int16 uint8 bool supported akg ops

mindspore/ops/_op_impl/akg/gpu/equal.py

@@ -23,6 +23,9 @@ equal_op_info = AkgGpuRegOp("Equal") \
     .dtype_format(DataType.F16_Default, DataType.F16_Default, DataType.BOOL_Default) \
     .dtype_format(DataType.F32_Default, DataType.F32_Default, DataType.BOOL_Default) \
     .dtype_format(DataType.I32_Default, DataType.I32_Default, DataType.BOOL_Default) \
+    .dtype_format(DataType.I16_Default, DataType.I16_Default, DataType.BOOL_Default) \
+    .dtype_format(DataType.U8_Default, DataType.U8_Default, DataType.BOOL_Default) \
+    .dtype_format(DataType.BOOL_Default, DataType.BOOL_Default, DataType.BOOL_Default) \
     .get_op_info()
 
 

mindspore/ops/_op_impl/akg/gpu/greater_equal.py

@@ -23,6 +23,8 @@ greater_equal_op_info = AkgGpuRegOp("GreaterEqual") \
     .dtype_format(DataType.F16_Default, DataType.F16_Default, DataType.BOOL_Default) \
     .dtype_format(DataType.F32_Default, DataType.F32_Default, DataType.BOOL_Default) \
     .dtype_format(DataType.I32_Default, DataType.I32_Default, DataType.BOOL_Default) \
+    .dtype_format(DataType.I16_Default, DataType.I16_Default, DataType.BOOL_Default) \
+    .dtype_format(DataType.U8_Default, DataType.U8_Default, DataType.BOOL_Default) \
     .get_op_info()
 
 

mindspore/ops/_op_impl/akg/gpu/lessequal.py

@@ -23,6 +23,8 @@ lessequal_op_info = AkgGpuRegOp("LessEqual") \
     .dtype_format(DataType.F16_Default, DataType.F16_Default, DataType.BOOL_Default) \
     .dtype_format(DataType.F32_Default, DataType.F32_Default, DataType.BOOL_Default) \
     .dtype_format(DataType.I32_Default, DataType.I32_Default, DataType.BOOL_Default) \
+    .dtype_format(DataType.I16_Default, DataType.I16_Default, DataType.BOOL_Default) \
+    .dtype_format(DataType.U8_Default, DataType.U8_Default, DataType.BOOL_Default) \
     .get_op_info()
 
 

mindspore/ops/_op_impl/akg/gpu/notequal.py

@@ -23,6 +23,9 @@ notequal_op_info = AkgGpuRegOp("NotEqual") \
     .dtype_format(DataType.F16_Default, DataType.F16_Default, DataType.BOOL_Default) \
     .dtype_format(DataType.F32_Default, DataType.F32_Default, DataType.BOOL_Default) \
     .dtype_format(DataType.I32_Default, DataType.I32_Default, DataType.BOOL_Default) \
+    .dtype_format(DataType.I16_Default, DataType.I16_Default, DataType.BOOL_Default) \
+    .dtype_format(DataType.U8_Default, DataType.U8_Default, DataType.BOOL_Default) \
+    .dtype_format(DataType.BOOL_Default, DataType.BOOL_Default, DataType.BOOL_Default) \
     .get_op_info()
 
 

mindspore/ops/_op_impl/akg/gpu/squeeze.py

@@ -22,6 +22,10 @@ squeeze_op_info = AkgGpuRegOp("Squeeze") \
     .attr("axis", "optional", "listInt") \
     .dtype_format(DataType.F16_Default, DataType.F16_Default) \
     .dtype_format(DataType.F32_Default, DataType.F32_Default) \
+    .dtype_format(DataType.I32_Default, DataType.I32_Default) \
+    .dtype_format(DataType.I16_Default, DataType.I16_Default) \
+    .dtype_format(DataType.U8_Default, DataType.U8_Default) \
+    .dtype_format(DataType.BOOL_Default, DataType.BOOL_Default) \
     .get_op_info()
 
 

mindspore/ops/_op_impl/akg/gpu/squeeze_grad.py

@@ -22,6 +22,10 @@ squeeze_grad_op_info = AkgGpuRegOp("SqueezeGrad") \
     .attr("x_shape", "required", "listInt") \
     .dtype_format(DataType.F16_Default, DataType.F16_Default) \
     .dtype_format(DataType.F32_Default, DataType.F32_Default) \
+    .dtype_format(DataType.I32_Default, DataType.I32_Default) \
+    .dtype_format(DataType.I16_Default, DataType.I16_Default) \
+    .dtype_format(DataType.U8_Default, DataType.U8_Default) \
+    .dtype_format(DataType.BOOL_Default, DataType.BOOL_Default) \
     .get_op_info()
 
