Nigel
/
mindspore
forked from mindspore-Ecosystem/mindspore
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

add test case about switch simplify

This commit is contained in:
chenfei 2022-06-17 15:12:31 +08:00
parent 8427a32034
commit b140512d4b
3 changed files with 190 additions and 35 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

64
tests/st/control/test_abstract_funcgraph.py Normal file
View File

@ -0,0 +1,64 @@
# Copyright 2022 Huawei Technologies Co., Ltd
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# ============================================================================
from mindspore.nn import Cell
from mindspore.common import Tensor, dtype
import mindspore.ops.operations as P
import mindspore.ops.functional as F
import numpy as np
import pytest
@pytest.mark.level0
@pytest.mark.platform_x86_gpu_training
@pytest.mark.env_onecard
def test_watch_get_func_graphs_from_abstract():
"""
Feature: Get func_graph from abstract.
Description: Watching the function of getting func graph from abstract.
Expectation: Output correct.
"""
class Net(Cell):
def __init__(self):
super().__init__()
self.op = P.Add()
def construct(self, x, y):
for t in range(2):
if y != x:
if x > 4:
x = y / x
y = 1 - x
y = y - y
elif x > 2:
y = x - 1
else:
y = 3 - y
y = t * x
elif x != 3:
x = x - x
if x == y:
continue
return self.op(x, y)
x = np.array([4], np.float32)
y = np.array([1], np.float32)
net = Net()
grad_net = F.grad(net, grad_position=(0, 1))
fgrad = grad_net(Tensor(x), Tensor(y))
assert fgrad[0] == Tensor([2], dtype.float32)
assert fgrad[1] == Tensor([0], dtype.float32)

35
tests/st/control/test_if_by_if.py
View File

@ -70,41 +70,6 @@ def test_if_by_if_basic():
assert np.allclose(out_ms.asnumpy(), out_np)
@pytest.mark.level0
@pytest.mark.platform_arm_ascend_training
@pytest.mark.platform_x86_ascend_training
@pytest.mark.env_onecard
def test_tensor_condition():
"""
Feature: control flow function.
Description: Switch condition is tensor determinate condition.
Expectation: Null.
"""
class Net(Cell):
def construct(self, x, y):
if x < 5:
x = y + 2
for p in range(1):
x = p * x
if x >= y:
x = 2 * x
if x <= 5:
x = 2 + y
elif x >= 2:
x = x * y
return x + y
context.set_context(mode=context.GRAPH_MODE)
x = np.array([3], np.float32)
y = np.array([1], np.float32)
net = Net()
out = net(Tensor(x), Tensor(y))
assert np.allclose(out.asnumpy(), np.array([4.], np.float32))
@pytest.mark.level0
@pytest.mark.platform_arm_ascend_training
@pytest.mark.platform_x86_ascend_training

126
tests/st/control/test_switch_simplify.py Normal file
View File

@ -0,0 +1,126 @@
# Copyright 2022 Huawei Technologies Co., Ltd
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# ============================================================================
from mindspore.nn import Cell
from mindspore.common import Tensor
import mindspore.ops.operations as P
import mindspore.ops.functional as F
from mindspore import context
import numpy as np
import pytest
@pytest.mark.level0
@pytest.mark.platform_x86_gpu_training
@pytest.mark.env_onecard
def test_switch_simplify_avoid_dead_node():
"""
Feature: Switch simplify pass.
Description: If switch simplify pass can't simplify constant tensor condition,
dead node will exist in backend.
Expectation: output correct.
"""
class Net(Cell):
def __init__(self):
super().__init__()
self.op = P.Add()
def construct(self, x, y):
if y != x:
x = y - 3
elif x == 4:
for r in range(2):
x = 1 / y
if x > 2:
y = y + 3
y = y - y
y = y * x
elif y >= x:
x = x * x
elif x > y:
x = y - r
else:
y = 2 + x
for _ in range(2):
x = x * y
x = x - 3
y = y + 2
if x > 3:
break
if x > 2:
break
elif x == y:
if y <= x:
y = x / 2
x = 3 + y
x = x * 2
elif x == 2:
x = y * y
elif x < y:
y = 2 * y
elif x != 2:
y = x * y
while x != 5:
break
return self.op(x, y)
x = np.array([4], np.float32)
y = np.array([4], np.float32)
net = Net()
out = net(Tensor(x), Tensor(y))
grad_net = F.grad(net, grad_position=(0, 1))
fgrad = grad_net(Tensor(x), Tensor(y))
sgrad_net = F.grad(grad_net)
sgrad = sgrad_net(Tensor(x), Tensor(y))
assert np.allclose(out.asnumpy(), np.array([-19.75], np.float32))
assert np.allclose(fgrad[0].asnumpy(), np.array([0.], np.float32))
assert np.allclose(fgrad[1].asnumpy(), np.array([-2.03125], np.float32))
assert np.allclose(sgrad.asnumpy(), np.array([0.], np.float32))
@pytest.mark.level0
@pytest.mark.platform_arm_ascend_training
@pytest.mark.platform_x86_ascend_training
@pytest.mark.env_onecard
def test_tensor_condition():
"""
Feature: control flow function.
Description: Switch condition is tensor determinate condition.
Expectation: Null.
"""
class Net(Cell):
def construct(self, x, y):
if x < 5:
x = y + 2
for p in range(1):
x = p * x
if x >= y:
x = 2 * x
if x <= 5:
x = 2 + y
elif x >= 2:
x = x * y
return x + y
context.set_context(mode=context.GRAPH_MODE)
x = np.array([3], np.float32)
y = np.array([1], np.float32)
net = Net()
out = net(Tensor(x), Tensor(y))
assert np.allclose(out.asnumpy(), np.array([4.], np.float32))