paper-with-me

Papers

Comet: Fine-grained Computation-communication Overlapping for Mixture-of-Experts

2025-02-27 · Shulai Zhang, Ningxin Zheng, Haibin Lin, Ziheng Jiang, Wenlei Bao, Chengquan Jiang, Qi Hou, Weihao Cui, Size Zheng, Li-Wen Chang, Quan Chen, Xin Liu

Mixture-of-experts (MoE) has been extensively employed to scale large language models to trillion-plus parameters while maintaining a fixed computational cost. The development of large MoE models in the distributed scenario encounters the problem of large communication overhead. The inter-device communication of a MoE layer can occupy 47% time of the entire model execution with popular models and frameworks. Therefore, existing methods suggest the communication in a MoE layer to be pipelined with the computation for overlapping. However, these coarse grained overlapping schemes introduce a notable impairment of computational efficiency and the latency concealing is sub-optimal. To this end, we present COMET, an optimized MoE system with fine-grained communication-computation overlapping. Leveraging data dependency analysis and task rescheduling, COMET achieves precise fine-grained overlapping of communication and computation. Through adaptive workload assignment, COMET effectively eliminates fine-grained communication bottlenecks and enhances its adaptability across various scenarios. Our evaluation shows that COMET accelerates the execution of a single MoE layer by $1.96\times$ and for end-to-end execution, COMET delivers a $1.71\times$ speedup on average. COMET has been adopted in the production environment of clusters with ten-thousand-scale of GPUs, achieving savings of millions of GPU hours.

📄 PDF Abstract BibTeX arXiv:2502.19811

Code (4)

bytedance/flux 공식 구현 pytorch
multi-swe-bench/magentless
multi-swe-bench/mopenhands
multi-swe-bench/mswe-agent

Tasks

Computational EfficiencyGPUMixture-of-Experts

Methods 이 논문이 사용한 방법론

MoE 설명 없음

Similar Papers 제목 키워드 기반

MGG: Accelerating Graph Neural Networks with Fine-grained intra-kernel Communication-Computation Pipelining on Multi-GPU Platforms

2022-09-14 · yuke wang, Boyuan Feng, Zheng Wang, Tong Geng 외

The increasing size of input graphs for graph neural networks (GNNs) highlights the demand for using multi-GPU platforms. However, existing multi-GPU GNN systems optimize the computation and communication individually ba…

GPULayout DesignManagement

Domino: Eliminating Communication in LLM Training via Generic Tensor Slicing and Overlapping

2024-09-23 · Guanhua Wang, Chengming Zhang, Zheyu Shen, Ang Li 외

Given the popularity of generative AI, Large Language Models (LLMs) often consume hundreds or thousands of GPUs for parallelizing and accelerating the training process. Communication overhead becomes more pronounced when…

Comet: A Communication-efficient and Performant Approximation for Private Transformer Inference

2024-05-24 · Xiangrui Xu, Qiao Zhang, Rui Ning, Chunsheng Xin 외

The prevalent use of Transformer-like models, exemplified by ChatGPT in modern language processing applications, underscores the critical need for enabling private inference essential for many cloud-based services relian…

Privacy Preserving

More Experts Than Galaxies: Conditionally-overlapping Experts With Biologically-Inspired Fixed Routing

2024-10-10 · Sagi Shaier, Francisco Pereira, Katharina von der Wense, Lawrence E Hunter 외

The evolution of biological neural systems has led to both modularity and sparse coding, which enables energy efficiency and robustness across the diversity of tasks in the lifespan. In contrast, standard neural networks…

image-classificationImage ClassificationLanguage ModelingLanguage Modelling+2

Comet: Accelerating Private Inference for Large Language Model by Predicting Activation Sparsity

2025-05-12 · Guang Yan, Yuhui Zhang, Zimu Guo, Lutan Zhao 외

With the growing use of large language models (LLMs) hosted on cloud platforms to offer inference services, privacy concerns about the potential leakage of sensitive information are escalating. Secure multi-party computa…

Language ModelingLanguage ModellingLarge Language Model