Add Asypadding for conv fusion. (#21041)

* Add Asypadding for conv fusion.

test=develop

reference: pr/20042

* Fix eigen build link error

* Change back file mode

* Use math function & add more checks.

paddle/fluid/operators/fused/conv_fusion_op.cc

@@ -73,15 +73,85 @@ class Conv2DFusionOpInferShape : public framework::InferShapeBase {
     std::vector<int> paddings = ctx->Attrs().Get<std::vector<int>>("paddings");
     std::vector<int> dilations =
         ctx->Attrs().Get<std::vector<int>>("dilations");
+    std::string padding_algorithm =
+        ctx->Attrs().Get<std::string>("padding_algorithm");
+    int groups = ctx->Attrs().Get<int>("groups");
 
-    std::vector<int64_t> oshape({in_dims[0], filter_dims[0]});
-    for (size_t i = 0; i < strides.size(); ++i) {
-      oshape.push_back(ConvOutputSize(in_dims[i + 2], filter_dims[i + 2],
-                                      dilations[i], paddings[i], strides[i]));
+    framework::DDim in_data_dims;
+    in_data_dims = framework::slice_ddim(in_dims, 2, in_dims.size());
+
+    PADDLE_ENFORCE_EQ(
+        in_dims.size() == 4 || in_dims.size() == 5, true,
+        "ShapeError: Conv_fusion input should be 4-D or 5-D tensor. But "
+        "received: %u-D Tensor,"
+        "the shape of Conv_fusion input is [%s]",
+        in_dims.size(), in_dims);
+
+    PADDLE_ENFORCE_EQ(in_dims.size(), filter_dims.size(),
+                      "ShapeError: Conv_fusion input dimension and filter "
+                      "dimension should be the "
+                      "equal."
+                      "But received: the shape of Conv_fusion input is [%s], "
+                      "input dimension of Conv_fusion "
+                      "input is [%d],"
+                      "the shape of filter is [%s],  the filter dimension of "
+                      "Conv_fusion is [%d]",
+                      in_dims, in_dims.size(), filter_dims, filter_dims.size());
+
+    int in_sub_stride_size = in_dims.size() - strides.size();
+    PADDLE_ENFORCE_EQ(
+        in_dims.size() - strides.size() == 2U, true,
+        "ShapeError: the dimension of input minus the dimension of "
+        "stride must be euqal to 2."
+        "But received: the dimension of input minus the dimension "
+        "of stride is [%d], the"
+        "input dimension of Conv_fusion is [%d], the shape of Conv_fusion "
+        "input "
+        "is [%s], the stride"
+        "dimension of Conv_fusion is [%d]",
+        in_sub_stride_size, in_dims.size(), in_dims, strides.size());
+
+    const auto input_channels = in_dims[1];
+
+    PADDLE_ENFORCE_EQ(
+        input_channels, filter_dims[1] * groups,
+        "ShapeError: The number of input channels should be equal to filter "
+        "channels * groups. But received: the input channels is [%d], the shape"
+        "of input is [%s], the filter channel is [%d], the shape of filter is "
+        "[%s],"
+        "the groups is [%d]",
+        in_dims[1], in_dims, filter_dims[1], filter_dims, groups);
+    PADDLE_ENFORCE_EQ(
+        filter_dims[0] % groups, 0,
+        "ShapeError: The number of output channels should be divided by groups."
+        "But received: the output channels is [%d], the shape of filter is [%s]"
+        "(the first dimension of filter is output channel), the groups is [%d]",
+        filter_dims[0], filter_dims, groups);
+
+    framework::DDim filter_data_dims =
+        framework::slice_ddim(filter_dims, 2, filter_dims.size());
+    std::vector<int> ksize = framework::vectorize<int>(filter_data_dims);
+    UpdatePaddingAndDilation(&paddings, &dilations, padding_algorithm,
+                             in_data_dims, strides, ksize);
+
+    std::vector<int64_t> output_shape({in_dims[0]});
+    output_shape.push_back(filter_dims[0]);
+
+    for (size_t i = 0; i < in_data_dims.size(); ++i) {
+      if ((!ctx->IsRuntime()) &&
+          (in_data_dims[i] <= 0 || filter_dims[i + 2] <= 0)) {
+        output_shape.push_back(-1);
+      } else {
+        output_shape.push_back(
+            ConvOutputSize(in_data_dims[i], filter_dims[i + 2], dilations[i],
+                           paddings[2 * i], paddings[2 * i + 1], strides[i]));
+      }
     }
+
     PADDLE_ENFORCE_EQ(ctx->HasOutput("Output"), true,
                       "Output(Output) of ConvOp should not be null.");
-    ctx->SetOutputDim("Output", framework::make_ddim(oshape));
+    ctx->SetOutputDim("Output", framework::make_ddim(output_shape));
+
     std::vector<int> channels =
         ctx->Attrs().Get<std::vector<int>>("split_channels");
     if (channels.size()) {
@@ -90,7 +160,8 @@ class Conv2DFusionOpInferShape : public framework::InferShapeBase {
       std::vector<framework::DDim> oshapes;
       oshapes.reserve(channels.size());
       for (size_t i = 0; i < channels.size(); ++i) {
-        oshapes.push_back({oshape[0], channels[i], oshape[2], oshape[3]});
+        oshapes.push_back(
+            {output_shape[0], channels[i], output_shape[2], output_shape[3]});
       }
       ctx->SetOutputsDim("Outputs", oshapes);
     }

