SA260324R-0294_r1

SA260324R-0294

① SA Source

• Source: 開啟完整 SA 文章
• Section: CMX
• Line hint: 374

Context Before

KV cache grows linearly with input sequence length and number of users and is the primary tradeoff when it comes to prefill performance (time to first token).

Evidence

At scale, on-device HBM does not have enough capacity.

Context After

Host DRAM extends beyond HBM capacity with an additional tier of cache, but also hits limits on total amount per node, memory bandwidth, and network bandwidth. Enter NVMe storage for additional KVcache offload.

② Atomic Claim

在大規模情境下，device 上的 HBM 容量不足。

• Epistemic Mode: ASSERTED
• Mapping Status: COMPLETE

③ Semantic Frame

{
  "attribute": "CAPACITY_SUFFICIENCY",
  "context_nodes": [],
  "entity": {
    "id": "04_knowledge_base/HBM",
    "label": "HBM"
  },
  "frame_type": "ATTRIBUTE",
  "qualifiers": {
    "condition_text": "at scale",
    "numeric_mentions": [],
    "temporal_mentions": []
  },
  "value": {
    "numeric_mentions": [],
    "value_text": "在大規模情境下容量不足。"
  }
}

④ Canonical Entity Mapping

Role	Surface Label	Canonical Target
entity	HBM	HBM

⑤ Human Review

請在 Properties 逐項確認：

• 原文 → Atomic Claim 是否忠實
• Atomic Claim → Semantic Frame 是否忠實
• Canonical Entity mapping 是否正確
• Epistemic mode 是否保留原文語氣
• 最後選擇 review_action

Review state

Markdown 內文不是正式 approval。只有 Apply bridge 寫入的 Decision Ledger event 才是正式決策。