Work

Distributed Halide

Year: 2016

Type: conference-paper

Abstract: Many image processing tasks are naturally expressed as a pipeline of small computational kernels known as stencils. Halide is a popular domain-specific language and compiler designed to implement image processing algorithms. Halide uses simple language constructs to express what to compute and a separate scheduling co-language for expressing when and where to perform the computation. This approach has demonstrated performance comparable to or better than hand-optimized code. Until now, however, Halide has been restricted to parallel shared memory execution, limiting its performance for memory-bandwidth-bound pipelines or large-scale image processing tasks.

Source: ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming

DOI: 10.1145/2851141.2851157

Authors: Tyler Denniston, Shoaib Kamil, Saman Amarasinghe

Institutions: Massachusetts Institute of Technology, Adobe Systems (United States)

Language: English

Topic: Parallel Computing and Optimization Techniques

Subfield: Hardware and Architecture

Field: Computer Science

Domain: Physical Sciences

Open Access status: hybrid

Funders: National Science Foundation, U.S. Department of Energy

Awards: XPS: FULL: DSD: Scalable High Performance with Halide and Simit Domain Specific Languages, TAS::89 0222::TAS NEW; ZETTABRICKS: A LANGUAGE COMPILER AND RUNTIME SYSTEM FOR ANYSCALE COMPUTING; PI - SAMAN AMARASINGHE, CAP3: A COMPUTER AIDED PERFORMANCE PROGRAMMING PLATFORM