SA-COV-SA-MEM-GDDR7-CAPACITY
① SA Source
- Source: 開啟完整 SA 文章
- Section:
SRAM and Memory Hierarchy - Line hint:
59
Context Before
We have written about the role of SRAM in the memory hierarchy, but the quick recap is that SRAM is very fast (low latency and high bandwidth) but this comes at the expense of density and therefore cost.
SRAM machines such as Groq’s LPU therefore enable very fast time to first token and tokens per second per user but at the expense of total throughput, as their limited SRAM capacity quickly gets saturated by weights, with little left over for KVcache that grows as more users are batched. GPUs win for throughput and cost as we have shown. This is why Nvidia has decided to combine these architectures to get the best of both worlds: accelerate parts of decode that are more latency sensitive and are not as memory heavy on a low-latency SRAM-heavy chip like the LPU, while memory hungry attention is performed on GPUs that come with a lot of fast (but not SRAM fast) memory capacity.
Evidence
Context After
Source: SemiAnalysis
This brings us to the Groq 3 LPU or LP30, with LPU gen 2 being skipped over. This chip has no Nvidia design involvement. The SerDes issues affecting v2 appear to be fixed. Behind the paywall, we will reveal the SerDes IP vendor which may come as a surprise. Nvidia also announced an LP35 which is a minor refresh of the LP30 which will remain on SF4 and will require a new tapeout. It will incorporate NVFP4 number format but given Nvidia is prioritizing time to market we don’t expect any other drastic design changes.
② Atomic Claim
In SemiAnalysis’s March 2026 memory comparison, GDDR7 capacity is ~16–48GB per GPU (~8–12 chips).
- Epistemic Mode:
ESTIMATED - Mapping Status:
COMPLETE
③ Semantic Frame
{
"attribute": "MARCH_2026_COMPARISON_CAPACITY",
"context_nodes": [],
"entity": {
"id": "04_knowledge_base/GDDR7",
"label": "GDDR7"
},
"frame_type": "ATTRIBUTE",
"qualifiers": {
"condition_text": null,
"numeric_mentions": [],
"temporal_mentions": []
},
"value": {
"numeric_mentions": [],
"value_text": "~16–48GB per GPU (~8–12 chips)"
}
}④ Canonical Entity Mapping
| Role | Surface Label | Canonical Target |
|---|---|---|
| entity | GDDR7 | GDDR7 |
⑤ V2 Draft（需要修正時直接編輯）
Inline repair
只編輯以下三個 Draft 區塊。完成後將
review_action設為REQUEST_REPAIR，再手動執行 Repair Preview。
執行腳本前，Draft 不會寫入 Claim Store 或 Graph。
V2 Atomic Claim Draft
In SemiAnalysis’s March 2026 memory comparison, GDDR7 capacity is ~16–48GB per GPU (~8–12 chips).
V2 Semantic Frame Draft
{
"attribute": "MARCH_2026_COMPARISON_CAPACITY",
"context_nodes": [],
"entity": {
"id": "04_knowledge_base/GDDR7",
"label": "GDDR7"
},
"frame_type": "ATTRIBUTE",
"qualifiers": {
"condition_text": null,
"numeric_mentions": [],
"temporal_mentions": []
},
"value": {
"numeric_mentions": [],
"value_text": "~16–48GB per GPU (~8–12 chips)"
}
}V2 Revision Metadata Draft
{
"epistemic_mode": "ESTIMATED",
"mapping_status": "COMPLETE"
}⑥ 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。