VR2R-0056_r1

VR2R-0056

① SA Source

• Source: 開啟完整 SA 文章
• Section: Extreme Co-design: 6 Silicon Product Offerings under the Rubin Platform - Chip Floorplans & Specifications
• Line hint: 49

Context Before

Evidence

Rubin’s dense FP4 and FP8 FLOPs increase by roughly ~3.5× versus GB200, while FP16 FLOPs rise by a more modest ~1.6x, underscoring NVIDIA’s continued emphasis on FP4/FP8 as the primary scaling vector.

Context After

On the memory side, HBM capacity remains flat from GB300, while HBM bandwidth scales more aggressively at ~2.8x. Overall, the architecture prioritizes bandwidth and low-precision compute.

② Atomic Claim

Rubin 的 dense FP4 與 FP8 FLOPs 相較 GB200 約提升 3.5 倍，而 FP16 FLOPs 約提升 1.6 倍，顯示 NVIDIA 持續以 FP4／FP8 作為主要 scaling vector。

• Epistemic Mode: ASSERTED
• Mapping Status: COMPLETE

③ Semantic Frame

{
  "comparison_expression": "Rubin dense FP4/FP8 FLOPs 約為 GB200 的 3.5x，FP16 FLOPs 約為 1.6x；NVIDIA 持續偏重 FP4/FP8 scaling。",
  "entities": [
    {
      "id": "04_knowledge_base/Vera Rubin",
      "label": "Rubin"
    },
    {
      "id": "04_knowledge_base/FP4",
      "label": "FP4"
    },
    {
      "id": "04_knowledge_base/FP8",
      "label": "FP8"
    },
    {
      "id": "04_knowledge_base/FP16",
      "label": "FP16"
    },
    {
      "id": "04_knowledge_base/GB200",
      "label": "GB200"
    },
    {
      "id": "02_companies/NVDA",
      "label": "NVIDIA"
    }
  ],
  "frame_type": "COMPARISON",
  "metric": "DENSE_FLOPS_SCALING",
  "operator": "MULTIPLE_OF",
  "qualifiers": {
    "condition_text": null,
    "numeric_mentions": [
      "3.5x",
      "1.6x"
    ],
    "temporal_mentions": []
  }
}

④ Canonical Entity Mapping

Role	Surface Label	Canonical Target
entity	Rubin	04_knowledge_base/Vera Rubin
entity	FP4	FP4
entity	FP8	FP8
entity	FP16	FP16
entity	GB200	GB200
entity	NVIDIA	NVDA

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。