Distributed Compute Labs is a Canadian non-profit organization responsible for developing the next-generation of compute networks.
Modern day research requires extensive computing power. Researchers are competing for limited resources, either in availability or cost. Meanwhile, the processing power of your computer is going to waste while your computer sits idle. Distributed Compute Labs is building you the tools needed to provide your compute power to the network as well as draw compute power from the network on-demand. Organizing processing power as a utility will accelerate compute-enabled research, innovation and discovery in Canada and across the globe.
Compute providers will receive Distributed Compute Credits (DCC) in exchange for their compute time. The protocol is designed to soak up unused cycles in otherwise active machines, or complete usage of a dedicated machine. DCC’s can then be used to deploy new compute tasks, or sold in the global marketplace.
Abundant compute resources are provided at a fraction of the cost of current commercial cloud computing services, disrupting existing market powers and accelerating compute-enabled research, innovation and discovery.
Connect your idle processing power from mobile devices to mainframes and help accelerate academic, research, institutional and enterprise computational objectives.
Deploy compute tasks from your computer and obtain results at an unprecedented cost savings and speed, all securely sourced from a decentralized compute supply.
Dr. Kristine Spekkens
Associate Professor, RMC Physics and Space Science
Dr. Gunnar Blohm
Associate Professor, Queen’s University Computational Neuroscience
Nizar Ladak
President & CEO Compute Ontario
Dr. Felicia Magpantay
Assistant Professor, Queen’s University Mathematics and Statistics
The DCP consists of a task scheduling system, an algorithm for efficiently distributing those tasks, and a compartment to securely compute them. Research groups can compute together using their own equipment, and reach out to a trusted network of global providers during peak demand.
The DCP will remain browser-compatible, enabling new opportunities in Distributed Computing. Micro-computations will provide passive income from compute power while online, which can be donated in lieu of advertisements or subscription costs.
Subsequent versions of the protocol will see the addition of more extensive implementation kits and auxiliary libraries that will plug into widely used and industry leading software.
DCP is standards-based and future-proof.
It is designed to be lightweight and of the simplest form, built on top of shared standards with a similar philosophy.
Javascript is the universal language which all websites and web applications are written. Javascript has been driven by rapid development from Google, Mozilla, Apple and Microsoft as they continuously compete for the fastest implementation in their browsers. It has the largest and youngest development community in the world. These factors make it the obvious choice for a distributed compute network.
Web Assembly was created for absolute performance without compromising security. Specific tasks can be hard-typed and compiled ahead of time allowing developers to benefit from the convenience and performance of both high and low level languages. Web assembly is a compile target for many native languages including C, C++, Julia, and Rust. Dozens of other compilers are under development by their respective communities.
DCP currently integrates WebGL 2 for rendering and general compute tasks. The browsers CUDA equivalent is under development; WebCompute Sharders. When they are released DCP will be their primary use case. GPU’s exist mostly at the edge of the network allowing us to gain more power than centralized GPU centers. We will enable GPU processing and in the future, GPGPU computation, parallel data and algorithms, physics simulations, machine learning, and neural networks.
