Distributed LLM Serving on Consumer-Grade GPUs by Reconciling Computation and Communication (2025.findings-emnlp)
Copied to clipboard
Lewei Jin, Kui Zhang, Yongqi Chen, null Zhuoyifan, Renjie Li, Yi Gao, Bowei Yang, Zhengong Cai, Wei Dong
| Challenge: | Large language models are reshaping internet services, and serving them is costly. |
| Approach: | They propose an efficient distributed LLM serving system that splits prefill and decode requests into smaller chunks . |
| Outcome: | The proposed system reduces TTFT, TPOT, and latency compared to the state-of-the-art system. |
Similar Papers
ScaleLLM: A Resource-Frugal LLM Serving Framework by Optimizing End-to-End Efficiency (2024.emnlp-industry)
Copied to clipboard
Yuhang Yao, Han Jin, Alay Shah, Shanshan Han, Zijian Hu, Dimitris Stripelis, Yide Ran, Zhaozhuo Xu, Salman Avestimehr, Chaoyang He
| Challenge: | Large language models (LLMs) are widely used in commercial applications . low latency is crucial due to system latency, query concurrency, and computational resources constraints. |
| Approach: | They propose a system that can be resource-efficiently served by addressing bottlenecks beyond LLM inference . they propose 4.3 speed up over vLLM and 1.5 higher throughput . |
| Outcome: | The proposed system outperforms state-of-the-arts with 1.5 higher throughput . it achieves 4.3 speed up with 64 concurrent requests on Mixtral 8x7B . |
Towards Efficient Large Language Model Serving: A Survey on System-Aware KV Cache Optimization (2026.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) are a powerful tool for high-performance inference serving. |
| Approach: | They focus on system-aware KV infrastructure for serving LLMs . they analyze cross-behavior co-design affinity and behavior-objective links . |
| Outcome: | The proposed key-value (KV) cache is crucial for low-latency, high-throughput LLM inference serving. |
Scaling Down, Serving Fast: Compressing and Deploying Efficient LLMs for Recommendation Systems (2025.emnlp-industry)
Copied to clipboard
Kayhan Behdin, Ata Fatahibaarzi, Qingquan Song, Yun Dai, Aman Gupta, Zhipeng Wang, Hejian Sang, Shao Tang, Gregory Dexter, Sirou Zhu, Siyu Zhu, Tejas Dharamsi, Vignesh Kothapalli, Zhoutong Fu, Yihan Cao, Pin-Lun Hsu, Fedor Borisyuk, Natesh S. Pillai, Luke Simon, Rahul Mazumder
| Challenge: | Large language models (LLMs) have demonstrated remarkable performance across a wide range of industrial applications. |
| Approach: | They propose two techniques for training and deploying small language models that deliver high performance for a variety of industry use cases. |
| Outcome: | The proposed techniques retain much of the quality of larger models while reducing training/serving costs and latency. |
Dovetail: A CPU/GPU Heterogeneous Speculative Decoding for LLM inference (2025.emnlp-main)
Copied to clipboard
| Challenge: | Large language models (LLMs) are demanding more memory and computational resources . however, these devices typically feature weaker GPUs and stronger CPUs . |
| Approach: | They propose a lossless inference acceleration method that leverages the characteristics of heterogeneous devices and the advantages of speculative decoding. |
| Outcome: | The proposed method achieves speedups ranging from 1.79 to 10.1 across different devices . it uses a draft model on the GPU to perform preliminary predictions, while a target model on CPU validates these outputs . |
DiSCo: Device-Server Collaborative LLM-based Text Streaming Services (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have introduced significant cost and quality of experience (QoE) challenges in serving millions of daily requests. |
| Approach: | They propose a device-server cooperative scheduler that optimizes users’ QoE by adaptively routing requests and migrating response generation between endpoints while maintaining cost constraints. |
| Outcome: | Evaluations on real-world workloads show that the proposed scheduler can reduce tail TTFT (11-52%) and mean TTTT (6-78%) while maintaining comparable QoE levels. |
High-Throughput and Memory-Efficient Zeroth-Order Fine-tuning LLMs with Distributed Parallel Computing (2026.findings-acl)
Copied to clipboard
| Challenge: | Zeroth-order (ZO) optimization is a memory-efficient alternative to fine-tuning large language models (LLMs). |
| Approach: | They propose a zeroth-order (ZO) optimization framework that offloads model parameters to CPU memory and overlapping transformer block transfer with dual forward computation on a single GPU. |
| Outcome: | The proposed framework achieves a 3x speedup over ZO2 on an OPT-175B model while maintaining memory efficiency and improving training throughput. |
StitchLLM: Serving LLMs, One Block at a Time (2025.acl-long)
Copied to clipboard
Bodun Hu, Shuozhe Li, Saurabh Agarwal, Myungjin Lee, Akshay Jajoo, Jiamin Li, Le Xu, Geon-Woo Kim, Donghyun Kim, Hong Xu, Amy Zhang, Aditya Akella
| Challenge: | Existing techniques like distillation and pruning are not efficient for large language models. |
| Approach: | They propose a dynamic model routing framework that uses a powerful bottom model to process all queries and a lightweight routing mechanism to allocate computational resources appropriately. |
| Outcome: | The proposed framework improves system throughput while minimizing performance degradation. |
LinguaLinked: Distributed Large Language Model Inference on Mobile Devices (2024.acl-demos)
Copied to clipboard
| Challenge: | Recent research shows that large language models demonstrate enhanced capabilities in various language tasks. |
| Approach: | They introduce a system for decentralized, distributed LLM inference on mobile devices . they use optimized model assignment technique to segment LLMs and linear optimization to align segments with each device . |
| Outcome: | The proposed system performs well on high-end to low-end Android devices. |
TrimLLM: Progressive Layer Dropping for Domain-Specific LLMs (2025.acl-long)
Copied to clipboard
| Challenge: | specialized language models do not show simultaneous memory saving and inference speedup at deployment time. |
| Approach: | They develop a layer-wise specialization technique that reduces the depth of LLMs by progressive layer dropping and compares it to other algorithms for inference. |
| Outcome: | The proposed model retains LLMs’ capacity in specific domains and achieves inference speedup irrespective of hardware and deep learning frameworks. |
LLMs on a Budget? Say HOLA (2025.emnlp-industry)
Copied to clipboard
Zohaib Hasan Siddiqui, Jiechao Gao, Ebad Shabbir, Mohammad Anas Azeez, Rafiq Ali, Gautam Siddharth Kashyap, Usman Naseem
| Challenge: | Current solutions such as quantization, pruning, and Retrieval-Augmented Generation (RAG) offer only partial optimizations and often sacrifice accuracy, speed, or generality. |
| Approach: | They propose an end-to-end optimization framework for efficient LLM deployment . it leverages Hierarchical Speculative Decoding (HSD) for faster inference without quality loss. |
| Outcome: | HOLA delivers +17.6% EMA on GSM8K, +10.5% MCA on ARC, and reduced latency and memory on edge devices like Jetson Nano. |