Large language models (LLMs) require increasing memory capacity to accommodate growing model weights and KV caches. High-Bandwidth Flash (HBF) offers high memory density and aggregate read bandwidth through massive plane-level parallelism, making it an attractive option for LLM serving. However, serving LLMs entirely f...
Shu-Zhang Zhong, Wei-Kai Xu, Yi-Fan Zhou et al.· 1 citation
A faster storage device should make serving faster. We find the opposite. High-Bandwidth Flash (HBF) stacks NAND behind a wide, package-local interface, promising flash-scale capacity with far lower read latency and higher bandwidth than an SSD. The obvious move is to keep an SSD-style Mooncake KV-offloading stack and...
Zhuo-Ran Li, Zhuo-Hang Bian, Xin Huang et al.· 5 citations
HDA-MoE is presented, a framework that optimizes MoE execution on NMP architectures through hybrid parallel deployment and runtime scheduling and integrates an offline hybrid parallel mapping algorithm with an online dynamic and adaptive scheduling mechanism to reduce communication overhead while improving computation...
Hao-Chen Huang, Shu-Zhang Zhong, Sheng-Xuan Qiu et al.· IEEE Transactions on Compute...· 0 citations
Deploying Video Diffusion Models (VDMs) on edge devices is appealing for localized and privacy-preserving generation, but their iterative Transformer-based denoising remains too slow for practical local inference. Cross-Timestep Caching (CTC) has emerged as a promising direction for reducing redundant computation, reus...
