I recently gave a talk about our programming language and compiler for image processing, Halide. Click here for the slides.

Most of the credit goes to my former student Jonathan Ragan-Kelley and former postdoc Andrew Adams. As Marc Levoy summarized, with a bit of portable Halide code you can get:

• faster matrix multiply than Eigen
• faster Gaussian blur than Intel Performance Primitives
• faster Fourier transform than FFTW
• faster Local Laplacian than Adobe’s

In addition:

• it runs on device and in the cloud
• is supported on Linux, Windows, OSX, iOS, Android
• compiles to x86, ARM, MIPS, native client, OpenCL, OpenGL, CUDA, JavaScript, RenderScript (ISPs soon)

Halide started as a research project, but we made the compiler open source and it is now in heavy use at Google and various other companies. It dramatically improves programmers’ productivity.

For more, read our upcoming CACM research highlight, the original Halide paper, the PLDI paper, Jonathan’s thesis, or go to http://halide-lang.org/