VR2R-0733_r1

VR2R-0733

① SA Source

• Source: 開啟完整 SA 文章
• Section: VR NVL72 TCO: BoM and Power Budget Analysis
• Line hint: 649

Context Before

Inference throughput too can diverge materially from marketed peak FLOPS, meaning spec-sheet compute does not directly translate into real-world token generation performance. While the B300 is rated at 4,500 Dense FP8 TFLOPs versus 5,000 Dense FP8 TFLOPs for MI355, implying a 10% theoretical compute disadvantage and with same marketed memory bandwidth of 8TB/s measured inference token throughput from our InferenceX benchmarks shows B300 delivering roughly 6.3x the performance at 100 interactivity for Deepseek R1, using 8k input tokens and 1k output tokens.

Evidence

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.

Context After

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.

② Atomic Claim

由於 total cost of ownership 只高約 1.75 倍，B300 即使 marketed figures 較保守，仍呈現更優的 performance-per-TCO profile。

• Epistemic Mode: ASSERTED
• Mapping Status: COMPLETE

③ Semantic Frame

{
  "attribute": "PERFORMANCE_PER_TCO_PROFILE",
  "context_nodes": [],
  "entity": {
    "id": "04_knowledge_base/NVIDIA B300",
    "label": "B300"
  },
  "frame_type": "ATTRIBUTE",
  "qualifiers": {
    "condition_text": "based on the preceding InferenceX performance comparison",
    "numeric_mentions": [
      "1.75x"
    ],
    "temporal_mentions": []
  },
  "value": {
    "numeric_mentions": [
      "1.75x"
    ],
    "value_text": "B300 has a superior performance-per-TCO profile even though total cost of ownership is about 1.75x higher."
  }
}

④ Canonical Entity Mapping

Role	Surface Label	Canonical Target
entity	B300	04_knowledge_base/NVIDIA B300

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。