Write a Blog >>
Mon 11 Oct 2021 14:30 - 14:40 - Helping Programmers Build Chair(s): Michael Coblenz

Writing high-performance image processing code is challenging and labor-intensive. Automatic compiler optimization is appealing, and has been the subject of much research, but human programmers are still often able to optimize their code in ways that significantly outperform automatic compilers. As a result, high-performance image processing code still relies on painstaking hand optimization. The Halide programming language simplifies this task by decoupling high-level algorithms from “schedules” which define how to optimize those computations, including choices that control parallelism and cache locality. However, even with this abstraction, it is still challenging for Halide programmers to understand complicated scheduling strategies and productively write valid, optimized schedules. To address this, we propose a programming support method called “guided optimization.” Guided optimization provides programmers a set of valid optimization options and interactive feedback about their current choices, which enables them to comprehend and efficiently optimize the image processing code without the time-consuming process of trial-and-error in traditional text editors. We implemented a proof-of-concept system, Roly-poly, which integrates guided optimization, program visualization, and scheduling cost estimation to support the comprehension and development of efficient Halide image processing codes. We conducted a user study with novice Halide programmers and confirmed that Roly-poly and its guided optimization was informative, increased productivity, and resulted in higher-performing schedules in less time.

Mon 11 Oct

Displayed time zone: Central Time (US & Canada) change

14:10 - 14:50
Helping Programmers BuildResearch Papers / Journal-First Presentations
Chair(s): Michael Coblenz University of Maryland at College Park
Where-provenance for Bidirectional Editing in SpreadsheetsFull paper
Research Papers
Jack Williams Microsoft Research, Cambridge, Andrew D. Gordon Microsoft Research and University of Edinburgh
Applications of model-driven engineering in cyber-physical systems: A systematic mapping studyJournal-first
Journal-First Presentations
Mustafa Abshir Mohamed Independent Researcher, Moharram Challenger University of Antwerp, Geylani Kardas Ege University
Guided Optimization for Image Processing PipelinesShort paper
Research Papers
Yuka Ikarashi MIT CSAIL, Jonathan Ragan-Kelley Massachusetts Institute of Technology, Tsukasa Fukusato The University of Tokyo, Jun Kato National Institute of Advanced Industrial Science and Technology (AIST), Takeo Igarashi The University of Tokyo
A Toolkit for Building More Adaptable User Interfaces for Vision-impaired UsersShort paper
Research Papers
Calvin Luy Monash University, Jeremy Law Monash University, Lily Ho Monash University, Richard Matheson Monash University, Tracey Cai Monash University, Anuradha Madugalla Monash University, Australia, John Grundy Monash University