{
  "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": "2609.39131",
  "title": "Characterizing High Bandwidth Flash for LLM Serving",
  "summary": "The method replaces HBM-only KV cache and weight storage with a three-tier HBM-HBF-host hierarchy plus scheduling that batches and places KV state to exploit high-bandwidth flash. It costs access to HBF hardware that is not commercially available on a 24 GB GPU setup, added data-placement and scheduling complexity, and in some light workloads higher energy consumption. Results come from trace-driven simulation rather than deployed systems, so realized gains depend on workload KV-reuse patterns, and unmanaged flash writes can exhaust device endurance.",
  "rationale": "The gains depend on HBF hardware and validated only in simulation, so the reader should track the technology and its scheduling ideas but cannot adopt it on current constrained infrastructure.",
  "technique": "HBM-HBF-host tiered KV cache with buffered cache-aware scheduling",
  "family": "cache_kv",
  "recommendation": "observar",
  "published": "2026-09-30",
  "as_of": "2026-10-18",
  "editorial_status": "indexed",
  "source_url": "https://arxiv.org/abs/2609.39131",
  "independent_implementations": 0,
  "total_implementations": 0,
  "stars": 0,
  "citations": 0,
  "claims": [
    {
      "claim_id": "claim-01",
      "statement": "The method replaces HBM-only KV cache and weight storage with a three-tier HBM-HBF-host hierarchy plus scheduling that batches and places KV state to exploit high-bandwidth flash. It costs access to HBF hardware that is not commercially available on a 24 GB GPU setup, added data-placement and scheduling complexity, and in some light workloads higher energy consumption. Results come from trace-driven simulation rather than deployed systems, so realized gains depend on workload KV-reuse patterns, and unmanaged flash writes can exhaust device endurance.",
      "basis": "inferred",
      "result": "",
      "baseline": "",
      "conditions": "",
      "source_url": "https://arxiv.org/abs/2609.39131",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "claim_id": "claim-02",
      "statement": "Trace-driven simulation reports 36.1-87.0% lower completion time versus HBM-only systems (up to roughly 7.7x), modeled energy savings up to 55.8%, and HBF write lifetime extended from 4.77 to 14.82 years via buffered cache-aware scheduling.",
      "basis": "inferred",
      "result": "",
      "baseline": "",
      "conditions": "",
      "source_url": "https://arxiv.org/abs/2609.39131",
      "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": ""
}