Research

Paper

TESTING March 10, 2026

Fully Symbolic Analysis of Loop Locality: Using Imaginary Reuse to Infer Real Performance

Authors

Yifan Zhu, Yekai Pan, Chen Ding, Yanghui Wu

Abstract

This paper presents a new theory of locality and its compiler support. The theory is fully symbolic and derives locality as polynomials, and the compiler analysis supports affine loop nests. They derive cache-performance scaling in quadratic and reciprocal expressions and are more general and precise than empirical scaling rules. Evaluated on a benchmark suite of 41 scientific kernels and tensor operations, the compiler requires an average of 41 seconds to derive the locality polynomials. After derivation, predicting the cache miss count for any given input size and cache configuration takes less than a millisecond. Across all tests--with and without loop fusion--the accuracy in the data movement prediction is 99.6\%, compared to simulated set-associative L1 data cache.

Metadata

arXiv ID: 2603.10196
Provider: ARXIV
Primary Category: cs.PL
Published: 2026-03-10
Fetched: 2026-03-12 04:21

Related papers

Raw Data (Debug)
{
  "raw_xml": "<entry>\n    <id>http://arxiv.org/abs/2603.10196v1</id>\n    <title>Fully Symbolic Analysis of Loop Locality: Using Imaginary Reuse to Infer Real Performance</title>\n    <updated>2026-03-10T19:51:31Z</updated>\n    <link href='https://arxiv.org/abs/2603.10196v1' rel='alternate' type='text/html'/>\n    <link href='https://arxiv.org/pdf/2603.10196v1' rel='related' title='pdf' type='application/pdf'/>\n    <summary>This paper presents a new theory of locality and its compiler support. The theory is fully symbolic and derives locality as polynomials, and the compiler analysis supports affine loop nests. They derive cache-performance scaling in quadratic and reciprocal expressions and are more general and precise than empirical scaling rules.\n  Evaluated on a benchmark suite of 41 scientific kernels and tensor operations, the compiler requires an average of 41 seconds to derive the locality polynomials. After derivation, predicting the cache miss count for any given input size and cache configuration takes less than a millisecond. Across all tests--with and without loop fusion--the accuracy in the data movement prediction is 99.6\\%, compared to simulated set-associative L1 data cache.</summary>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.PL'/>\n    <published>2026-03-10T19:51:31Z</published>\n    <arxiv:primary_category term='cs.PL'/>\n    <author>\n      <name>Yifan Zhu</name>\n    </author>\n    <author>\n      <name>Yekai Pan</name>\n    </author>\n    <author>\n      <name>Chen Ding</name>\n    </author>\n    <author>\n      <name>Yanghui Wu</name>\n    </author>\n  </entry>"
}