Redwood forest



Copperhead is a functional data-parallel Python dialect, along with a runtime that currently supports CUDA-enabled GPUs.

from copperhead import *

@cu def saxpy(a, x, y): return [a * xi + yi for xi, yi in zip(x, y)]

Copperhead is available at Google Code.


Damascene is an reexamination of the gPb image contour detector. By reevaluating components of the gPb algorithm in a parallel context, and creating an eigensolver specially suited to Normalized Cuts image segmentation, we have made high-quality image contour detection more efficient. Damascene is a parallel implementation of these ideas using CUDA. On an Nvidia GTX280, we can perform image contour detection 130x faster than the original serial CPU gPb detector, while still providing the same accuracy on the contour benchmark of the Berkeley Segmentation Dataset. For more details, please reference our ICCV 2009 paper.

Download Damascene here.


GPUSVM is an implementation of Support Vector Machine training and classification, using Nvidia graphics processors programmed through CUDA. On an Nvidia 8800GTX, we achieve up to 35x faster training, and 24x faster classification over CPU implementations, while achieving the same accuracy. For more details see our ICML paper.

Download GPUSVM here: GPUSVM

PyCUDA and CodePy

I'm a fan of Andreas Klöckner's CodePy and PyCUDA projects, and use them in my own work. I keep my development branches published on GitHub.