paddle/fluid/operators/fused/conv_fusion_op.cu.cc → paddle/fluid/operators/fused/conv_fusion_op.cu

@@ -13,7 +13,10 @@ See the License for the specific language governing permissions and
 limitations under the License. */
 
 #include "paddle/fluid/framework/op_registry.h"
+#include "paddle/fluid/operators/conv_cudnn_helper.h"
 #include "paddle/fluid/operators/conv_cudnn_op_cache.h"
+#include "paddle/fluid/operators/conv_op.h"
+#include "paddle/fluid/operators/math/padding.h"
 #include "paddle/fluid/platform/cudnn_helper.h"
 
 DECLARE_int64(cudnn_exhaustive_search_times);
@@ -44,6 +47,7 @@ class CUDNNConvFusionOpKernel : public framework::OpKernel<T> {
     PADDLE_ENFORCE_NOT_NULL(bias, "The bias should not be null.");
     auto* residual = ctx.Input<Tensor>("ResidualData");
     auto* output = ctx.Output<Tensor>("Output");
+    output->mutable_data<T>(ctx.GetPlace());
 
     std::vector<int> strides = ctx.Attr<std::vector<int>>("strides");
     std::vector<int> paddings = ctx.Attr<std::vector<int>>("paddings");
@@ -55,11 +59,96 @@ class CUDNNConvFusionOpKernel : public framework::OpKernel<T> {
     bool exhaustive_search =
         FLAGS_cudnn_exhaustive_search || ctx.Attr<bool>("exhaustive_search");
 
