SA260324R-0101
① SA Source
- Source: 開啟完整 SA 文章
- Section:
GPU and LPU Integration: Attention FFN Disaggregation (AFD) - Line hint:
111
Context Before
As we explained in our InferenceX article, LLM inference involves two phases: prefill and decode. Prefill processes the full input context: It is compute-intensive, which is suitable for GPUs. On the other hand, decode predicts new tokens and is memory-bounded. Decode is latency-sensitive because the model predicts new tokens one by one, and LPU’s high SRAM bandwidth and low-latency capabilities can help accelerate this iterative process.
Evidence
Attention and FFN are subsets of operations in a model.
Context After
In a model forward pass, attention’s output feeds into a token router, and the token router assigns each token to k experts, where each expert is an FFN. Attention and FFN have very different performance properties. During decode phase, the GPU utilization of attention barely improves when scaling batch size due to being bounded by loading KV cache. In contrast, the GPU utilization of FFN scales with batch size comparatively better.
② Atomic Claim
Attention 與 FFN 都是模型中的 operation subset。
- Epistemic Mode:
ASSERTED - Mapping Status:
PARTIAL
③ Semantic Frame
{
"frame_type": "CLAIM_ONLY",
"qualifiers": {
"condition_text": null,
"numeric_mentions": [],
"temporal_mentions": []
},
"reason": "ATTENTION_HAS_NO_CANONICAL_ENTITY_IN_THIS_CLAIM",
"statement": "Attention 與 FFN 都是模型中的 operation subset。"
}④ Canonical Entity Mapping
| Role | Surface Label | Canonical Target |
|---|---|---|
| — | — | — |
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。