 

mindspore/ops/_op_impl/akg/gpu/tile.py

@@ -22,6 +22,8 @@ tile_op_info = AkgGpuRegOp("Tile") \
     .attr("multiples", "required", "listInt") \
     .dtype_format(DataType.F16_Default, DataType.F16_Default) \
     .dtype_format(DataType.F32_Default, DataType.F32_Default) \
+    .dtype_format(DataType.I32_Default, DataType.I32_Default) \
+    .dtype_format(DataType.I16_Default, DataType.I16_Default) \
     .get_op_info()
 
 

tests/st/ops/gpu/test_equal_op.py

@@ -65,6 +65,21 @@ def test_equal():
     y2_np = np.array([0, 1, -3]).astype(np.int32)
     y2 = Tensor(y2_np)
     expect2 = np.equal(x2_np, y2_np)
+    x3_np = np.array([0, 1, 3]).astype(np.int16)
+    x3 = Tensor(x3_np)
+    y3_np = np.array([0, 1, -3]).astype(np.int16)
+    y3 = Tensor(y3_np)
+    expect3 = np.equal(x3_np, y3_np)
+    x4_np = np.array([0, 1, 4]).astype(np.uint8)
+    x4 = Tensor(x4_np)
+    y4_np = np.array([0, 1, 3]).astype(np.uint8)
+    y4 = Tensor(y4_np)
+    expect4 = np.equal(x4_np, y4_np)
+    x5_np = np.array([True, False, True]).astype(bool)
+    x5 = Tensor(x5_np)
+    y5_np = np.array([True, False, False]).astype(bool)
+    y5 = Tensor(y5_np)
+    expect5 = np.equal(x5_np, y5_np)
 
     context.set_context(mode=context.PYNATIVE_MODE, device_target="GPU")
     equal = NetEqual()
@@ -77,6 +92,17 @@ def test_equal():
     output2 = equal(x2, y2)
     assert np.all(output2.asnumpy() == expect2)
     assert output2.shape == expect2.shape
+    output3 = equal(x3, y3)
+    assert np.all(output3.asnumpy() == expect3)
+    assert output3.shape == expect3.shape
+    output4 = equal(x4, y4)
+    assert np.all(output4.asnumpy() == expect4)
+    assert output4.shape == expect4.shape
+    output5 = equal(x5, y5)
+    assert np.all(output5.asnumpy() == expect5)
+    assert output5.shape == expect5.shape
+
+
 
     context.set_context(mode=context.GRAPH_MODE, device_target="GPU")
     equal = NetEqual()
@@ -89,6 +115,15 @@ def test_equal():
     output2 = equal(x2, y2)
     assert np.all(output2.asnumpy() == expect2)
     assert output2.shape == expect2.shape
+    output3 = equal(x3, y3)
+    assert np.all(output3.asnumpy() == expect3)
+    assert output3.shape == expect3.shape
+    output4 = equal(x4, y4)
+    assert np.all(output4.asnumpy() == expect4)
+    assert output4.shape == expect4.shape
+    output5 = equal(x5, y5)
+    assert np.all(output5.asnumpy() == expect5)
+    assert output5.shape == expect5.shape
 
 
 @pytest.mark.level0
@@ -98,18 +133,45 @@ def test_notequal():
     x0 = Tensor(np.array([[1.2, 1], [1, 0]]).astype(np.float32))
     y0 = Tensor(np.array([[1, 2]]).astype(np.float32))
     expect0 = np.array([[True, True], [False, True]])
+    x1 = Tensor(np.array([[2, 1], [1, 1]]).astype(np.int16))
+    y1 = Tensor(np.array([[1, 2]]).astype(np.int16))
+    expect1 = np.array([[True, True], [False, True]])
+    x2 = Tensor(np.array([[2, 1], [1, 2]]).astype(np.uint8))
+    y2 = Tensor(np.array([[1, 2]]).astype(np.uint8))
+    expect2 = np.array([[True, True], [False, False]])
+    x3 = Tensor(np.array([[False, True], [True, False]]).astype(bool))
+    y3 = Tensor(np.array([[True, False]]).astype(bool))
+    expect3 = np.array([[True, True], [False, False]])
 
