clm_847dc5204c844ec081ed13355157dedd_r1

NIEK2-0327

① SA Source

• Source: 開啟完整 SA 文章
• Section: Feynman NVL1152 Networking Topologies
• Line hint: 399

Context Before

Feynman NVL1152 Networking Topologies

Within each Feynman Kyber rack, we tentatively assume double the bandwidth per logical GPU and double the NVLink Switch bandwidth to 28.8T and 57.6T respectively. Though Jensen, in the Financial Q+A the day after the GTC Keynote, characterized NVL1152 as “all CPO”, the key technical blog outlining the new rack form factors ↗ only strictly referenced CPO for rack to rack interconnect. We will discuss the potential topography for both options.

Evidence

To double the scale-up bandwidth using copper interconnect, NVIDIA would have to achieve a per lane bandwidth of 448Gbit/s uni-di (and implemented with simulatanous bi-directional SerDes so that each physical channel carries 448G of RX and 448G of TX)

Context After

Source: SemiAnalysis

② Atomic Claim

若要用 copper interconnect 將 scale-up 頻寬加倍，NVIDIA 必須把每 lane 頻寬提升至 448Gbit/s uni-di，且配合 simultaneous bi-directional SerDes，讓每個實體 channel 同時承載 448G RX 與 448G TX。

• Epistemic Mode: HYPOTHETICAL
• Mapping Status: PARTIAL

③ Semantic Frame

{
  "comparison_expression": "若要用 copper interconnect 將 scale-up 頻寬加倍，NVIDIA 必須把每 lane 頻寬提升至 448Gbit/s uni-di，且配合 simultaneous bi-directional SerDes，讓每個實體 channel 同時承載 448G RX 與 448G TX。",
  "entities": [
    {
      "id": "04_knowledge_base/Copper",
      "label": "copper"
    },
    {
      "id": "02_companies/NVDA",
      "label": "NVIDIA"
    },
    {
      "id": "04_knowledge_base/SerDes",
      "label": "SerDes"
    }
  ],
  "frame_type": "COMPARISON",
  "metric": "BANDWIDTH",
  "operator": "MULTIPLE_OF",
  "qualifiers": {
    "condition_text": null,
    "numeric_mentions": [
      "448Gb",
      "448"
    ],
    "temporal_mentions": []
  }
}

④ Canonical Entity Mapping

Role	Surface Label	Canonical Target
comparison_entity_0	copper	Copper
comparison_entity_1	NVIDIA	NVDA
comparison_entity_2	SerDes	SerDes

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。