-    const T* input_data = input->data<T>();
+    // const T* input_data = input->data<T>();
     const T* filter_data = filter->data<T>();
     const T* bias_data = bias->data<T>();
-    T* output_data = output->mutable_data<T>(ctx.GetPlace());
+    // T* output_data = output->mutable_data<T>(ctx.GetPlace());
+
+    const std::string padding_algorithm =
+        ctx.Attr<std::string>("padding_algorithm");
+    const std::string data_format = ctx.Attr<std::string>("data_format");
+
+    Tensor transformed_input_channel(input->type());
+    Tensor transformed_output(output->type());
+    T* output_data = nullptr;
+
+    transformed_input_channel = *input;
+    transformed_output = *output;
+    output_data = transformed_output.data<T>();
     const T* residual_data = residual ? residual->data<T>() : output_data;
+    // update padding and dilation
+    auto in_dims = transformed_input_channel.dims();
+    auto filter_dims = filter->dims();
+    framework::DDim in_data_dims;
+    in_data_dims = framework::slice_ddim(in_dims, 2, in_dims.size());
+
+    framework::DDim filter_data_dims =
+        framework::slice_ddim(filter_dims, 2, filter_dims.size());
+    std::vector<int> ksize = framework::vectorize<int>(filter_data_dims);
+    UpdatePaddingAndDilation(&paddings, &dilations, padding_algorithm,
+                             in_data_dims, strides, ksize);
+
+    int data_dim = strides.size();  // 2d or 3d
+    bool is_sys_pad = math::IsSymmetricPadding(paddings, data_dim);
+
+    Tensor transformed_input;
+    std::vector<int> padding_common(data_dim, 0);
+    if (!is_sys_pad) {
+      std::vector<int> padding_diff(data_dim);
+      std::vector<int> new_input_shape_vec(data_dim + 2);
+      new_input_shape_vec[0] = transformed_input_channel.dims()[0];
+      new_input_shape_vec[1] = transformed_input_channel.dims()[1];
+
+      std::vector<int> input_pad(transformed_input_channel.dims().size() * 2,
+                                 0);
+      for (size_t i = 0; i < data_dim; ++i) {
+        padding_diff[i] = std::abs(paddings[2 * i] - paddings[2 * i + 1]);
+        padding_common[i] = std::min(paddings[2 * i], paddings[2 * i + 1]);
+        new_input_shape_vec[i + 2] =
+            transformed_input_channel.dims()[i + 2] + padding_diff[i];
+        input_pad[2 * i + 4] = paddings[2 * i] - padding_common[i];
+        input_pad[2 * i + 4 + 1] = paddings[2 * i + 1] - padding_common[i];
+      }
+      framework::DDim new_input_shape(
+          framework::make_ddim(new_input_shape_vec));
+      transformed_input.Resize(new_input_shape);
+      auto& dev_ctx =
+          ctx.template device_context<paddle::platform::CUDADeviceContext>();
+
+      transformed_input =
+          ctx.AllocateTmpTensor<T, paddle::platform::CUDADeviceContext>(
+              new_input_shape, dev_ctx);
+      const int rank = transformed_input_channel.dims().size();
+      T pad_value(0.0);
+      switch (rank) {
+        case 4: {
+          math::PadFunction<paddle::platform::CUDADeviceContext, T, 4>(
+              ctx, input_pad, transformed_input_channel, pad_value,
+              &transformed_input);
+        } break;
+        case 5: {
+          math::PadFunction<paddle::platform::CUDADeviceContext, T, 5>(
+              ctx, input_pad, transformed_input_channel, pad_value,
+              &transformed_input);
+        } break;
+        default:
+          PADDLE_THROW("ConvOp only support tensors with 4 or 5 dimensions.");
+      }
+
+    } else {
+      transformed_input = transformed_input_channel;
+      if (paddings.size() == data_dim) {
+        for (size_t i = 0; i < data_dim; ++i) {
+          padding_common[i] = paddings[i];
+        }
+      } else {
+        for (size_t i = 0; i < data_dim; ++i) {
+          padding_common[i] = paddings[2 * i];
+        }
+      }
+    }
+
+    const T* input_data = transformed_input.data<T>();
 
     // ------------------- cudnn descriptors ---------------------
     ScopedTensorDescriptor input_desc;
@@ -74,18 +163,19 @@ class CUDNNConvFusionOpKernel : public framework::OpKernel<T> {
     }
 
     cudnnConvolutionDescriptor_t cudnn_conv_desc =
-        conv_desc.descriptor<T>(paddings, strides, dilations);
+        conv_desc.descriptor<T>(padding_common, strides, dilations);
     CUDNN_ENFORCE(platform::dynload::cudnnSetConvolutionGroupCount(
         cudnn_conv_desc, groups));
 
     cudnnTensorDescriptor_t cudnn_input_desc = input_desc.descriptor<T>(
-        layout, framework::vectorize<int>(input->dims()));
+        layout, framework::vectorize<int>(transformed_input.dims()));
     cudnnTensorDescriptor_t cudnn_output_desc = output_desc.descriptor<T>(
-        layout, framework::vectorize<int>(output->dims()));
+        layout, framework::vectorize<int>(transformed_output.dims()));
     cudnnFilterDescriptor_t cudnn_filter_desc = filter_desc.descriptor<T>(
         layout, framework::vectorize<int>(filter->dims()));
     // Now only support NCHW
-    std::vector<int> bias_dim = {1, static_cast<int>(output->dims()[1]), 1, 1};
+    std::vector<int> bias_dim = {
+        1, static_cast<int>(transformed_output.dims()[1]), 1, 1};
     cudnnTensorDescriptor_t cudnn_bias_desc =
         bias_desc.descriptor<T>(layout, bias_dim);
     cudnnActivationDescriptor_t cudnn_act_desc =
@@ -109,7 +199,7 @@ class CUDNNConvFusionOpKernel : public framework::OpKernel<T> {
     CUDNN_ENFORCE(platform::dynload::cudnnSetConvolutionMathType(
         cudnn_conv_desc, CUDNN_DEFAULT_MATH));
 
