GB2-0028

① SA Source

Context Before

image

Source: Nvidia

Evidence

In NVL72, they keep the same flat 1 tier NVLink topology as HGX Hopper/Blackwell such that you can talk to any of the GPUs within the same rack with only 1 hop through an NVSwitch.

Context After

image

Source: SemiAnalysis

② Atomic Claim

NVL72 採 flat 1-tier NVLink topology,同 rack 任一 GPU 之間僅需經過 1 個 NVSwitch hop。

  • Epistemic Mode: ASSERTED
  • Mapping Status: COMPLETE

③ Semantic Frame

{
  "attribute": "nvlink_topology",
  "context_nodes": [
    {
      "id": "04_knowledge_base/NVLink",
      "label": "NVLink"
    },
    {
      "id": "04_knowledge_base/NVSwitch",
      "label": "NVSwitch"
    }
  ],
  "entity": {
    "id": "04_knowledge_base/GB200 NVL72",
    "label": "GB200 NVL72"
  },
  "frame_type": "ATTRIBUTE",
  "qualifiers": {
    "condition_text": null,
    "numeric_mentions": [
      "1"
    ],
    "temporal_mentions": []
  },
  "value": {
    "numeric_mentions": [
      "1"
    ],
    "value_text": "flat 1-tier topology; any same-rack GPU reachable with 1 NVSwitch hop"
  }
}

④ Canonical Entity Mapping

RoleSurface LabelCanonical Target
entityGB200 NVL7204_knowledge_base/GB200 NVL72
context_0NVLinkNVLink
context_1NVSwitchNVSwitch

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。