add padding support for max pool

[donjuanplatinum]

No description provided.

[ivarflakstad]

Nice!
If you want this to work on cuda / metal you'll have to add padding functionality to the actual kernels as well.
You can see the cuda kernel here and the metal kernel here.
And update the code where the kernels are called of course, like here for metal.

[donjuanplatinum]

Nice! If you want this to work on cuda / metal you'll have to add padding functionality to the actual kernels as well. You can see the cuda kernel here and the metal kernel here. And update the code where the kernels are called of course, like here for metal.

I have modified the code :)