Unifying Spatial Accelerator Compilation With Idiomatic and Modular Transformations

Jian Weng, Sihao Liu, Dylan Kupsh, Tony Nowatzki

Research output: Contribution to journalArticlepeer-review

3 Scopus citations

Abstract

Spatial accelerators provide high performance, energy efficiency, and flexibility. Recent design frameworks enable these architectures to be quickly designed and customized to a domain. However, constructing a compiler for this immense design space is challenging, first because accelerators express programs with high-level idioms that are difficult to recognize. Second, it is unpredictable whether certain transformations are beneficial or will lead to infeasible hardware mappings. Our work develops a general spatial-accelerator compiler with two key ideas. First, we propose an approach to recognize and represent useful dataflow idioms, along with a novel idiomatic memory representation. Second, we propose the principle of modular compilation, which combines hardware-aware transformation selection and an iterative approach to handle uncertainty. Our compiler achieves 2.3 × speedup, and 98.7 × area-normalized speedup over high-end server central processing unit (CPU).

Original languageEnglish (US)
Pages (from-to)59-69
Number of pages11
JournalIEEE Micro
Volume42
Issue number5
DOIs
StatePublished - 2022

Bibliographical note

Publisher Copyright:
© 1981-2012 IEEE.

ASJC Scopus subject areas

  • Software
  • Hardware and Architecture
  • Electrical and Electronic Engineering

Fingerprint

Dive into the research topics of 'Unifying Spatial Accelerator Compilation With Idiomatic and Modular Transformations'. Together they form a unique fingerprint.

Cite this