Sushi2 library

Matrix Library for JavaScript

This library is intended to be the fastest matrix library for JavaScript, with the power of GPU computing. To gain best performance, WebCL technology is used to access GPU from JavaScript.

Interactive getting started on the browser

Documents (work in progress)

Build for use in node.js

Since this project is written in TypeScript, transpiling to JavaScript is necessary.

Package from npm repository contains transplied JavaScript, so simply installing them is enough.

npm install milsushi2

If you intend to modify code, download and build.

git clone https://github.com/mil-tokyo/sushi2
cd sushi2
npm install
npm run build

Sushi2 depends on node-opencl for GPU computing which allows dramatically faster computation. This dependency is optional, so even the installation of node-opencl fails, Sushi2 can work without it.

In my environment (Ubuntu 14.04 + NVIDIA CUDA 7.5), installation with node-opencl requires additional environment variables.

CPLUS_INCLUDE_PATH=/usr/local/cuda/include LIBRARY_PATH=/usr/local/cuda/lib64 npm install milsushi2

Build for use in web browser

For simply using the library, download the js file from releases page.

To make single JavaScript file for web browsers, type the following commands:

git clone https://github.com/mil-tokyo/sushi2
cd sushi2
npm install
npm run build
npm run browserify

This will generate browser/milsushi2.js (without WebCL support), and browser/milsushi2_cl.js (WebCL support version).

Usage in node.js

You can import the module by require('milsushi2').

Hello world in node shell

var $M = require('milsushi2');
$M.initcl();//OpenCL initialization, true if succeeds
var x = $M.jsa2mat([[1,2],[3,4]]);
var y = $M.jsa2mat([[0.1,0.5],[0.7,0.0]]);
$M.plus(x, y);

Usage in web browser

By loading them from html page (<script src="milsushi2.js"></script>), milsushi2 global object is generated.

Hello world in html

<script>
var $M = milsushi2;
var x = $M.jsa2mat([[1,2],[3,4]]);
var y = $M.jsa2mat([[0.1,0.5],[0.7,0.0]]);
alert($M.plus(x, y));
</script>

To use WebCL for GPU computing, use milsushi2_cl.js instead of milsushi2.

Unfortunately, currently a plugin is needed to enable WebCL. We tested on webcl-firefox plugin with Firefox 32. Compiled version of webcl-firefox plugin for Linux is here (Ubuntu 14.04 + CUDA 7.5, commit d87447f, License: MPL 2.0).

If WebCL is enabled, $M.initcl() should return true.

License

MIT