maciejczyzewski / fast_gpu_voronoi Goto Github PK

Hey Maciej,

Your code does not seem to work under Linux. Here's what i got from your example. Any clues to how I could fix it ?

pyopencl._cl.RuntimeError: clBuildProgram failed: BUILD_PROGRAM_FAILURE - clBuildProgram failed: BUILD_PROGRAM_FAILURE - clBuildProgram failed: BUILD_PROGRAM_FAILURE

Build on <pyopencl.Device 'Quadro P4000' on 'NVIDIA CUDA' at 0x55f3a6f046a0>:

:2:26: warning: unknown OpenCL extension 'cl_ext_device_fission' - ignoring
#pragma OPENCL EXTENSION cl_ext_device_fission : enable
^
:35:29: error: expected expression
int A = (step * cos(float( ((6.28/12) * i) )));
^
:36:29: error: expected expression
int B = (step * sin(float( ((6.28/12) * i) )));

the versions are not working

@maciejczyzewski and @KamilPiechowiak ,

I was wondering if there is a way to add a few more attributes to the output. So far you are storing X/Y and the matrix of points labelled by the region it belongs to.

I noticed that Scipy Voronoi implementation (https://docs.scipy.org/doc/scipy/reference/generated/scipy.spatial.Voronoi.html) actually has really convenient access to vertices and regions. While I know it is possible to extract similar attributes likely via scipy.ConvexHull, it might also add the overhead to the calculation that is "almost" done in openCL performant code.

Do you think it would be difficult to actually do the calc directly in OpenCL and expose similar attributes?

Thanks,
Anton.

PS. Why am I asking?
I am trying to make an open-source library for stippling in python that relies on:
ReScience-Archives/Rougier-2017#3
They are using Voronoi diagrams there obtained via Scipy, and so far according to the code profiler that is one of the slower spots (I have sped up some of the calculation via numba).