# Efficiently Serving Large Multimodal Models Using EPD Disaggregation

arXiv: https://arxiv.org/abs/2501.05460 · HTML: https://mlsanity.com/paper/2501.05460/

## Key facts

- Cited by 16 papers in the map (data through 14 Oct 2026).
- Placed in Visual token pruning and efficiency, one of 16 directions in Multimodal LLMs / vision-language models.
- Builds on 22 of its 28 references that are in the map.
- First posted on arXiv 25 Dec 2024.

## Placed in

- Multimodal LLMs / vision-language models / [Visual token pruning and efficiency](https://mlsanity.com/mllm/visual-token-pruning-and-efficiency/)

## Builds on (most cited first, 22 of 22)

- [GPT-4 Technical Report](https://mlsanity.com/paper/2303.08774/) (2023)
- [Visual Instruction Tuning](https://mlsanity.com/paper/2304.08485/) (2023)
- [Efficient Memory Management for Large Language Model Serving with PagedAttention](https://mlsanity.com/paper/2309.06180/) (2023)
- [InternVL: Scaling up Vision Foundation Models and Aligning for Generic Visual-Linguistic Tasks](https://mlsanity.com/paper/2312.14238/) (2023)
- [Video-MME: The First-Ever Comprehensive Evaluation Benchmark of Multi-modal LLMs in Video Analysis](https://mlsanity.com/paper/2405.21075/) (2024)
- [A Survey on Evaluation of Large Language Models](https://mlsanity.com/paper/2307.03109/) (2023)
- [MiniCPM-V: A GPT-4V Level MLLM on Your Phone](https://mlsanity.com/paper/2408.01800/) (2024)
- [SnapKV: LLM Knows What You are Looking for Before Generation](https://mlsanity.com/paper/2404.14469/) (2024)
- [NExT-QA:Next Phase of Question-Answering to Explaining Temporal Actions](https://mlsanity.com/paper/2105.08276/) (2021)
- [DistServe: Disaggregating Prefill and Decoding for Goodput-optimized Large Language Model Serving](https://mlsanity.com/paper/2401.09670/) (2024)
- [Splitwise: Efficient generative LLM inference using phase splitting](https://mlsanity.com/paper/2311.18677/) (2023)
- [Mooncake: A KVCache-centric Disaggregated Architecture for LLM Serving](https://mlsanity.com/paper/2407.00079/) (2024)
- [Multimodal Large Language Models: A Survey](https://mlsanity.com/paper/2311.13165/) (2023)
- [SARATHI: Efficient LLM Inference by Piggybacking Decodes with Chunked Prefills](https://mlsanity.com/paper/2308.16369/) (2023)
- [GEAR: An Efficient KV Cache Compression Recipe for Near-Lossless Generative Inference of LLM](https://mlsanity.com/paper/2403.05527/) (2024)
- [Inference without Interference: Disaggregate LLM Inference for Mixed Downstream Workloads](https://mlsanity.com/paper/2401.11181/) (2024)
- [InstaHide: Instance-hiding Schemes for Private Distributed Learning](https://mlsanity.com/paper/2010.02772/) (2020)
- [A Survey of Resource-efficient LLM and Multimodal Foundation Models](https://mlsanity.com/paper/2401.08092/) (2024)
- [D'ej`aVu: KV-cache Streaming for Fast, Fault-tolerant Generative LLM Serving](https://mlsanity.com/paper/2403.01876/) (2024)
- [P/D-Serve: Serving Disaggregated Large Language Model at Scale](https://mlsanity.com/paper/2408.08147/) (2024)
- [Inf-MLLM: Efficient Streaming Inference of Multimodal Large Language Models on a Single GPU](https://mlsanity.com/paper/2409.09086/) (2024)
- [IPProtect: protecting the intellectual property of visual datasets during data valuation](https://mlsanity.com/paper/2212.11468/) (2022)

## Built on by (most cited first, 16 of 16)

- [ElasticMM: Efficient Multimodal LLMs Serving with Elastic Multimodal Parallelism](https://mlsanity.com/paper/2507.10069/) (2025)
- [Arrow: Adaptive Scheduling Mechanisms for Disaggregated LLM Inference Architecture](https://mlsanity.com/paper/2505.11916/) (2025)
- [Cornfigurator: Automated Planning for Any-to-Any Multimodal Model Serving](https://mlsanity.com/paper/2512.14098/) (2025)
- [xLLM Technical Report](https://mlsanity.com/paper/2510.14686/) (2025)
- [VoxServe: Streaming-Centric Serving System for Speech Language Models](https://mlsanity.com/paper/2602.00269/) (2026)
- [FlowPrefill: Decoupling Preemption from Prefill Scheduling Granularity to Mitigate Head-of-Line Blocking in LLM Serving](https://mlsanity.com/paper/2602.16603/) (2026)
- [M*: A Modular, Extensible, Serving System for Multimodal Models](https://mlsanity.com/paper/2606.12688/) (2026)
- [Coordinated Scheduling for MoE LLM Serving](https://mlsanity.com/paper/2606.15177/) (2026)
- [Enabling Disaggregated Multi-Stage MLLM Inference via GPU-Internal Scheduling and Resource Sharing](https://mlsanity.com/paper/2512.17574/) (2025)
- [SLO-Aware Compute Resource Allocation for Prefill-Decode Disaggregated LLM Inference](https://mlsanity.com/paper/2603.04716/) (2026)
- [Cornserve: A Distributed Serving System for Any-to-Any Multimodal Models](https://mlsanity.com/paper/2603.12118/) (2026)
- [From Models to Operators: Rethinking Autoscaling Granularity for Large Generative Models](https://mlsanity.com/paper/2511.02248/) (2025)
- [OmniInfer: System-Wide Acceleration Techniques for Optimizing LLM Serving Throughput and Latency](https://mlsanity.com/paper/2511.22481/) (2025)
- [QuickGrasp: Responsive Video-Language Querying Service via Accelerated Tokenization and Edge-Augmented Inference](https://mlsanity.com/paper/2603.00126/) (2026)
- [Cost-Efficient Multimodal LLM Inference via Cross-Tier GPU Heterogeneity](https://mlsanity.com/paper/2603.12707/) (2026)
- [OpWeave: Flexible Operator Disaggregation for Heterogeneous LLM Serving](https://mlsanity.com/paper/2609.14237/) (2026)

## Abstract (arXiv)

Large Multimodal Models (LMMs) extend Large Language Models (LLMs) by handling diverse inputs such as images, audio, and video, but at the cost of adding a multimodal encoding stage that increases both computational and memory overhead. This step negatively affects key Service Level Objectives (SLOs), such as time to first token (TTFT) and time per output token (TPOT). We introduce Encode-Prefill-Decode (EPD) Disaggregation, a novel framework that separates the encoding, prefill, and decode stages onto dedicated resources. Unlike current systems, which bundle encoding and prefill together, our approach decouples these steps, unlocking new opportunities and optimizations. These include a mechanism to cache multimedia tokens for efficient transfer, a novel way to parallelize the encoding load within a request, a module for optimal resource allocation for disaggregated serving, and a novel role-switching method to handle changing workload characteristics. Experimental evaluations with popular LMMs show substantial gains in memory efficiency (up to 15x lower peak memory utilization), batch sizes (up to 22x larger), 10x more images per request, and 2.2x larger KV caches. Furthermore, it leads to significant improvements in SLO attainment (up to 90-100% improvement) and TTFT (up to 71% reduction), compared to systems that do not disaggregate. The code is available at https://github.com/vbdi/epdserve.