def compute_with_sca(type, xin, N, is_out=True, factor=None, coef=None, prefix=True,
optimize=False, change_coef=False):
if optimize:
xin = nlcpy.sca.convert_optimized_array(xin, xin.dtype)
if is_out:
xout = nlcpy.zeros_like(xin)
if optimize:
xout = nlcpy.sca.convert_optimized_array(xout, xout.dtype)
dxin, dxout = nlcpy.sca.create_descriptor((xin, xout))
else:
dxin = nlcpy.sca.create_descriptor((xin))
xout = None
_coef = nlcpy.zeros_like(coef) if change_coef else coef
if type == 'xya':
if xin.ndim < 2:
return True, True
desc = create_description_xya(
dxin, N, factor=factor, coef=_coef, prefix=prefix)
elif type == 'xza':
if xin.ndim < 3:
return True, True
desc = create_description_xza(
dxin, N, factor=factor, coef=_coef, prefix=prefix)
elif type == 'xwa':
if xin.ndim < 4:
return True, True
desc = create_description_xwa(
dxin, N, factor=factor, coef=_coef, prefix=prefix)
elif type == 'yza':
if xin.ndim < 3:
return True, True
desc = create_description_yza(
dxin, N, factor=factor, coef=_coef, prefix=prefix)
elif type == 'ywa':
if xin.ndim < 4:
return True, True
desc = create_description_ywa(
dxin, N, factor=factor, coef=_coef, prefix=prefix)
elif type == 'zwa':
if xin.ndim < 4:
return True, True
desc = create_description_zwa(
dxin, N, factor=factor, coef=_coef, prefix=prefix)
else:
raise TypeError
if is_out:
kern = nlcpy.sca.create_kernel(desc, dxout[...])
else:
kern = nlcpy.sca.create_kernel(desc)
if change_coef:
_coef[...] = coef
res = kern.execute()
return res, xout
def compute_with_naive(type, xin, N, factor=None, coef=None):
xout = nlcpy.zeros_like(xin)
if type == 'xyd':
if xin.ndim < 2:
return True
naive_xyd(xin, xout, N, factor=factor, coef=coef)
elif type == 'xzd':
if xin.ndim < 3:
return True
naive_xzd(xin, xout, N, factor=factor, coef=coef)
elif type == 'xwd':
if xin.ndim < 4:
return True
naive_xwd(xin, xout, N, factor=factor, coef=coef)
elif type == 'yzd':
if xin.ndim < 3:
return True
naive_yzd(xin, xout, N, factor=factor, coef=coef)
elif type == 'ywd':
if xin.ndim < 4:
return True
naive_ywd(xin, xout, N, factor=factor, coef=coef)
elif type == 'zwd':
if xin.ndim < 4:
return True
naive_zwd(xin, xout, N, factor=factor, coef=coef)
else:
raise TypeError
return xout
def compute_with_naive(type, xin, N, factor=None, coef=None, div=False, sub=False):
xout = nlcpy.zeros_like(xin)
if type == 'xa':
naive_xa(xin, xout, N, factor=factor, coef=coef, div=div, sub=sub)
else:
raise TypeError
return xout
def compute_with_sca(type, xin, N, is_out=True, factor=None, coef=None, prefix=True,
optimize=False, change_coef=False, div=False, sub=False):
if optimize:
xin = nlcpy.sca.convert_optimized_array(xin, xin.dtype)
if is_out:
xout = nlcpy.zeros_like(xin)
if optimize:
xout = nlcpy.sca.convert_optimized_array(xout, xout.dtype)
dxin, dxout = nlcpy.sca.create_descriptor((xin, xout))
else:
dxin = nlcpy.sca.create_descriptor((xin))
xout = None
_coef = nlcpy.ones_like(coef) if change_coef else coef
if type == 'xa':
desc = create_description_xa(
dxin, N, factor=factor, coef=_coef, prefix=prefix, div=div, sub=sub)
else:
raise TypeError
if is_out:
kern = nlcpy.sca.create_kernel(desc, dxout[...])
else:
kern = nlcpy.sca.create_kernel(desc)
if change_coef:
_coef[...] = coef
res = kern.execute()
return res, xout
def test_me_case_1():
a = ny.array([[10, 7, 4], [3, 2, 1]])
na = ny.array(a)
ans = ny.std(na, axis=0)
out = ny.zeros_like(ans)
dst = ny.std(na, axis=0, out=out)
assert_array_almost_equal(dst.get(), out.get())
def compute_with_naive(type, xin, N, factor=None, coef=None):
xout = nlcpy.zeros_like(xin)
if type == 'xyzw':
if xin.ndim < 4:
return True
naive_xyzw(xin, xout, N, factor=factor, coef=coef)
else:
raise TypeError
return xout
def compute_with_naive(type, xin, N, factor=None, coef=None):
xout = nlcpy.zeros_like(xin)
if type == 'xa':
naive_xa(xin, xout, N, factor=factor, coef=coef)
elif type == 'ya':
if xin.ndim < 2:
return True
naive_ya(xin, xout, N, factor=factor, coef=coef)
elif type == 'za':
if xin.ndim < 3:
return True
naive_za(xin, xout, N, factor=factor, coef=coef)
elif type == 'wa':
if xin.ndim < 4:
return True
naive_wa(xin, xout, N, factor=factor, coef=coef)
else:
raise TypeError
return xout
def ca4(arg):
d = arg[0]
t = arg[1]
a = arg[2]
ss = 1
for x in d:
ss *= x
if t == 'int32' or t == 'int64' or t == 'uint32' or t == 'uint64':
seed = [random.randint(1, ss * ss) for i in range(ss)]
elif t == 'float64' or t == 'float32':
seed = [random.random() for i in range(ss)]
else:
pass
na = ny.array(seed)
ans = ny.mean(na, axis=a)
out = ny.zeros_like(ans)
dst = ny.mean(na, axis=a, out=out)
return dst, out
def test_zeros_like_reshape_nlcpy_only(self, dtype, order):
a = testing.shaped_arange((2, 3, 4), nlcpy, dtype)
b = nlcpy.zeros_like(a, shape=self.shape)
c = nlcpy.zeros(self.shape, order=order, dtype=dtype)
testing.assert_array_equal(b, c)
def test_zeros_like_subok(self):
a = nlcpy.ndarray((2, 3, 4))
return nlcpy.zeros_like(a, subok=True)
