English

Axe: A Simple Unified Layout Abstraction for Machine Learning Compilers

Distributed, Parallel, and Cluster Computing 2026-01-30 v2 Artificial Intelligence Machine Learning Programming Languages

Abstract

Scaling modern deep learning workloads demands coordinated placement of data and compute across device meshes, memory hierarchies, and heterogeneous accelerators. We present Axe Layout, a hardware-aware abstraction that maps logical tensor coordinates to a multi-axis physical space via named axes. Axe unifies tiling, sharding, replication, and offsets across inter-device distribution and on-device layouts, enabling collective primitives to be expressed consistently from device meshes to threads. Building on Axe, we design a multi-granularity, distribution-aware DSL and compiler that composes thread-local control with collective operators in a single kernel. Experiments show that our unified approach can bring performance close to hand-tuned kernels on across latest GPU devices and multi-device environments and accelerator backends.

Keywords

Cite

@article{arxiv.2601.19092,
  title  = {Axe: A Simple Unified Layout Abstraction for Machine Learning Compilers},
  author = {Bohan Hou and Hongyi Jin and Guanjie Wang and Jinqi Chen and Yaxing Cai and Lijie Yang and Zihao Ye and Yaoyao Ding and Ruihang Lai and Tianqi Chen},
  journal= {arXiv preprint arXiv:2601.19092},
  year   = {2026}
}
R2 v1 2026-07-01T09:21:28.512Z