clm_d9960300b65245d0bf915242cf2c5934_r1

VR2-0773

① SA Source

• Source: 開啟完整 SA 文章
• Section: VR NVL72 TCO: BoM and Power Budget Analysis
• Line hint: 981

Context Before

Source: SemiAnalysis AI TCO Model ↗

Evidence

5,000 Dense FP8 TFLOPs for MI355

Context After

Given that total cost of ownership is only 1.75x higher, this results in a superior performance-per-TCO profile for B300 despite the more modest marketed figures. Such a wildly different result despite very similar specs underscores that real world performance is not dictated by peak FLOPS or memory bandwidth alone. Software and network capabilities are also major factors that contribute to training and token throughput in real workloads.

Rubin and MI4XX will ship with new microarchitectures, real world performance is especially difficult to predict without and benchmarking like we do with InferenceX.

② Atomic Claim

MI355 額定為 5,000 dense FP8 TFLOPs。

• Epistemic Mode: ASSERTED
• Mapping Status: COMPLETE

③ Semantic Frame

{
  "attribute": "COMPUTE_PERFORMANCE",
  "context_nodes": [
    {
      "id": "04_knowledge_base/FP8",
      "label": "FP8"
    },
    {
      "id": "04_knowledge_base/TFLOPS",
      "label": "TFLOPs"
    }
  ],
  "entity": {
    "id": "04_knowledge_base/MI355",
    "label": "MI355"
  },
  "frame_type": "ATTRIBUTE",
  "qualifiers": {
    "condition_text": null,
    "numeric_mentions": [
      "5,000"
    ],
    "temporal_mentions": []
  },
  "value": {
    "numeric_mentions": [
      "5,000"
    ],
    "value_text": "MI355 額定為 5,000 dense FP8 TFLOPs。"
  }
}

④ Canonical Entity Mapping

Role	Surface Label	Canonical Target
entity	MI355	MI355
context_0	FP8	FP8
context_1	TFLOPs	TFLOPS

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。