in model architectures exacerbates pipeline bubbles, thereby reducing training efficiency. Prior approaches typically optimize a single phase of the pipeline schedule (i.e., partitioning, placement, or scheduling), leaving substantial pipeline bubbles. While promising, co-optimization poses three key challenges: (1) complex performance modeling, (2) a combinatorial search space, and (3) irregular execution orders. To address these challenges, we propose OctoPipe, a pipeline parallelism system to jointly optimize partitioning, placement, and scheduling. First, we build a graph-based pipeline simulator to model heterogeneous pipeline execution for co-optimization. Second, on top of the simulator, we develop an iterative bubble-aware tuner to efficiently explore the combinatorial search space. Third, we implement a unified pipeline executor that dynamically orchestrates computation and communication to support irregular execution orders without deadlocks while maximizing communication-computation overlap. Experiments show that OctoPipe achieves 1.09--1.49$\times$ throughput improvement over the state-of-the-art pipeline parallelism approaches across various heterogeneous model configurations and GPU cluster scales.
Distributed, Parallel, and Cluster Computing (cs.DC); Artificial Intelligence (cs.AI)
arXiv:2509.23722 [cs.DC]
(or arXiv:2509.23722v3 [cs.DC] for this version)
https://doi.org/10.48550/arXiv.2509.23722 Focus to learn more arXiv-issued DOI via DataCite
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.
Full text under the publisher's open licence · 261 words
Related coverage
Scaling MoE reinforcement learning on Amazon EKS with EFA and DeepEP with 40% more throughput
AWS · Sep 25, 2026
AI Model Co-Design: Hardware-Friendly LLM Design
NVIDIA · Jul 10, 2026
News
Quantum Computing Report · Jul 25, 2026
VSA: Accelerating Video Diffusion Inference with Sparse Attention on AMD GPUs
AMD · Aug 4, 2026