ConvolutionSeperable example

Accelerated Computing CUDA CUDA Programming and Performance
1

Hi all,

I am trying to run some tests on convolutionSeperable from SDK examples with different kernel radius. How do I change the kernel radius aligned parameter in relative to kernel radius?

Thanks,
ngs

2

I think the aligned radius is just the next largest multiple of 16.

Kernel radius 1-16 is 16 aligned. Radius 17-32 is 32 aligned.

If I remember correctly, it’s simple, but it seems complex.

Hope that helps!