IX2-0207
① SA Source
- Source: 開啟完整 SA 文章
- Section:
Nvidia Disagg Prefill and WideEP - Line hint:
364
Context Before
EP requires all-to-all communication, where every GPU needs to send tokens to every other GPU. This is extremely bandwidth hungry. Recall that Nvidia’s servers have two separate networking domains – the scale-up NVLink domain, and the Scale-out Domain, usually using InfiniBand or Ethernet as the networking protocol.
Evidence
This is roughly 7-10x the bandwidth of the InfiniBand/Ethernet based scale-out network
Context After
InfiniBand/RoCEv2 Ethernet (outside of the NVL72 rack): Typically 400-800 Gbit/s per GPU uni-directional (50-100 GB/s). Note that all our testing for Nvidia was conducted on InfiniBand based clusters.
② Atomic Claim
這大約是基於 InfiniBand／Ethernet 的 scale-out network 頻寬的 7～10 倍。
- Epistemic Mode:
ASSERTED - Mapping Status:
COMPLETE
③ Semantic Frame
{
"comparison_expression": "這大約是基於 InfiniBand／Ethernet 的 scale-out network 頻寬的 7～10 倍。",
"entities": [
{
"id": "04_knowledge_base/InfiniBand",
"label": "InfiniBand"
},
{
"id": "04_knowledge_base/Ethernet",
"label": "Ethernet"
},
{
"id": "04_knowledge_base/Scale-out networking",
"label": "scale-out network"
}
],
"frame_type": "COMPARISON",
"metric": "BANDWIDTH",
"operator": "MULTIPLE_OF",
"qualifiers": {
"condition_text": null,
"numeric_mentions": [
"7",
"10"
],
"temporal_mentions": []
}
}④ Canonical Entity Mapping
| Role | Surface Label | Canonical Target |
|---|---|---|
| comparison_entity_0 | InfiniBand | InfiniBand |
| comparison_entity_1 | Ethernet | Ethernet |
| comparison_entity_2 | scale-out network | 04_knowledge_base/Scale-out networking |
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。