{ "schema_version": 1, "builder_plan": { "scenario": "Long customer conversations exhaust memory or limit concurrent requests.", "change": "Change only the attention cache policy, after checking model and runtime support.", "baseline": "The same model with its ordinary attention cache.", "steps": [ "Locate the paper's implementation and check its license, model support, data needs, and hardware requirements. Stop here if you cannot obtain the required artifacts.", "Save a small set of representative product tasks and expected outcomes. Keep evaluation tasks out of any training or tuning, and choose quality and cost limits before running.", "Run the baseline and the proposed change on identical inputs. Keep other settings fixed and repeat paired runs to expose variation.", "Save per-task outputs, failures, timings, and all usage costs. Review regressions before expanding the test." ], "measure": "Peak cache memory, long-context answer quality, first-token latency, and requests completed.", "decision": "Keep the change only if it meets your preselected quality and latency limits and improves the metric you care about after setup and operating costs. Otherwise retain the baseline. This is an exploration plan, not a validated recommendation." }, "builder_review": null, "arxiv_id": "2610.29820", "title": "UlaCache: Unified Logical Addressing for Sparse LLM Serving", "summary": "UlaCache replaces conventional physically-tied KV cache management with a unified logical address space spanning GPU and optional CPU memory, combined with a Parallel Bucketed LRU for shared GPU pool eviction, so sparse-attention KV selection is decoupled from physical placement. The cost is substantial systems complexity: address-mapping infrastructure, CPU-GPU transfer coordination, local buffer management across decoding steps, and integration with sparse attention and speculative decoding, with gains demonstrated only on H200 and B300 hardware in DP/TP configurations. It can fail when the workload does not already use sparse attention and speculative decoding, when transfer bandwidth between CPU and GPU becomes the bottleneck, or when the reader's single 24 GB GPU cannot realize the distributed-storage benefits the system is designed around.", "rationale": "The reported gains depend on sparse attention, speculative decoding, DP/TP serving, and datacenter GPUs, none of which match a single 24 GB GPU or API-based production setup, so this is worth monitoring rather than adopting.", "technique": "Unified logical addressing for sparse KV cache management", "family": "cache_kv", "recommendation": "observar", "published": "2026-10-21", "as_of": "2026-10-22", "editorial_status": "indexed", "source_url": "https://arxiv.org/abs/2610.29820", "independent_implementations": 0, "total_implementations": 0, "stars": 0, "citations": null, "claims": [ { "claim_id": "claim-01", "statement": "UlaCache replaces conventional physically-tied KV cache management with a unified logical address space spanning GPU and optional CPU memory, combined with a Parallel Bucketed LRU for shared GPU pool eviction, so sparse-attention KV selection is decoupled from physical placement. The cost is substantial systems complexity: address-mapping infrastructure, CPU-GPU transfer coordination, local buffer management across decoding steps, and integration with sparse attention and speculative decoding, with gains demonstrated only on H200 and B300 hardware in DP/TP configurations. It can fail when the workload does not already use sparse attention and speculative decoding, when transfer bandwidth between CPU and GPU becomes the bottleneck, or when the reader's single 24 GB GPU cannot realize the distributed-storage benefits the system is designed around.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2610.29820", "source_page": null, "source_excerpt": "" }, { "claim_id": "claim-02", "statement": "Peak decoding throughput speedups of 1.16x (DP) and 1.70x (TP) on H200, and 1.73x (DP) and 2.51x (TP) on B300, with speculative decoding enabled under matched CPU and GPU memory budgets, versus a state-of-the-art baseline.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2610.29820", "source_page": null, "source_excerpt": "" } ], "exposure_map": [ { "dimension": "quality", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "compute", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "latency", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "operations", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "compatibility", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "security", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "data_and_training", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "reproducibility", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" } ], "risks": [], "minimum_test": [], "open_questions": [], "independent_tests": [], "lab_reproductions": [], "report_available": false, "validation_tier": "unknown", "evidence_tier": "unknown", "infrastructure_basis": "unknown", "equations": [], "equations_status": "not_fetched", "core_kind": "", "equations_fetched_at": "", "equations_source_url": "", "equations_source_sha256": "", "exposures_reviewed_at": "", "exposures_source_sha256": "" }