clm_cbdac41e94f748fc9f8f821966b4c178_r1

VR2-0636

① SA Source

• Source: 開啟完整 SA 文章
• Section: Connecting GPUs in the Scale-Out InfiniBand Network
• Line hint: 767

Context Before

Illustratively, consider two hypothetical VR NVL deployments with 1.6T and 800G logical ports respectively. A 1-plane, 3-layer network with one 1.6T logical ports achieves only a maximum cluster size of 93,312 GPUs, or:

2(115,2001,6002)3=2(722)3=93,312

Evidence

By one 1.6T logical port, we mean that the two 800G OSFP cages connected to each GPU are connected to a single, dual-port 1.6T transceiver at the leaf layer because the two 800G ports are effectively performing the function of one 1.6T port – and hence the term “logical”

Context After

Source: SemiAnalysis AI Networking Model ↗

② Atomic Claim

所謂一個 1.6T logical port，是指每顆 GPU 連接的兩個 800G OSFP cages，在 leaf layer 連到同一個 dual-port 1.6T transceiver；兩個 800G ports 實際共同執行一個 1.6T port 的功能，因此稱為 logical port。

• Epistemic Mode: ASSERTED
• Mapping Status: PARTIAL

③ Semantic Frame

{
  "frame_type": "NARY_RELATION",
  "participants": [
    {
      "node": {
        "id": "04_knowledge_base/1.6T",
        "label": "1.6T"
      },
      "role": "partner"
    },
    {
      "node": {
        "id": "04_knowledge_base/GPU",
        "label": "GPU"
      },
      "role": "partner"
    },
    {
      "node": {
        "id": "04_knowledge_base/800G",
        "label": "800G"
      },
      "role": "partner"
    },
    {
      "node": {
        "id": "04_knowledge_base/OSFP (Octal Small Form-factor Pluggable)",
        "label": "OSFP"
      },
      "role": "partner"
    }
  ],
  "qualifiers": {
    "condition_text": null,
    "numeric_mentions": [
      "1.6",
      "800"
    ],
    "temporal_mentions": []
  },
  "relation_type": "PARTNERS_WITH"
}

④ Canonical Entity Mapping

Role	Surface Label	Canonical Target
partner	1.6T	1.6T
partner	GPU	GPU
partner	800G	800G
partner	OSFP	04_knowledge_base/OSFP (Octal Small Form-factor Pluggable)

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。