Graphite is a professional 2D graphics editor for photo editing, image manipulation, graphic design, illustration, data visualization, batch processing, and technical art. It is designed to run on a variety of machines, from mobile hardware like iPads or web browsers on midrange laptops up to beefy workstations with dozens of CPU cores and multiple GPUs.
To provide a responsive user experience, its architecture is made to support the use of distributed computation to make up for deficiencies in local compute power. Resulting productivity benefits will scale for users on hardware ranging from low-end mobile devices up to high-end workstations because documents and use cases can grow to great complexity.
This article explores the current thinking about the problems and potential engineering solutions involved in Graphite and Graphene for building a high-performance distributed computing runtime environment. We are only just embarking on the graph engine implementation, meaning this post describes theoretical approaches to theoretical challenges. The aim is to shed light on what will need to be built and what we currently believe is the trajectory for this work. We hope this post prompts discussions that evolve the concepts and approaches described herein. If this topic sounds interesting and you have feedback, please get in touch.