     context.set_context(mode=context.PYNATIVE_MODE, device_target="GPU")
     notequal = NetNotEqual()
     output0 = notequal(x0, y0)
     assert np.all(output0.asnumpy() == expect0)
     assert output0.shape == expect0.shape
+    output1 = notequal(x1, y1)
+    assert np.all(output1.asnumpy() == expect1)
+    assert output1.shape == expect1.shape
+    output2 = notequal(x2, y2)
+    assert np.all(output2.asnumpy() == expect2)
+    assert output2.shape == expect2.shape
+    output3 = notequal(x3, y3)
+    assert np.all(output3.asnumpy() == expect3)
+    assert output3.shape == expect3.shape
 
     context.set_context(mode=context.GRAPH_MODE, device_target="GPU")
     notequal = NetNotEqual()
     output0 = notequal(x0, y0)
     assert np.all(output0.asnumpy() == expect0)
     assert output0.shape == expect0.shape
+    output1 = notequal(x1, y1)
+    assert np.all(output1.asnumpy() == expect1)
+    assert output1.shape == expect1.shape
+    output2 = notequal(x2, y2)
+    assert np.all(output2.asnumpy() == expect2)
+    assert output2.shape == expect2.shape
+    output3 = notequal(x3, y3)
+    assert np.all(output3.asnumpy() == expect3)
+    assert output3.shape == expect3.shape
 
 
 @pytest.mark.level0
@@ -119,15 +181,33 @@ def test_greaterqual():
     x0 = Tensor(np.array([[1.2, 1], [1, 0]]).astype(np.float32))
     y0 = Tensor(np.array([[1, 2]]).astype(np.float32))
     expect0 = np.array([[True, False], [True, False]])
+    x1 = Tensor(np.array([[2, 1], [1, 1]]).astype(np.int16))
+    y1 = Tensor(np.array([[1, 2]]).astype(np.int16))
+    expect1 = np.array([[True, False], [True, False]])
+    x2 = Tensor(np.array([[2, 1], [1, 2]]).astype(np.uint8))
+    y2 = Tensor(np.array([[1, 2]]).astype(np.uint8))
+    expect2 = np.array([[True, False], [True, True]])
 
     context.set_context(mode=context.PYNATIVE_MODE, device_target="GPU")
     gequal = NetGreaterEqual()
     output0 = gequal(x0, y0)
     assert np.all(output0.asnumpy() == expect0)
     assert output0.shape == expect0.shape
+    output1 = gequal(x1, y1)
+    assert np.all(output1.asnumpy() == expect1)
+    assert output1.shape == expect1.shape
+    output2 = gequal(x2, y2)
+    assert np.all(output2.asnumpy() == expect2)
+    assert output2.shape == expect2.shape
 
     context.set_context(mode=context.GRAPH_MODE, device_target="GPU")
     gequal = NetGreaterEqual()
     output0 = gequal(x0, y0)
     assert np.all(output0.asnumpy() == expect0)
     assert output0.shape == expect0.shape
+    output1 = gequal(x1, y1)
+    assert np.all(output1.asnumpy() == expect1)
+    assert output1.shape == expect1.shape
+    output2 = gequal(x2, y2)
+    assert np.all(output2.asnumpy() == expect2)
+    assert output2.shape == expect2.shape

tests/st/ops/gpu/test_lessequal_op.py

@@ -38,12 +38,27 @@ def test_lessequal():
     x = Tensor(np.array([[1, 2, 3]]).astype(np.float32))
     y = Tensor(np.array([[2]]).astype(np.float32))
     expect = [[True, True, False]]
+    x1 = Tensor(np.array([[1, 2, 3]]).astype(np.int16))
+    y1 = Tensor(np.array([[2]]).astype(np.int16))
+    expect = [[True, True, False]]
+    x2 = Tensor(np.array([[1, 2, 3]]).astype(np.uint8))
+    y2 = Tensor(np.array([[2]]).astype(np.uint8))
+    expect = [[True, True, False]]
+
     context.set_context(mode=context.PYNATIVE_MODE, device_target="GPU")
     lessequal = Net()
     output = lessequal(x, y)
     assert np.all(output.asnumpy() == expect)
+    output = lessequal(x1, y1)
+    assert np.all(output.asnumpy() == expect)
+    output = lessequal(x2, y2)
+    assert np.all(output.asnumpy() == expect)
 
