# Faster Neighborhood Attention: Reducing the O(n^2) Cost of Self Attention at the Threadblock Level

arXiv: https://arxiv.org/abs/2403.04690 · HTML: https://mlsanity.com/paper/2403.04690/

## Key facts

- Cited by 14 papers in the map (data through 14 Oct 2026).
- Builds on 20 of its 26 references that are in the map.
- First posted on arXiv 7 Mar 2024.

## Builds on (most cited first, 20 of 20)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [Deep Residual Learning for Image Recognition](https://mlsanity.com/paper/1512.03385/) (2015)
- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [PyTorch: An Imperative Style, High-Performance Deep Learning Library](https://mlsanity.com/paper/1912.01703/) (2019)
- [Scalable Diffusion Models with Transformers](https://mlsanity.com/paper/2212.09748/) (2022)
- [FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness](https://mlsanity.com/paper/2205.14135/) (2022)
- [Longformer: The Long-Document Transformer](https://mlsanity.com/paper/2004.05150/) (2020)
- [Mistral 7B](https://mlsanity.com/paper/2310.06825/) (2023)
- [FlashAttention-2: Faster Attention with Better Parallelism and Work Partitioning](https://mlsanity.com/paper/2307.08691/) (2023)
- [Big Bird: Transformers for Longer Sequences](https://mlsanity.com/paper/2007.14062/) (2020)
- [ViViT: A Video Vision Transformer](https://mlsanity.com/paper/2103.15691/) (2021)
- [Generating Long Sequences with Sparse Transformers](https://mlsanity.com/paper/1904.10509/) (2019)
- [Image Transformer](https://mlsanity.com/paper/1802.05751/) (2018)
- [Stand-Alone Self-Attention in Vision Models](https://mlsanity.com/paper/1906.05909/) (2019)
- [Neighborhood Attention Transformer](https://mlsanity.com/paper/2204.07143/) (2022)
- [Scaling Local Self-Attention for Parameter Efficient Visual Backbones](https://mlsanity.com/paper/2103.12731/) (2021)
- [Online normalizer calculation for softmax](https://mlsanity.com/paper/1805.02867/) (2018)
- [Self-attention Does Not Need O(n2) Memory](https://mlsanity.com/paper/2112.05682/) (2021)
- [Dilated Neighborhood Attention Transformer](https://mlsanity.com/paper/2209.15001/) (2022)
- [Efficient Image Generation with Variadic Attention Heads](https://mlsanity.com/paper/2211.05770/) (2022)

## Built on by (most cited first, 14 of 14)

- [Flex Attention: A Programming Model for Generating Optimized Attention Kernels](https://mlsanity.com/paper/2412.05496/) (2024)
- [Generalized Neighborhood Attention: Multi-dimensional Sparse Attention at the Speed of Light](https://mlsanity.com/paper/2504.16922/) (2025)
- [InvFusion: Bridging Supervised and Zero-shot Diffusion for Inverse Problems](https://mlsanity.com/paper/2504.01689/) (2025)
- [Attention on the Sphere](https://mlsanity.com/paper/2505.11157/) (2025)
- [NAF: Zero-Shot Feature Upsampling via Neighborhood Attention Filtering](https://mlsanity.com/paper/2511.18452/) (2025)
- [WeatherMesh-3: Fast and accurate operational global weather forecasting](https://mlsanity.com/paper/2503.22235/) (2025)
- [Enhancing 3D Semantic Scene Completion with a Refinement Module](https://mlsanity.com/paper/2512.18363/) (2025)
- [LaplacianFormer:Rethinking Linear Attention with Laplacian Kernel](https://mlsanity.com/paper/2604.20368/) (2026)
- [Native Segmentation Vision Transformers](https://mlsanity.com/paper/2505.16993/) (2025)
- [SurGe: Improved Surface Geometry in Point Maps](https://mlsanity.com/paper/2605.31577/) (2026)
- [Optimizing Vision Transformers with Data-Free Knowledge Transfer](https://mlsanity.com/paper/2408.05952/) (2024)
- [NABLA: Neighborhood Adaptive Block-Level Attention](https://mlsanity.com/paper/2507.13546/) (2025)
- [Hilbert-Guided Sparse Local Attention](https://mlsanity.com/paper/2511.05832/) (2025)
- [Le-DETR: Revisiting Real-Time Detection Transformer with Efficient Encoder Design](https://mlsanity.com/paper/2602.21010/) (2026)

## Abstract (arXiv)

Neighborhood attention reduces the cost of self attention by restricting each token's attention span to its nearest neighbors. This restriction, parameterized by a window size and dilation factor, draws a spectrum of possible attention patterns between linear projection and self attention. Neighborhood attention, and more generally sliding window attention patterns, have long been bounded by infrastructure, particularly in higher-rank spaces (2-D and 3-D), calling for the development of custom kernels, which have been limited in either functionality, or performance, if not both. In this work, we aim to massively improve upon existing infrastructure by providing two new methods for implementing neighborhood attention. We first show that neighborhood attention can be represented as a batched GEMM problem, similar to standard attention, and implement it for 1-D and 2-D neighborhood attention. These kernels on average provide 895% and 272% improvement in full precision runtime compared to existing naive CUDA kernels for 1-D and 2-D neighborhood attention respectively. We find that aside from being heavily bound by memory bandwidth, certain inherent inefficiencies exist in all unfused implementations of neighborhood attention, which in most cases undo their theoretical efficiency gain. Motivated by the progress made into fused dot-product attention kernels, we developed fused neighborhood attention; an adaptation of fused dot-product attention kernels that allow fine-grained control over attention across different spatial axes. Known for reducing the quadratic time complexity of self attention to a linear complexity, neighborhood attention can now enjoy a reduced and constant memory footprint, and record-breaking half precision runtime. We observe that our fused implementation successfully circumvents some of the unavoidable inefficiencies in unfused implementations...