TileCas lets AI coding agents write faster GPU kernels across levels
TileCas: Cascading Kernel Development Across Programming Abstractions
Software Engineering
Summary
Writing computer programs for graphics and AI processors is tricky because programmers use different levels of detail, from simple to complex. The authors designed TileCas, a system that helps AI coding agents move smoothly between these different levels to improve programs. TileCas tracks how well a program is doing and suggests fixes at whichever level is best. When tested, AI agents using TileCas made programs that ran noticeably faster than before. This shows smarter AI tools can write better computer code when they work across multiple programming styles.
What this means in practice
- •For gpu software engineers: Use TileCas to enable AI assistants to refine GPU kernel code across different abstraction levels for faster hardware performance.
- •For ai accelerator developers: Integrate TileCas with AI coding workflows to improve kernel optimization and speed on specialized AI hardware devices.
Authors
Xinyu Xiao, Zihe Song, Dezhi Ran, Wei Yang, Tao Xie
Abstract
Programming abstractions have traditionally been designed for humans, but code generation is increasingly performed by models and autonomous agents. We investigate abstractions for machine-driven programming through GPU and AI-accelerator kernel generation. High-level domain-specific languages (DSLs) simplify search by delegating implementation details to compilers, while lower-level representations provide finer-grained control at the cost of a larger and more complex search space. In this paper, we present TileCas, a unified, pluggable interface that enables coding agents to develop kernels across multiple editable abstraction levels. Rather than committing to a single representation, TileCas allows an agent to transfer and refine candidates across levels through Cross-Level Diagnosis, which maps correctness and performance feedback to actionable controls at different abstraction levels; Abstraction Scheduling, which selects both the candidate and representation from which subsequent refinement should proceed; and a Heterogeneous Search Tree, which preserves and revisits alternative search trajectories spanning multiple representations. We integrate TileCas with two code agents AKO and KDA, and evaluate both integrations on 50 end-to-end KernelBench Level 3 workloads across two accelerator platforms and two language models. Giving equal weight to the four accelerator-model configurations, AKO with TileCas achieves an arithmetic-mean speedup of $4.84\times$, compared with $3.75\times$ for the strongest evaluated baseline without TileCas, an improvement of 29.0%. Integrating the same interface with KDA improves its mean speedup by 69.2%, with gains in every configuration. These results show that allowing coding agents to navigate across programming abstractions can improve machine-driven software development and can be reused across different base workflows.