English

QUILL: An Algorithm-Architecture Co-Design for Cache-Local Deformable Attention

Hardware Architecture 2025-11-18 v1 Computer Vision and Pattern Recognition Machine Learning

Abstract

Deformable transformers deliver state-of-the-art detection but map poorly to hardware due to irregular memory access and low arithmetic intensity. We introduce QUILL, a schedule-aware accelerator that turns deformable attention into cache-friendly, single-pass work. At its core, Distance-based Out-of-Order Querying (DOOQ) orders queries by spatial proximity; the look-ahead drives a region prefetch into an alternate buffer--forming a schedule-aware prefetch loop that overlaps memory and compute. A fused MSDeformAttn engine executes interpolation, Softmax, aggregation, and the final projection (W''m) in one pass without spilling intermediates, while small tensors are kept on-chip and surrounding dense layers run on integrated GEMMs. Implemented as RTL and evaluated end-to-end, QUILL achieves up to 7.29x higher throughput and 47.3x better energy efficiency than an RTX 4090, and exceeds prior accelerators by 3.26-9.82x in throughput and 2.01-6.07x in energy efficiency. With mixed-precision quantization, accuracy tracks FP32 within <=0.9 AP across Deformable and Sparse DETR variants. By converting sparsity into locality--and locality into utilization--QUILL delivers consistent, end-to-end speedups.

Keywords

Cite

@article{arxiv.2511.13679,
  title  = {QUILL: An Algorithm-Architecture Co-Design for Cache-Local Deformable Attention},
  author = {Hyunwoo Oh and Hanning Chen and Sanggeon Yun and Yang Ni and Wenjun Huang and Tamoghno Das and Suyeon Jang and Mohsen Imani},
  journal= {arXiv preprint arXiv:2511.13679},
  year   = {2025}
}

Comments

Accepted to DATE 2026

R2 v1 2026-07-01T07:41:46.628Z