     context.set_context(mode=context.GRAPH_MODE, device_target="GPU")
     lessequal = Net()
     output = lessequal(x, y)
     assert np.all(output.asnumpy() == expect)
+    output = lessequal(x1, y1)
+    assert np.all(output.asnumpy() == expect)
+    output = lessequal(x2, y2)
+    assert np.all(output.asnumpy() == expect)

tests/st/ops/gpu/test_squeeze_op.py

@@ -0,0 +1,79 @@
+# Copyright 2019 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.
+# ============================================================================
+import numpy as np
+
+import mindspore.context as context
+import mindspore.nn as nn
+from mindspore import Tensor
+from mindspore.ops import operations as P
+
+context.set_context(mode=context.PYNATIVE_MODE, device_target="GPU")
+
+
+class Net(nn.Cell):
+    def __init__(self):
+        super(Net, self).__init__()
+        self.squeeze = P.Squeeze()
+
+    def construct(self, tensor):
+        return self.squeeze(tensor)
+
+
+def test_net_bool():
+    x = np.random.randn(1, 16, 1, 1).astype(np.bool)
+    net = Net()
+    output = net(Tensor(x))
+    print(output.asnumpy())
+    assert np.all(output.asnumpy() == x.squeeze())
+
+
+def test_net_uint8():
+    x = np.random.randn(1, 16, 1, 1).astype(np.uint8)
+    net = Net()
+    output = net(Tensor(x))
+    print(output.asnumpy())
+    assert np.all(output.asnumpy() == x.squeeze())
+
+
+def test_net_int16():
+    x = np.random.randn(1, 16, 1, 1).astype(np.int16)
+    net = Net()
+    output = net(Tensor(x))
+    print(output.asnumpy())
+    assert np.all(output.asnumpy() == x.squeeze())
+
+
+def test_net_int32():
+    x = np.random.randn(1, 16, 1, 1).astype(np.int32)
+    net = Net()
+    output = net(Tensor(x))
+    print(output.asnumpy())
+    assert np.all(output.asnumpy() == x.squeeze())
+
+
+def test_net_float16():
+    x = np.random.randn(1, 16, 1, 1).astype(np.float16)
+    net = Net()
+    output = net(Tensor(x))
+    print(output.asnumpy())
+    assert np.all(output.asnumpy() == x.squeeze())
+
+
+def test_net_float32():
+    x = np.random.randn(1, 16, 1, 1).astype(np.float32)
+    net = Net()
+    output = net(Tensor(x))
+    print(output.asnumpy())
+    assert np.all(output.asnumpy() == x.squeeze())

tests/st/ops/gpu/test_tile_op.py

@@ -33,6 +33,27 @@ mul1 = (2, 2, 2)
 input_x2 = np.arange(1).reshape((1, 1, 1)).astype(np.float32)
 mul2 = (1, 1, 1)
 
+input_32_x0 = np.arange(2).reshape((2, 1, 1)).astype(np.int32)
+mul_32_0 = (8, 1, 1)
+input_32_x1 = np.arange(32).reshape((2, 4, 4)).astype(np.int32)
+mul_32_1 = (2, 2, 2)
+input_32_x2 = np.arange(1).reshape((1, 1, 1)).astype(np.int32)
+mul_32_2 = (1, 1, 1)
+
+input_16_x0 = np.arange(2).reshape((2, 1, 1)).astype(np.int16)
+mul_16_0 = (8, 1, 1)
+input_16_x1 = np.arange(32).reshape((2, 4, 4)).astype(np.int16)
+mul_16_1 = (2, 2, 2)
+input_16_x2 = np.arange(1).reshape((1, 1, 1)).astype(np.int16)
+mul_16_2 = (1, 1, 1)
+
+input_8_x0 = np.arange(2).reshape((2, 1, 1)).astype(np.uint8)
+mul_8_0 = (8, 1, 1)
+input_8_x1 = np.arange(32).reshape((2, 4, 4)).astype(np.int8)
+mul_8_1 = (2, 2, 2)
+input_8_x2 = np.arange(1).reshape((1, 1, 1)).astype(np.uint8)
+mul_8_2 = (1, 1, 1)
+
 
 class Net(Cell):
     def __init__(self):
@@ -54,6 +75,46 @@ class Net(Cell):
         return output
 
