Repository navigation
Re-add convolution kernel #117
Description
I have looked at the convolution! code but there are some details I am missing.
- Can the kernel have any shape?
- It seems the method returns the same shape of the original but there is no option for padding or not padding.
Wouldn't it be nice to have some documentation and several tests showing how to use it?
I have done some tests:
https://github.com/davidbp/learn_julia/blob/master/GPU_compute/GPUArrays/GPUArrays_convolution.ipynb
Activity
SimonDanisch commentedon Jun 1, 2018
That was just a very simple first stab at it...
@americast, this could be one of your next tasks :)
davidbp commentedon Jun 1, 2018
In the documentation it might be ilustrative to add some gifs:
[-]convolution! documentation[/-][+]Re-add convolution kernel[/+]maleadt commentedon Jan 28, 2020
Update: this kernel has been removed, so would need to be re-added if somebody is interested. Old code here: 0868d9f
Would probably need to integrate with DSP.jl, but there doesn't seem to be an AbstractDSP.jl (cfr. FFT.jl).
maleadt commentedon Oct 19, 2026
Convolutions are provided by NNlib.jl, with generic GPU kernels and vendor-library back-ends (cuDNN, MIOpen), so this does not belong in GPUArrays.