paper-with-me

홈 › Papers

Accelerating MoE Model Inference with Expert Sharding

2025-03-11 · Oana Balmau, Anne-Marie Kermarrec, Rafael Pires, André Loureiro Espírito Santo, Martijn de Vos, Milos Vujasinovic

Mixture of experts (MoE) models achieve state-of-the-art results in language modeling but suffer from inefficient hardware utilization due to imbalanced token routing and communication overhead. While prior work has focused on optimizing MoE training and decoder architectures, inference for encoder-based MoE models in a multi-GPU with expert parallelism setting remains underexplored. We introduce MoEShard, an inference system that achieves perfect load balancing through tensor sharding of MoE experts. Unlike existing approaches that rely on heuristic capacity factors or drop tokens, MoEShard evenly distributes computation across GPUs and ensures full token retention, maximizing utilization regardless of routing skewness. We achieve this through a strategic row- and column-wise decomposition of expert matrices. This reduces idle time and avoids bottlenecks caused by imbalanced expert assignments. Furthermore, MoEShard minimizes kernel launches by fusing decomposed expert computations, significantly improving throughput. We evaluate MoEShard against DeepSpeed on encoder-based architectures, demonstrating speedups of up to 6.4$\times$ in time to first token (TTFT). Our results show that tensor sharding, when properly applied to experts, is a viable and effective strategy for efficient MoE inference.

📄 PDF Abstract BibTeX arXiv:2503.08467

Code (0)

등록된 구현이 없습니다.

Tasks

DecoderGPULanguage ModelingLanguage ModellingMixture-of-Expertsmodel

Methods 이 논문이 사용한 방법론

MoE 설명 없음

Similar Papers 제목 키워드 기반

Efficient, VRAM-Constrained xLM Inference on Clients

2026-04-29 · Aditya Ukarande, Deep Shekhar, Marc Blackstein, Ram Rangan arxiv

To usher in the next round of client AI innovation, there is an urgent need to enable efficient, lossless inference of high-accuracy large language models (LLMs) and vision language models (VLMs), jointly referred to as …

Adjoint sharding for very long context training of state space models

2025-01-01 · Xingzi Xu, Amir Tavanaei, Kavosh Asadi, Karim Bouyarmane

Despite very fast progress, efficiently training large language models (LLMs) in very long contexts remains challenging. Existing methods fall back to training LLMs with short contexts (a maximum of a few thousands token…

GPULarge Language ModelState Space Models

Seesaw: High-throughput LLM Inference via Model Re-sharding

2025-03-09 · Qidong Su, Wei Zhao, Xin Li, Muralidhar Andoorveedu 외

To improve the efficiency of distributed large language model (LLM) inference, various parallelization strategies, such as tensor and pipeline parallelism, have been proposed. However, the distinct computational characte…

Computational EfficiencyLanguage ModelingLanguage ModellingLarge Language Model+1

Learning to Shard: RL for Co-optimizing the Parallelism Degrees and Per-operator Sharding Dimensions in Distributed LLM Inference

2025-08-29 · Ruokai Yin, Sattwik Deb Mishra, Xuan Zuo, Hokchhay Tann 외 arxiv

Distributed LLM inference requires careful coordination of parallelization strategies across hundreds to thousands of NPUs to meet production SLOs. Current systems like Megatron-LM rely on static heuristics that separate…

Automap: Towards Ergonomic Automated Parallelism for ML Models

2021-12-06 · Michael Schaarschmidt, Dominik Grewe, Dimitrios Vytiniotis, Adam Paszke 외

The rapid rise in demand for training large neural network architectures has brought into focus the need for partitioning strategies, for example by using data, model, or pipeline parallelism. Implementing these methods …