clm_2c8395fa0c5f4ddea6992d6f3382c8fb_r1

VR2-0768

① SA Source

• Source: 開啟完整 SA 文章
• Section: VR NVL72 TCO: BoM and Power Budget Analysis
• Line hint: 971

Context Before

NVIDIA is offering sparsity for FP4, marketing 50 PFLOPS of FP4, while AMD has opsted to remove sparsity support since CDNA4 for inference dtypes. Harnessing the 50 PFLOPS sparse vs 35 PFLOPS dense drops the cost per perf in units of $/hr per Marketed PFLOP by 35% – a valid comparison if AI Labs can indeed successfully harness Sparse FP4 on the VR NVL72.

As always, one caveat is that this comparison is done based on a marketed dense PFLOP basis. Effective dense PFLOP (i.e. the real world chip throughput) can differ based on Model Flops Utilization % (MFU), and in general we have seen NVIDIA chip operate at a higher MFU % vs AMD chips, suggesting that performance per TCO based on effective dense PFLOPs could be better for NVIDIA systems vs AMD – however, MFU is dependent on actual workloads with no one-size-fits-all MFU % that is consistently applicable to either systems.

Evidence

Indeed, real world use of FP4 Sparsity will probably not reach 50 PFLOPS but it will probably deliver better effective FLOPs than FP4 Dense

Context After

Source: SemiAnalysis AI TCO Model ↗

② Atomic Claim

實務上，FP4 Sparsity 很可能無法真正達到 50 PFLOPS，但有效 FLOPs 可能仍高於 FP4 Dense。

• Epistemic Mode: ASSERTED
• Mapping Status: PARTIAL

③ Semantic Frame

{
  "comparison_expression": "實務上，FP4 Sparsity 很可能無法真正達到 50 PFLOPS，但有效 FLOPs 可能仍高於 FP4 Dense。",
  "entities": [
    {
      "id": "04_knowledge_base/FP4",
      "label": "FP4"
    },
    {
      "id": "04_knowledge_base/Sparsity",
      "label": "Sparsity"
    }
  ],
  "frame_type": "COMPARISON",
  "metric": "COMPUTE_PERFORMANCE",
  "operator": "GREATER_THAN",
  "qualifiers": {
    "condition_text": null,
    "numeric_mentions": [
      "50 PFLOPS"
    ],
    "temporal_mentions": []
  }
}

④ Canonical Entity Mapping

Role	Surface Label	Canonical Target
comparison_entity_0	FP4	FP4
comparison_entity_1	Sparsity	Sparsity

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。