-    auto x_dims = framework::vectorize(input->dims());
+    auto x_dims = framework::vectorize(transformed_input.dims());
     auto f_dims = framework::vectorize(filter->dims());
     if (!exhaustive_search) {
       CUDNN_ENFORCE(platform::dynload::cudnnGetConvolutionForwardAlgorithm(

python/paddle/fluid/tests/unittests/test_conv2d_fusion_op.py

@@ -23,23 +23,47 @@ from op_test import OpTest
 from test_conv2d_op import conv2d_forward_naive
 
 
+def create_test_padding_SAME_class(parent):
+    class TestPaddingSMAECase(parent):
+        def init_paddings(self):
+            self.pad = [0, 0]
+            self.padding_algorithm = "SAME"
+
+    cls_name = "{0}_{1}".format(parent.__name__, "PaddingSAMEOp")
+    TestPaddingSMAECase.__name__ = cls_name
+    globals()[cls_name] = TestPaddingSMAECase
+
+
+def create_test_padding_VALID_class(parent):
+    class TestPaddingVALIDCase(parent):
+        def init_paddings(self):
+            self.pad = [1, 1]
+            self.padding_algorithm = "VALID"
+
+    cls_name = "{0}_{1}".format(parent.__name__, "PaddingVALIDOp")
+    TestPaddingVALIDCase.__name__ = cls_name
+    globals()[cls_name] = TestPaddingVALIDCase
+
+
 class TestConv2dFusionOp(OpTest):
     def setUp(self):
         self.op_type = "conv2d_fusion"
         self.exhaustive_search = False
-        self.data_format = "AnyLayout"
+        self.data_format = "NCHW"
         self.dtype = np.float32
         self.activation = 'relu'
         self.add_bias = True
         self.add_residual_data = True
         self.channels = None
         self.outputs = None
+        self.padding_algorithm = "EXIPLICIT"
 
         self.init_group()
         self.init_dilation()
         self.init_test_case()
         self.init_bias_residual()
         self.init_activation()
+        self.init_paddings()
         self.set_search_method()
 
         conv2d_param = {
@@ -52,7 +76,9 @@ class TestConv2dFusionOp(OpTest):
         filter = np.random.random(self.filter_size).astype(self.dtype)
 
         self.output, _, _, _, _ = conv2d_forward_naive(
-            input, filter, self.groups, conv2d_param)
+            input, filter, self.groups, conv2d_param, self.padding_algorithm,
+            self.data_format)
+
         self.output = self.output.astype(self.dtype)
 
         self.inputs = {
@@ -84,7 +110,8 @@ class TestConv2dFusionOp(OpTest):
             'data_format': self.data_format,
             'exhaustive_search': self.exhaustive_search,
             'activation': self.activation,
-            'split_channels': self.channels
+            'split_channels': self.channels,
+            'padding_algorithm': self.padding_algorithm
         }
         self.outputs = {'Output': self.output}
 
@@ -127,6 +154,10 @@ class TestConv2dFusionOp(OpTest):
     def set_outputs(self):
         pass
 
+    def init_paddings(self):
+        self.pad = [0, 0]
+        self.padding_algorithm = "EXPLICIT"
+
 
 class TestWithoutResidual(TestConv2dFusionOp):
     def init_bias_residual(self):
@@ -186,5 +217,156 @@ class TestMultipleOutputs(TestConv2dFusionOp):
         self.outputs['Outputs'] = [('out1', out1), ('out2', out2)]
 
