s[OL].reorder(rco, ryo, rxo, rcm, rym, rxm, rci, ryi, rxi, n, f, y, x)
In tune_conv2d_cuda.py, reoder OL‘s [n, f, y, x] as the inner axis, what it means?
why not reorder [n, f, y, x] as the outer axis?
I get the lower code for OL, there is no differences if or not reorder with [n, f, y, x].