# SpecPipe: Accelerating Pipeline Parallelism-based LLM Inference with Speculative Decoding

arXiv: https://arxiv.org/abs/2504.04104 · HTML: https://mlsanity.com/paper/2504.04104/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- Placed in Speculative decoding, one of 12 directions in Efficient LLM inference and serving.
- Builds on 31 of its 56 references that are in the map.
- First posted on arXiv 5 Apr 2025.

## Placed in

- Efficient LLM inference and serving / [Speculative decoding](https://mlsanity.com/inference/speculative-decoding/)

## Builds on (most cited first, 30 of 31)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [Training Verifiers to Solve Math Word Problems](https://mlsanity.com/paper/2110.14168/) (2021)
- [Measuring Massive Multitask Language Understanding](https://mlsanity.com/paper/2009.03300/) (2020)
- [Efficient Memory Management for Large Language Model Serving with PagedAttention](https://mlsanity.com/paper/2309.06180/) (2023)
- [OPT: Open Pre-trained Transformer Language Models](https://mlsanity.com/paper/2205.01068/) (2022)
- [TriviaQA: A Large Scale Distantly Supervised Challenge Dataset for Reading Comprehension](https://mlsanity.com/paper/1705.03551/) (2017)
- [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://mlsanity.com/paper/1909.08053/) (2019)
- [Fast Inference from Transformers via Speculative Decoding](https://mlsanity.com/paper/2211.17192/) (2022)
- [DROP: A Reading Comprehension Benchmark Requiring Discrete Reasoning Over Paragraphs](https://mlsanity.com/paper/1903.00161/) (2019)
- [Medusa: Simple LLM Inference Acceleration Framework with Multiple Decoding Heads](https://mlsanity.com/paper/2401.10774/) (2024)
- [Aligning AI With Shared Human Values](https://mlsanity.com/paper/2008.02275/) (2020)
- [EAGLE: Speculative Sampling Requires Rethinking Feature Uncertainty](https://mlsanity.com/paper/2401.15077/) (2024)
- [Taming Throughput-Latency Tradeoff in LLM Inference with Sarathi-Serve](https://mlsanity.com/paper/2403.02310/) (2024)
- [EAGLE-2: Faster Inference of Language Models with Dynamic Draft Trees](https://mlsanity.com/paper/2406.16858/) (2024)
- [SpecInfer: Accelerating Generative Large Language Model Serving with Tree-based Speculative Inference and Verification](https://mlsanity.com/paper/2305.09781/) (2023)
- [LayerSkip: Enabling Early Exit Inference and Self-Speculative Decoding](https://mlsanity.com/paper/2404.16710/) (2024)
- [CodeGeeX: A Pre-Trained Model for Code Generation with Multilingual Benchmarking on HumanEval-X](https://mlsanity.com/paper/2303.17568/) (2023)
- [Break the Sequential Dependency of LLM Inference Using Lookahead Decoding](https://mlsanity.com/paper/2402.02057/) (2024)
- [Unlocking Efficiency in Large Language Model Inference: A Comprehensive Survey of Speculative Decoding](https://mlsanity.com/paper/2401.07851/) (2024)
- [FlashInfer: Efficient and Customizable Attention Engine for LLM Inference Serving](https://mlsanity.com/paper/2501.01005/) (2025)
- [Accelerating LLM Inference with Staged Speculative Decoding](https://mlsanity.com/paper/2308.04623/) (2023)
- [Reward-Guided Speculative Decoding for Efficient LLM Reasoning](https://mlsanity.com/paper/2501.19324/) (2025)
- [FALCON: Honest-Majority Maliciously Secure Framework for Private Deep Learning](https://mlsanity.com/paper/2004.02229/) (2020)
- [SWIFT: On-the-Fly Self-Speculative Decoding for LLM Inference Acceleration](https://mlsanity.com/paper/2410.06916/) (2024)
- [EE-LLM: Large-Scale Training and Inference of Early-Exit Large Language Models with 3D Parallelism](https://mlsanity.com/paper/2312.04916/) (2023)
- [Evaluating Modern GPU Interconnect: PCIe, NVLink, NV-SLI, NVSwitch and GPUDirect](https://mlsanity.com/paper/1903.04611/) (2019)
- [Large Language Models in Cybersecurity: State-of-the-Art](https://mlsanity.com/paper/2402.00891/) (2024)
- [Large language models in medicine: the potentials and pitfalls](https://mlsanity.com/paper/2309.00087/) (2023)
- [PipeInfer: Accelerating LLM Inference using Asynchronous Pipelined Speculation](https://mlsanity.com/paper/2407.11798/) (2024)

## Built on by (most cited first, 4 of 4)

- [Hardware Co-Design Scaling Laws via Roofline Modelling for On-Device LLMs](https://mlsanity.com/paper/2602.10377/) (2026)
- [SpecHop: Continuous Speculation for Accelerating Multi-Hop Retrieval Agents](https://mlsanity.com/paper/2605.21965/) (2026)
- [FASER: Fine-Grained Phase Management for Speculative Decoding in Dynamic LLM Serving](https://mlsanity.com/paper/2604.20503/) (2026)
- [Speculative Pipeline Decoding: Higher-Accuracy Drafting with Hidden Latency via Pipeline Parallelism](https://mlsanity.com/paper/2605.30852/) (2026)

## Abstract (arXiv)

The demand for large language model inference is rapidly increasing. Pipeline parallelism offers a cost-effective deployment strategy for distributed inference but suffers from high service latency. While incorporating speculative decoding to pipeline parallelism improves performance, it still faces challenges of low hardware utilization and narrow speculative window. Inspired by branch prediction in instruction pipelining, we introduce SpecPipe, which fills the pipeline with speculative tokens of a request step-by-step. By maximizing the hardware utilization, SpecPipe decodes one token per pipeline step ideally. Specifically, SpecPipe comprises a dynamic speculative token tree and a pipelined inference framework. The tree dynamically accepts tokens from a speculative token source and outputs the tokens to the inference pipeline. Since the speculative window relaxed in our framework, a high-accuracy draft model is integrated without fine-tuning. The pipeline inference framework follows node-wise computation, pruning propagation, and inter-node communication stages. We implement SpecPipe and a variant SpecPipe-DB with dynamic batching for single- and multi-request inference, respectively. On an 8-stage pipeline, SpecPipe improves time between tokens on diverse single-request workloads by 4.19×-5.53× over standard pipeline parallelism and by 2.08×-2.38× over prior tree-based speculative decoding methods. For multi-request workloads, SpecPipe-DB achieves 1.64×-2.08× higher throughput and 1.61×-2.06× lower time between tokens than vLLM.