dpt.sum over axis for transposed input array #1391
Description
Due to merging gh-1364 the below example has been broken:
import numpy, dpctl, dpctl.tensor as dpt
from math import prod
sh = (1, 2, 3)
a = dpt.arange(prod(sh), dtype='i4')
a = dpt.reshape(a, sh)
aT = dpt.permute_dims(a, (2, 1, 0))
aT.shape
# Out: (3, 2, 1)
dpt.sum(aT, axis=2)
# Out:
# usm_ndarray([[0, 1],
# [2, 3],
# [4, 5]])
b = numpy.arange(prod(sh), dtype='i4').reshape(sh)
bT = b.T
bT.shape
# Out: (3, 2, 1)
bT.sum(axis=2)
# Out:
# array([[0, 3],
# [1, 4],
# [2, 5]])Reactions are currently unavailable