IX2-0500
① SA Source
- Source: 開啟完整 SA 文章
- Section:
Optimizing Inference with Wide EP + Disaggregated Serving - Line hint:
698
Context Before
Source: SemiAnalysis InferenceX ↗
We observe the same general pattern when extending to wide EP with disaggregated prefill. Prefill and decode run with separate parallelism strategies and node counts, both tuned to the workload and target interactivity level. Take an 8k/1k workload (prefill heavy) at the high-throughput, low-interactivity end of the frontier. Prefill is the bottleneck as each request requires a forward pass of 8192 input tokens, which is computationally expensive. Recipes in this region allocate more prefill nodes than decode (4P1D, 7P2D, 4P3D) to sustain high prefill throughput. These prefill nodes run DEP configurations, replicating attention weights across independent data-parallel ranks so that multiple long-context prefills can be processed simultaneously. Decode nodes are fewer but run wide DEP with large batch sizes by the same principle as with single node.
Evidence
expert parallelism handles MoE routing at the moderate batch sizes where EP load balance is sufficient
Context After
② Atomic Claim
expert parallelism 則在中等 batch size、EP load balance 已足夠時負責 MoE routing。
- Epistemic Mode:
ASSERTED - Mapping Status:
COMPLETE
③ Semantic Frame
{
"additional_nodes": [
{
"id": "04_knowledge_base/Batch size",
"label": "batch size"
}
],
"frame_type": "RELATION",
"object": {
"id": "04_knowledge_base/MoE Routing",
"label": "MoE routing"
},
"predicate": "RESPONSIBLE_FOR",
"qualifiers": {
"condition_text": null,
"numeric_mentions": [],
"temporal_mentions": []
},
"subject": {
"id": "04_knowledge_base/Expert Parallelism",
"label": "expert parallelism"
}
}④ Canonical Entity Mapping
| Role | Surface Label | Canonical Target |
|---|---|---|
| subject | expert parallelism | 04_knowledge_base/Expert Parallelism |
| object | MoE routing | 04_knowledge_base/MoE Routing |
| additional_0 | batch size | 04_knowledge_base/Batch size |
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。