clm_b884c332134f4a27822459b2e03cd3ad_r1

IX2-0513

① SA Source

• Source: 開啟完整 SA 文章
• Section: Dive into DeepSeek R1 Single Node Results
• Line hint: 732

Context Before

Source: SemiAnalysis InferenceX ↗

In comparing DeepSeek R1 FP8 perf between H200 (SGLang) and MI325X (SGLang), not much has changed since our initial release of InferenceXv1 last October. The MI325X data was captured on Feb 12th, 2026 with SGLang 0.5.8 whereas the B200 data was captured Jan 23, 2026 with SGLang 0.5.7.

Evidence

One thing we note is the considerably smaller interactivity range for MI325X than H200, with H200 ranging from 30-90 tok/sec/user whereas MI325X ranges from only 13-35 tok/sec/user

Context After

Source: SemiAnalysis InferenceX ↗

② Atomic Claim

MI325X 的 interactivity 範圍明顯小於 H200：H200 約為 30–90 tok/sec/user，而 MI325X 僅約 13–35 tok/sec/user。

• Epistemic Mode: ASSERTED
• Mapping Status: COMPLETE

③ Semantic Frame

{
  "attribute": "THROUGHPUT",
  "context_nodes": [
    {
      "id": "04_knowledge_base/H200",
      "label": "H200"
    }
  ],
  "entity": {
    "id": "04_knowledge_base/MI325X",
    "label": "MI325X"
  },
  "frame_type": "ATTRIBUTE",
  "qualifiers": {
    "condition_text": null,
    "numeric_mentions": [
      "30–90 tok/s",
      "13–35 tok/s"
    ],
    "temporal_mentions": []
  },
  "value": {
    "numeric_mentions": [
      "30–90 tok/s",
      "13–35 tok/s"
    ],
    "value_text": "MI325X 的 interactivity 範圍明顯小於 H200：H200 約為 30–90 tok/sec/user，而 MI325X 僅約 13–35 tok/sec/user。"
  }
}

④ Canonical Entity Mapping

Role	Surface Label	Canonical Target
entity	MI325X	MI325X
context_0	H200	H200

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。