SwiftKV: An Edge-Oriented Attention Algorithm and Multi-Head Accelerator for Fast, Efficient LLM Decoding

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Zhang, Junming, Zhang, Qinyan, Sun, Huajun, Gao, Feiyang, Hu, Sheng, Nie, Rui, Miao, Xiangshui
Format: Preprint
Veröffentlicht: 2026
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866908769582055424
author Zhang, Junming
Zhang, Qinyan
Sun, Huajun
Gao, Feiyang
Hu, Sheng
Nie, Rui
Miao, Xiangshui
author_facet Zhang, Junming
Zhang, Qinyan
Sun, Huajun
Gao, Feiyang
Hu, Sheng
Nie, Rui
Miao, Xiangshui
contents Edge acceleration for large language models is crucial for their widespread application; however, achieving fast attention inference and efficient decoding on resource-constrained edge accelerators remains challenging. This paper presents SwiftKV Attention, a per-token pipelined, low-latency single-pass attention inference algorithm, where every (kt, vt) in the KV cache is processed exactly once in a uniform per-token pipeline without score materialization, blockwise softmax, or a second pass, thereby enabling fast execution on edge accelerators with a single hardware set and no resource-intensive parallelism. Furthermore, to address the limited support for multi-head LLM decoding in existing accelerators, we design the SwiftKV-MHA accelerator, which enables high precision attention and low precision GEMV on the same processor array, achieving fast and efficient multi-head parallel decoding. Experimental results show that, on the edge accelerator, the SwiftKV Attention algorithm achieves a 7.16* speedup over native attention and significantly outperforms other attention algorithms. SwiftKV-MHA further reduces attention latency by 13.48*; under the same settings, it improves generation speed by 17.4% and increases token efficiency by 1.98* compared with state-of-the-art works.
format Preprint
id arxiv_https___arxiv_org_abs_2601_10953
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle SwiftKV: An Edge-Oriented Attention Algorithm and Multi-Head Accelerator for Fast, Efficient LLM Decoding
Zhang, Junming
Zhang, Qinyan
Sun, Huajun
Gao, Feiyang
Hu, Sheng
Nie, Rui
Miao, Xiangshui
Hardware Architecture
Edge acceleration for large language models is crucial for their widespread application; however, achieving fast attention inference and efficient decoding on resource-constrained edge accelerators remains challenging. This paper presents SwiftKV Attention, a per-token pipelined, low-latency single-pass attention inference algorithm, where every (kt, vt) in the KV cache is processed exactly once in a uniform per-token pipeline without score materialization, blockwise softmax, or a second pass, thereby enabling fast execution on edge accelerators with a single hardware set and no resource-intensive parallelism. Furthermore, to address the limited support for multi-head LLM decoding in existing accelerators, we design the SwiftKV-MHA accelerator, which enables high precision attention and low precision GEMV on the same processor array, achieving fast and efficient multi-head parallel decoding. Experimental results show that, on the edge accelerator, the SwiftKV Attention algorithm achieves a 7.16* speedup over native attention and significantly outperforms other attention algorithms. SwiftKV-MHA further reduces attention latency by 13.48*; under the same settings, it improves generation speed by 17.4% and increases token efficiency by 1.98* compared with state-of-the-art works.
title SwiftKV: An Edge-Oriented Attention Algorithm and Multi-Head Accelerator for Fast, Efficient LLM Decoding
topic Hardware Architecture
url https://arxiv.org/abs/2601.10953