clm_adc8f388b0ec4527824b3f552ba83a3f_r1

IX2-0206

① SA Source

• Source: 開啟完整 SA 文章
• Section: Nvidia Disagg Prefill and WideEP
• Line hint: 364

Context Before

EP requires all-to-all communication, where every GPU needs to send tokens to every other GPU. This is extremely bandwidth hungry. Recall that Nvidia’s servers have two separate networking domains – the scale-up NVLink domain, and the Scale-out Domain, usually using InfiniBand or Ethernet as the networking protocol.

Evidence

NVLink domain (within the NVL72 rack): 72 GPUs connected via NVLink with 900 GB/s uni-directional bandwidth per GPU

Context After

InfiniBand/RoCEv2 Ethernet (outside of the NVL72 rack): Typically 400-800 Gbit/s per GPU uni-directional (50-100 GB/s). Note that all our testing for Nvidia was conducted on InfiniBand based clusters.

② Atomic Claim

在 NVL72 機櫃內的 NVLink domain 中，72 顆 GPUs 透過 NVLink 互連，每顆 GPU 提供 900 GB/s uni-directional bandwidth。

• Epistemic Mode: ASSERTED
• Mapping Status: COMPLETE

③ Semantic Frame

{
  "additional_nodes": [],
  "frame_type": "RELATION",
  "object": {
    "id": "04_knowledge_base/GPU",
    "label": "GPUs"
  },
  "predicate": "PROVIDES",
  "qualifiers": {
    "condition_text": null,
    "numeric_mentions": [
      "72",
      "900 GB/s"
    ],
    "temporal_mentions": []
  },
  "subject": {
    "id": "04_knowledge_base/NVLink",
    "label": "NVLink"
  }
}

④ Canonical Entity Mapping

Role	Surface Label	Canonical Target
subject	NVLink	NVLink
object	GPUs	GPU

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。