![[2602.12422] CacheMind: From Miss Rates to Why -- Natural-Language, Trace-Grounded Reasoning for Cache Replacement](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.12422] CacheMind: From Miss Rates to Why -- Natural-Language, Trace-Grounded Reasoning for Cache Replacement
Summary
CacheMind introduces a novel tool for cache replacement, leveraging natural language processing and trace-grounded reasoning to enhance CPU microarchitecture performance.
Why It Matters
This research addresses the limitations of traditional cache replacement methods by providing a conversational interface that allows architects to ask complex questions about cache performance. By improving cache hit rates and offering actionable insights, CacheMind has the potential to significantly enhance CPU efficiency and performance in computing systems.
Key Takeaways
- CacheMind enables natural language queries for cache analysis, improving accessibility for architects.
- The tool achieves significant accuracy in trace-grounded reasoning tasks, outperforming existing methods.
- Actionable insights from CacheMind can lead to measurable improvements in cache performance.
- CacheMindBench serves as a benchmark suite for evaluating LLM-based reasoning in cache replacement.
- The integration of Retrieval-Augmented Generation enhances the effectiveness of cache data analysis.
Computer Science > Hardware Architecture arXiv:2602.12422 (cs) [Submitted on 12 Feb 2026] Title:CacheMind: From Miss Rates to Why -- Natural-Language, Trace-Grounded Reasoning for Cache Replacement Authors:Kaushal Mhapsekar, Azam Ghanbari, Bita Aslrousta, Samira Mirbagher-Ajorpaz View a PDF of the paper titled CacheMind: From Miss Rates to Why -- Natural-Language, Trace-Grounded Reasoning for Cache Replacement, by Kaushal Mhapsekar and 3 other authors View PDF HTML (experimental) Abstract:Cache replacement remains a challenging problem in CPU microarchitecture, often addressed using hand-crafted heuristics, limiting cache performance. Cache data analysis requires parsing millions of trace entries with manual filtering, making the process slow and non-interactive. To address this, we introduce CacheMind, a conversational tool that uses Retrieval-Augmented Generation (RAG) and Large Language Models (LLMs) to enable semantic reasoning over cache traces. Architects can now ask natural language questions like, "Why is the memory access associated with PC X causing more evictions?", and receive trace-grounded, human-readable answers linked to program semantics for the first time. To evaluate CacheMind, we present CacheMindBench, the first verified benchmark suite for LLM-based reasoning for the cache replacement problem. Using the SIEVE retriever, CacheMind achieves 66.67% on 75 unseen trace-grounded questions and 84.80% on 25 unseen policy-specific reasoning tasks; with RANGER,...
