跳到正文
arXiv cs.AI· Peizhe Liu, Akanksha Jain, Gerasimos Gerogiannis, Vidushi Dadu, Parthasarathy Ranganathan, Jovan Stojkovic·· 1 小时前AI 评分43

Lodestone:自动化语义发现数据中心硬件专用化机会的智能体框架

Lodestone: Automated Semantic Discovery of Datacenter Hardware Specialization

AI 导读

arXiv 论文提出 Lodestone,一个通过语义推理自动发现数据中心硬件专用化机会的智能体框架。它从执行轨迹提取热点控制流区域,由智能体合并相同计算形成计算族并验证语义正确性,再对候选族提出 ISA 扩展、近核单元或独立加速器等硬件方案。在 6 类 20 个工作负载上模拟 30 个计算族并在 RTL 实现,平均加速 12.6%,最高达 80.2%。

正文 · 原文

View PDF HTML (experimental)

Abstract:Hardware specialization can improve datacenter efficiency. Unfortunately, identifying computations that justify dedicated hardware is difficult. Existing approaches analyze hot functions, instruction patterns, or overheads such as datacenter tax. They can miss fine-grained computations whose aggregate cost is spread across applications and libraries, especially when different pieces of code obscure the same underlying computation. We observe that seemingly unrelated code regions often perform the same computations, creating specialization opportunities that require semantic reasoning. The rise of agentic AI makes it possible to automate this. However, conventional AI agents cannot efficiently and reliably analyze large scale datacenter software. To unlock this opportunity, we present Lodestone, an agentic framework that automatically discovers hardware specialization opportunities by reasoning about what the executed code computes. First, Lodestone extracts hot control-flow regions from execution traces across workloads and organizes them into compact analysis packages that make large workloads tractable for agents. Second, agents describe what each region computes, merge those that perform the same computation into computational families, and check their semantic correctness in a loop. Third, to verify the discovered family, Lodestone runs every piece of code in a family against a set of reference inputs and outputs on recorded executions. Finally, for selected families, Lodestone proposes hardware, e.g., ISA extensions, near-core units, or detached accelerators, and human experts decide which proposals to develop further. We evaluate Lodestone on 20 workloads across six datacenter workload categories. We simulate 30 discovered families in a cycle-accurate simulator and implement them in RTL. These optimizations speed up the workloads by 12.6% on average and by up to 80.2%.
Subjects: Hardware Architecture (cs.AR); Artificial Intelligence (cs.AI)
Cite as: arXiv:2611.23663 [cs.AR]
  (or arXiv:2611.23663v1 [cs.AR] for this version)
  https://doi.org/10.48550/arXiv.2611.23663

arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Jovan Stojkovic [view email]
[v1] Wed, 18 Nov 2026 00:39:06 UTC (413 KB)

来源:arXiv cs.AI · arxiv.org