DiffusionSSD: Achieving Extreme Visual Compression with Dual-stage Diffusion and Cross-level Near-Data Processing
Authors: Kuai Yu, Wen Li, Shengwen Liang, Jie Zhang, Ying Wang, Huawei Li, Xin Yao, Renhai Chen, Gong Zhang
Organizations: Institute of Computing Technology, Chinese Academy of Sciences · University of Chinese Academy of Sciences · School of Artificial Intelligence, Shanxi University · Peking University · Huawei Technologies Co., Ltd
Growing visual data overwhelms storage on resource-constrained edge devices, making compression essential. Traditional codecs offer inadequate rate-distortion performance, motivating the exploration of diffusion codecs. Diffusion codecs face two limitations: (1) At the algorithm level, diffusion decoders preserve perceptual quality under extreme compression, but require substantial computation. (2) At the architecture level, in NPU-centered devices, neural codecs occupy the NPU, interfering with other AI applications. Meanwhile, the complex codec pipeline incurs long I/O paths, leading to additional overhead and energy cost. To address these issues, we propose DiffusionSSD, a cross-level near-data processing framework for extreme visual data compression on edge devices, using an algorithm-architecture co-design method. Specifically, (1) at algorithm level, we present Dual-stage Diffusion, a lightweight solution tailored for diffusion codecs that employs patchification and single-step diffusion, reducing computation cost while preserving perceptual quality; (2) at architecture level, we introduce a Cross-level Near Data Processing architecture that integrates the proposed computational universal flash storage (cUFS) and LPDDR-PIM, enabling codecs to be entirely offloaded to data side while reducing visual-data movement and avoiding SoC redesign; (3) at system level, combining both the algorithm and the underlying architecture, we propose a Patch-Workload co-aware scheduling scheme to enable more efficient system-level execution. Experiments show that Dual-stage Diffusion maintains comparable perceptual quality while reducing computation cost by up to 92.7%. Compared with NPU+DRAM and near-data-processing design such as SmartSSD, cUFS+PIM improves coding throughput by 2.32 times and 2.91 times on average, respectively.
Figures & tables
Fig. 1: Neural codecs offer excellent compression quality but lower throughput. While diffusion-based codecs deliver better perceptual results, their frame rate usually drops below 1 fps. Our approach matches the perceptual quality of diffusion codecs while maintaining an acceptable frame rate.
Fig. 2: Encoder-Decoder Architecture of Neural Codec.
Fig. 3: (a) uMCP overview. (b) The architecture of UFS chip.
Algorithm
Enc.Thpt(FPS)
Dec.Thpt(FPS)
H.264
35.29
387.09
Jpeg
143.69
142.64
balle
21.75
36.51
CDC-30
15.49
1.47
TABLE I: Performance of different codecs
Fig. 4: (a) Performance Collapse: For both traditional and neural approaches, when bpp≤0.5 , the perceptual quality (SSIM) sharply falls. (b) Roofline analysis showing performance bottlenecks across different input resolutions, models, and hardware architectures.
Fig. 6: Diffusion codec pipeline with dual-stage diffusion.
Fig. 7: Cross-level NDP architecture in uMCP with vertically stacked NAND and PIM chips above cUFS, enabled by LPWIO interfaces and VFO-based advanced packaging
Fig. 8: Computational UFS Details
Fig. 9: LPDDR-PIM design
Fig. 10: Patch-Workload Co-aware Scheduling.
Fig. 11: Algorithm performance of different tasks and benchmarks.
Fig. 12: The emulation results and platform configurations.