XTC, A Research Platform for Optimizing AI Workload Operators

TL;DR

XTC is a unified platform that standardizes scheduling and performance evaluation for AI workloads, enabling fair comparison, reuse, and accelerated research across different compiler ecosystems.

Contribution

It introduces a common API and measurement framework that decouples scheduling from code generation, facilitating portable experimentation and optimization research.

Findings

Enables cross-compiler scheduling comparison

Improves reproducibility of performance measurements

Accelerates AI workload optimization research

Abstract

Achieving high efficiency on AI operators demands precise control over computation and data movement. However, existing scheduling languages are locked into specific compiler ecosystems, preventing fair comparison, reuse, and evaluation across frameworks. No unified interface currently decouples scheduling specification from code generation and measurement. We introduce XTC, a platform that unifies scheduling and performance evaluation across compilers. With its common API and reproducible measurement framework, XTC enables portable experimentation and accelerates research on optimization strategies.