IX2-0171
① SA Source
- Source: 開啟完整 SA 文章
- Section:
Nvidia versus AMD Disagg Prefill - Line hint:
288
Context Before
Source: SemiAnalysis InferenceX ↗
Evidence
Context After
Source: SemiAnalysis InferenceX ↗
② Atomic Claim
比較 Dynamo TRTLLM B200 disagg prefill 與 SGLang MoRI MI355 disagg prefill，AMD 明顯落後，主要因為 TRTLLM 的 disagg prefill implementation 更成熟。
- Epistemic Mode:
ASSERTED - Mapping Status:
COMPLETE
③ Semantic Frame
{
"attribute": "MATURITY",
"context_nodes": [
{
"id": "04_knowledge_base/NVIDIA B200",
"label": "B200"
},
{
"id": "04_knowledge_base/Prefill",
"label": "prefill"
},
{
"id": "04_knowledge_base/SGLang",
"label": "SGLang"
},
{
"id": "04_knowledge_base/AMD MoRI backend",
"label": "MoRI"
},
{
"id": "04_knowledge_base/MI355",
"label": "MI355"
},
{
"id": "02_companies/AMD",
"label": "AMD"
}
],
"entity": {
"id": "04_knowledge_base/TensorRT-LLM",
"label": "TRTLLM"
},
"frame_type": "ATTRIBUTE",
"qualifiers": {
"condition_text": null,
"numeric_mentions": [],
"temporal_mentions": []
},
"value": {
"numeric_mentions": [],
"value_text": "比較 Dynamo TRTLLM B200 disagg prefill 與 SGLang MoRI MI355 disagg prefill，AMD 明顯落後，主要因為 TRTLLM 的 disagg prefill implementation 更成熟。"
}
}④ Canonical Entity Mapping
| Role | Surface Label | Canonical Target |
|---|---|---|
| entity | TRTLLM | TensorRT-LLM |
| context_0 | B200 | 04_knowledge_base/NVIDIA B200 |
| context_1 | prefill | Prefill |
| context_2 | SGLang | SGLang |
| context_3 | MoRI | 04_knowledge_base/AMD MoRI backend |
| context_4 | MI355 | MI355 |
| context_5 | AMD | AMD |
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。