DiSC: Resolution-Scalable Acceleration of Diffusion Models by Exploiting Sparsity and Cached Token Reuse with Hash-based Distribution
Fuente:
arXiv
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866910255096528896 |
|---|---|
| author | Yoon, Jieon Lee, Hangyeol Heo, Jaehoon Kim, Joo-Young |
| author_facet | Yoon, Jieon Lee, Hangyeol Heo, Jaehoon Kim, Joo-Young |
| contents | Transformer-based diffusion models offer superior scalability and performance but suffer from high computational overhead due to the iterative nature and quadratic complexity of self-attention at high resolutions. In this paper, we propose DiSC, a resolution-scalable, sparsity-aware hardware accelerator. At the software level, DiSC introduces two algorithms: Cached Token Reuse (CTR), and Softmax Thresholding with Sparsity Mask Reuse (ST). CTR introduces a mechanism that translates spatial variations in the input latent difference across steps into a token-level reuse decision, effectively eliminating redundant token computation. ST induces sparsity in attention operations by reusing a generated sparsity pattern, leveraging temporal similarity to bypass costly prediction overhead. Together, these algorithms provide resolution-scalable computational benefits and yield a moderate sparsity and hybrid dense-sparse workload.
To exploit this efficiently, we design a specialized hardware architecture and unified dataflow. This architecture avoids dedicated sparsity-handling components; instead, a hash-based distribution over on-chip memory banks allows DiSC to reuse its existing compute engines for sparse operations, efficiently exploiting the induced sparsity with minimal hardware overhead. Evaluated on DiT and PixArt-Sigma, DiSC achieves 3.47-4.74x and 2.48-3.50x speedups over NVIDIA A100 and H100 GPUs, respectively, with energy savings ranging from 46.4% to 68.1%. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_25798 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | DiSC: Resolution-Scalable Acceleration of Diffusion Models by Exploiting Sparsity and Cached Token Reuse with Hash-based Distribution Yoon, Jieon Lee, Hangyeol Heo, Jaehoon Kim, Joo-Young Hardware Architecture Transformer-based diffusion models offer superior scalability and performance but suffer from high computational overhead due to the iterative nature and quadratic complexity of self-attention at high resolutions. In this paper, we propose DiSC, a resolution-scalable, sparsity-aware hardware accelerator. At the software level, DiSC introduces two algorithms: Cached Token Reuse (CTR), and Softmax Thresholding with Sparsity Mask Reuse (ST). CTR introduces a mechanism that translates spatial variations in the input latent difference across steps into a token-level reuse decision, effectively eliminating redundant token computation. ST induces sparsity in attention operations by reusing a generated sparsity pattern, leveraging temporal similarity to bypass costly prediction overhead. Together, these algorithms provide resolution-scalable computational benefits and yield a moderate sparsity and hybrid dense-sparse workload. To exploit this efficiently, we design a specialized hardware architecture and unified dataflow. This architecture avoids dedicated sparsity-handling components; instead, a hash-based distribution over on-chip memory banks allows DiSC to reuse its existing compute engines for sparse operations, efficiently exploiting the induced sparsity with minimal hardware overhead. Evaluated on DiT and PixArt-Sigma, DiSC achieves 3.47-4.74x and 2.48-3.50x speedups over NVIDIA A100 and H100 GPUs, respectively, with energy savings ranging from 46.4% to 68.1%. |
| title | DiSC: Resolution-Scalable Acceleration of Diffusion Models by Exploiting Sparsity and Cached Token Reuse with Hash-based Distribution |
| topic | Hardware Architecture |
| url | https://arxiv.org/abs/2605.25798 |