IX2-0204

① SA Source

Context Before

Source: SemiAnalysis InferenceX

Nvidia Disagg Prefill and WideEP

Evidence

Recall that Nvidia’s servers have two separate networking domains – the scale-up NVLink domain

Context After

NVLink domain (within the NVL72 rack): 72 GPUs connected via NVLink with 900 GB/s uni-directional bandwidth per GPU. This is roughly 7-10x the bandwidth of the InfiniBand/Ethernet based scale-out network.

② Atomic Claim

需要注意,Nvidia 的伺服器有兩個分離的 networking domains,其中之一是 scale-up NVLink domain。

  • Epistemic Mode: ASSERTED
  • Mapping Status: COMPLETE

③ Semantic Frame

{
  "attribute": "COUNT",
  "context_nodes": [
    {
      "id": "04_knowledge_base/NVLink",
      "label": "NVLink"
    }
  ],
  "entity": {
    "id": "02_companies/NVDA",
    "label": "Nvidia"
  },
  "frame_type": "ATTRIBUTE",
  "qualifiers": {
    "condition_text": null,
    "numeric_mentions": [],
    "temporal_mentions": []
  },
  "value": {
    "numeric_mentions": [],
    "value_text": "需要注意,Nvidia 的伺服器有兩個分離的 networking domains,其中之一是 scale-up NVLink domain。"
  }
}

④ Canonical Entity Mapping

RoleSurface LabelCanonical Target
entityNvidiaNVDA
context_0NVLinkNVLink

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。