 
+class Net32(Cell):
+    def __init__(self):
+        super(Net32, self).__init__()
+        self.Tile = Tile()
+
+        self.input_32_x0 = Parameter(initializer(Tensor(input_32_x0), input_32_x0.shape), name='x0')
+        self.mul_32_0 = mul_32_0
+        self.input_32_x1 = Parameter(initializer(Tensor(input_32_x1), input_32_x1.shape), name='x1')
+        self.mul_32_1 = mul_32_1
+        self.input_32_x2 = Parameter(initializer(Tensor(input_32_x2), input_32_x2.shape), name='x2')
+        self.mul_32_2 = mul_32_2
+
+    @ms_function
+    def construct(self):
+        output = (self.Tile(self.input_32_x0, self.mul_32_0),
+                  self.Tile(self.input_32_x1, self.mul_32_1),
+                  self.Tile(self.input_32_x2, self.mul_32_2))
+        return output
+
+
+class Net16(Cell):
+    def __init__(self):
+        super(Net16, self).__init__()
+        self.Tile = Tile()
+
+        self.input_16_x0 = Parameter(initializer(Tensor(input_16_x0), input_16_x0.shape), name='x0')
+        self.mul_16_0 = mul_16_0
+        self.input_16_x1 = Parameter(initializer(Tensor(input_16_x1), input_16_x1.shape), name='x1')
+        self.mul_16_1 = mul_16_1
+        self.input_16_x2 = Parameter(initializer(Tensor(input_16_x2), input_16_x2.shape), name='x2')
+        self.mul_16_2 = mul_16_2
+
+    @ms_function
+    def construct(self):
+        output = (self.Tile(self.input_16_x0, self.mul_16_0),
+                  self.Tile(self.input_16_x1, self.mul_16_1),
+                  self.Tile(self.input_16_x2, self.mul_16_2))
+        return output
+
+
 @pytest.mark.level0
 @pytest.mark.platform_x86_gpu_training
 @pytest.mark.env_onecard
@@ -78,3 +139,55 @@ def test_tile():
     error2 = np.ones(shape=expect2.shape) * 1.0e-5
     assert np.all(diff2 < error2)
     assert output[2].shape == expect2.shape
+
+
+@pytest.mark.level0
+@pytest.mark.platform_x86_gpu_training
+@pytest.mark.env_onecard
+def test_tile_32():
+    net = Net32()
+    output = net()
+
+    expect0 = np.tile(input_32_x0, mul_32_0)
+    diff0 = output[0].asnumpy() - expect0
+    error0 = np.ones(shape=expect0.shape) * 1.0e-5
+    assert np.all(diff0 < error0)
+    assert output[0].shape == expect0.shape
+
+    expect1 = np.tile(input_32_x1, mul_32_1)
+    diff1 = output[1].asnumpy() - expect1
+    error1 = np.ones(shape=expect1.shape) * 1.0e-5
+    assert np.all(diff1 < error1)
+    assert output[1].shape == expect1.shape
+
+    expect2 = np.tile(input_32_x2, mul_32_2)
+    diff2 = output[2].asnumpy() - expect2
+    error2 = np.ones(shape=expect2.shape) * 1.0e-5
+    assert np.all(diff2 < error2)
+    assert output[2].shape == expect2.shape
+
+
+@pytest.mark.level0
+@pytest.mark.platform_x86_gpu_training
+@pytest.mark.env_onecard
+def test_tile_16():
+    net = Net16()
+    output = net()
+
+    expect0 = np.tile(input_16_x0, mul_16_0)
+    diff0 = output[0].asnumpy() - expect0
+    error0 = np.ones(shape=expect0.shape) * 1.0e-5
+    assert np.all(diff0 < error0)
+    assert output[0].shape == expect0.shape
+
+    expect1 = np.tile(input_16_x1, mul_16_1)
+    diff1 = output[1].asnumpy() - expect1
+    error1 = np.ones(shape=expect1.shape) * 1.0e-5
+    assert np.all(diff1 < error1)
+    assert output[1].shape == expect1.shape
+
+    expect2 = np.tile(input_16_x2, mul_16_2)
+    diff2 = output[2].asnumpy() - expect2
+    error2 = np.ones(shape=expect2.shape) * 1.0e-5
+    assert np.all(diff2 < error2)
+    assert output[2].shape == expect2.shape