aiqbench analysis
Adaptive DHiSparse addresses the critical bottleneck of KV memory exhaustion in long-context LLM serving by integrating speculative decoding with a load-adaptive hierarchical storage strategy. By dynamically managing residency between host and GPU memory, the system significantly boosts concurrency and throughput without compromising latency bounds, offering a practical path to scaling inference beyond current hardware limits.
Adaptive DHiSparse: Speculative Decoding over a Load-Adaptive KV Hierarchy
arXiv:2610.13708v1 Announce Type: cross Abstract: Long-context serving runs out of KV memory before the hardware reaches its useful throughput. HiSparse stores history in host memory and fetches only the selected entries, but it uses one cache budget at every load and does not handle draft-model speculative decoding such as DSpark and DFlash2. We present Adaptive DHiSparse, which supports speculative decoding over hierarchical sparse KV and adapts residency to load. It protects the required working set, retains additional history while GPU memory is spare, and recovers the rest through exact fused retrieval. DSpark and DFlash2 keep their proposal and acceptance rules: queries share missing-row retrieval, draft KV uses a window-bounded ring, and verification staging is reused across layer groups. At 100K context and concurrency 128, GLM-5.3 with DFlash2 delivers 793 output tokens/s per decode GPU, 1.45$\times$ HiSparse and 1.27$\times$ non-speculative adaptive serving. Held to the same latency bounds as HiSparse, it sustains up to 4$\times$ the concurrency and 2.3 to 3.7$\times$ the output, depending on the per-token bound. On DeepSeek-V4.1-Flash at 500K context and concurrency 512, Adaptive DHiSparse with DSpark delivers 1.85$\times$ the output of resident serving with DSpark. In a load step, optional residency reduces host fetches from 14.2% to 1.4% of selected rows at low load and releases memory under pressure.
Abstract:Long-context serving runs out of KV memory before the hardware reaches its useful throughput. HiSparse stores history in host memory and fetches only the selected entries, but it uses one cache budget at every load and does not handle draft-model speculative decoding such as DSpark and DFlash2. We present Adaptive DHiSparse, which supports speculative decoding over hierarchical sparse KV and adapts residency to load. It protects the required working set, retains additional history while GPU memory is spare, and recovers the rest through exact fused retrieval. DSpark and DFlash2 keep their proposal and acceptance rules: queries share missing-row retrieval, draft KV uses a window-bounded ring, and verification staging is reused across layer groups. At 100K context and concurrency 128, GLM-5.3 with DFlash2 delivers 793 output tokens/s per decode GPU, 1.45$\times$ HiSparse and 1.27$\times$ non-speculative adaptive serving. Held to the same latency bounds as HiSparse, it sustains up to 4$\times$ the concurrency and 2.3 to 3.7$\times$ the output, depending on the per-token bound. On DeepSeek-V4.1-Flash at 500K context and concurrency 512, Adaptive DHiSparse with DSpark delivers 1.85$\times$ the output of resident serving with DSpark. In a load step, optional residency reduces host fetches from 14.2% to 1.4% of selected rows at low load and releases memory under pressure.
Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG)
arXiv:2610.13708 [cs.AI]
(or arXiv:2610.13708v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2610.13708 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 314 words
Related coverage
Nightjar: Dynamic Adaptive Speculative Decoding for Large Language Models Serving
arXiv · Jun 16, 2026
TreeSpark: Calibrated, Load-Adaptive Draft Trees for Semi-Autoregressive Speculative Decoding
arXiv · Sep 22, 2026
Optimizing ATOM and vLLM-ATOM for High-Interactivity Inference
AMD · Sep 1, 2026
[AINews] High Taste Testers Ep. 1
Latent Space · Oct 17, 2026