VR2R-0088_r1

VR2R-0088

① SA Source

• Source: 開啟完整 SA 文章
• Section: Rubin
• Line hint: 75

Context Before

Blackwell Ultra GB300 added 50% more dense FP4 while keeping sparse FP4 FLOPs the same, while AMD’s MI355X stopped supporting structured sparsity on MXFP8, MXFP6 and MXFP4 formats to save silicon area.

Evidence

Rubin’s adaptive compression engine in the improved Transformer Engine is a key feature to re-boost naturally sparser inference performance by doing dynamic computation of sparsity in-flight and eliminating zeros in the data stream without zeroing out non-zero values, thus maintaining model accuracy while still boosting performance.

Context After

This is done automatically on existing models built for Blackwell without the need for a new programming model or specific optimizations.

② Atomic Claim

Rubin 改良版 Transformer Engine 中的 adaptive compression engine，會在資料流中即時動態計算 sparsity 並消除 zeros，但不把非零值強制歸零，因此可維持 model accuracy，同時提升天然較 sparse 的 inference performance。

• Epistemic Mode: INFERRED
• Mapping Status: COMPLETE

③ Semantic Frame

{
  "frame_type": "NARY_RELATION",
  "participants": [
    {
      "node": {
        "id": "04_knowledge_base/Vera Rubin",
        "label": "Rubin"
      },
      "role": "platform"
    },
    {
      "node": {
        "id": "04_knowledge_base/Transformer Engine",
        "label": "Transformer Engine"
      },
      "role": "engine"
    },
    {
      "node": {
        "id": "04_knowledge_base/Sparsity",
        "label": "sparsity"
      },
      "role": "sparsity_mechanism"
    }
  ],
  "qualifiers": {
    "condition_text": "dynamic in-flight zero elimination without zeroing non-zero values",
    "numeric_mentions": [],
    "temporal_mentions": []
  },
  "relation_type": "ADAPTIVE_COMPRESSION_SPARSE_INFERENCE_MECHANISM"
}

④ Canonical Entity Mapping

Role	Surface Label	Canonical Target
platform	Rubin	04_knowledge_base/Vera Rubin
engine	Transformer Engine	04_knowledge_base/Transformer Engine
sparsity_mechanism	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。