clm_670bbe900445425380fdfc31382bf230_r3

SA260324R-0110

回到這篇 SA Coverage

① SA Source

• Source: 開啟完整 SA 文章
• Section: GPU and LPU Integration: Attention FFN Disaggregation (AFD)
• Line hint: 101

Context Before

Source: MegaScale-Infer, SemiAnalysis

Evidence

As state-of-the-art mixture-of-expert (MoE) models grow increasingly sparse, tokens can choose experts from a larger expert pool. As a result, each expert receives fewer tokens, leading to lower utilization. This motivates attention and FFN disaggregation. If a GPU only performs attention operations, its HBM capacity can be fully allocated to KV cache, increasing the total number of tokens it can process, which then increases the tokens each expert processes on average.

Context After

Source: SemiAnalysis

② Atomic Claim

如果 GPU 只執行 attention operation，其 HBM 容量就能全部分配給 KV cache，提高可處理的 token 總數，進而增加每個 expert 平均處理的 token 數。

• Epistemic Mode: HYPOTHETICAL
• Mapping Status: PARTIAL

③ Semantic Frame

{
  "frame_type": "NARY_RELATION",
  "participants": [
    {
      "node": {
        "id": "04_knowledge_base/GPU",
        "label": "GPU"
      },
      "role": "compute_device"
    },
    {
      "node": {
        "id": "04_knowledge_base/HBM",
        "label": "HBM"
      },
      "role": "memory"
    },
    {
      "node": {
        "id": "04_knowledge_base/KV cache",
        "label": "KV cache"
      },
      "role": "allocation_target"
    }
  ],
  "qualifiers": {
    "condition_text": "如果 GPU 只執行 attention operation；attention 與 expert 在此 claim 沒有 canonical entity",
    "numeric_mentions": [],
    "temporal_mentions": []
  },
  "relation_type": "ALLOCATES_MEMORY_TO_KV_CACHE_UNDER_ATTENTION_ONLY"
}

④ Canonical Entity Mapping

Role	Surface Label	Canonical Target
compute_device	GPU	GPU
memory	HBM	HBM
allocation_target	KV cache	04_knowledge_base/KV cache

⑤ V2 Draft（需要修正時直接編輯）

Inline repair

只編輯以下三個 Draft 區塊。完成後將 review_action 設為 REQUEST_REPAIR，再手動執行 Repair Preview。
執行腳本前，Draft 不會寫入 Claim Store 或 Graph。

V2 Atomic Claim Draft

如果 GPU 只執行 attention operation，其 HBM 容量就能全部分配給 KV cache，提高可處理的 token 總數，進而增加每個 expert 平均處理的 token 數。

V2 Semantic Frame Draft

{
  "frame_type": "NARY_RELATION",
  "participants": [
    {
      "node": {
        "id": "04_knowledge_base/GPU",
        "label": "GPU"
      },
      "role": "compute_device"
    },
    {
      "node": {
        "id": "04_knowledge_base/HBM",
        "label": "HBM"
      },
      "role": "memory"
    },
    {
      "node": {
        "id": "04_knowledge_base/KV cache",
        "label": "KV cache"
      },
      "role": "allocation_target"
    }
  ],
  "qualifiers": {
    "condition_text": "如果 GPU 只執行 attention operation；attention 與 expert 在此 claim 沒有 canonical entity",
    "numeric_mentions": [],
    "temporal_mentions": []
  },
  "relation_type": "ALLOCATES_MEMORY_TO_KV_CACHE_UNDER_ATTENTION_ONLY"
}

V2 Revision Metadata Draft

{
  "epistemic_mode": "HYPOTHETICAL",
  "mapping_status": "PARTIAL"
}

⑥ 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。