 
+class TestAsyPadding(TestConv2dFusionOp):
+    def init_paddings(self):
+        self.pad = [0, 0, 1, 2]
+        self.padding_algorithm = "EXPLICIT"
+
+
+class TestWithPad_AsyPadding(TestConv2dFusionOp):
+    def init_test_case(self):
+        self.stride = [1, 1]
+        self.input_size = [2, 3, 10, 10]  # NCHW
+        assert np.mod(self.input_size[1], self.groups) == 0
+        f_c = self.input_size[1] // self.groups
+        self.filter_size = [6, f_c, 3, 3]
+
+    def init_paddings(self):
+        self.pad = [2, 1, 3, 2]
+        self.padding_algorithm = "EXPLICIT"
+
+
+class TestWithStride_AsyPadding(TestConv2dFusionOp):
+    def init_test_case(self):
+        self.stride = [2, 2]
+        self.input_size = [2, 3, 6, 6]  # NCHW
+        assert np.mod(self.input_size[1], self.groups) == 0
+        f_c = self.input_size[1] // self.groups
+        self.filter_size = [6, f_c, 3, 3]
+
+    def init_paddings(self):
+        self.pad = [2, 1, 3, 2]
+        self.padding_algorithm = "EXPLICIT"
+
+
+class TestWith1x1_AsyPadding(TestConv2dFusionOp):
+    def init_test_case(self):
+        self.stride = [1, 1]
+        self.input_size = [2, 3, 5, 5]  # NCHW
+        assert np.mod(self.input_size[1], self.groups) == 0
+        f_c = self.input_size[1] // self.groups
+        self.filter_size = [6, f_c, 1, 1]
+
+    def init_group(self):
+        self.groups = 3
+
+    def init_paddings(self):
+        self.pad = [2, 2, 4, 0]
+        self.padding_algorithm = "EXPLICIT"
+
+
+class TestWithGroup_AsyPadding(TestConv2dFusionOp):
+    def init_group(self):
+        self.groups = 3
+
+
+class TestWithDepthWise3x3_AsyPadding(TestConv2dFusionOp):
+    def init_test_case(self):
+        self.stride = [1, 1]
+        self.input_size = [3, 4, 10, 10]  # NCHW
+        assert np.mod(self.input_size[1], self.groups) == 0
+        f_c = self.input_size[1] // self.groups
+        self.filter_size = [8, f_c, 3, 3]
+
+    def init_dilation(self):
+        self.dilations = [2, 2]
+
+    def init_group(self):
+        self.groups = 4
+
+    def init_paddings(self):
+        self.pad = [1, 3, 2, 1]
+        self.padding_algorithm = "EXPLICIT"
+
+
+class TestWithDepthWise5x5_AsyPadding(TestConv2dFusionOp):
+    def init_test_case(self):
+        self.stride = [1, 1]
+        self.input_size = [2, 4, 10, 10]  # NCHW
+        assert np.mod(self.input_size[1], self.groups) == 0
+        f_c = self.input_size[1] // self.groups
+        self.filter_size = [8, f_c, 5, 5]
+
+    def init_group(self):
+        self.groups = 4
+
+    def init_paddings(self):
+        self.pad = [0, 1, 1, 0]
+        self.padding_algorithm = "EXPLICIT"
+
+
+class TestWithDepthWise7x7_AsyPadding(TestConv2dFusionOp):
+    def init_test_case(self):
+        self.stride = [2, 2]
+        self.input_size = [2, 8, 10, 10]  # NCHW
+        assert np.mod(self.input_size[1], self.groups) == 0
+        f_c = self.input_size[1] // self.groups
+        self.filter_size = [16, f_c, 7, 7]
+
+    def init_group(self):
+        self.groups = 8
+
+    def init_paddings(self):
+        self.pad = [1, 3, 4, 1]
+        self.padding_algorithm = "EXPLICIT"
+
+
+class TestWithDilation_AsyPadding(TestConv2dFusionOp):
+    def init_test_case(self):
+        self.stride = [1, 1]
+        self.input_size = [2, 3, 10, 10]  # NCHW
+        assert np.mod(self.input_size[1], self.groups) == 0
+        f_c = self.input_size[1] // self.groups
+        self.filter_size = [6, f_c, 3, 3]
+
+    def init_dilation(self):
+        self.dilations = [2, 2]
+
+    def init_group(self):
+        self.groups = 3
+
+    def init_paddings(self):
+        self.pad = [0, 1, 3, 0]
+        self.padding_algorithm = "EXPLICIT"
+
+
+class TestWithInput1x1Filter1x1_AsyPadding(TestConv2dFusionOp):
+    def init_test_case(self):
+        self.stride = [1, 1]
+        self.input_size = [2, 3, 1, 1]  # NCHW
+        assert np.mod(self.input_size[1], self.groups) == 0
+        f_c = self.input_size[1] // self.groups
+        self.filter_size = [6, f_c, 1, 1]
+
+    def init_group(self):
+        self.groups = 3
+
+    def init_paddings(self):
+        self.pad = [0, 3, 4, 0]
+        self.padding_algorithm = "EXPLICIT"
+
+
+create_test_padding_SAME_class(TestAsyPadding)
+create_test_padding_SAME_class(TestWithPad_AsyPadding)
+create_test_padding_SAME_class(TestWithStride_AsyPadding)
+create_test_padding_SAME_class(TestWithGroup_AsyPadding)
+create_test_padding_SAME_class(TestWithInput1x1Filter1x1_AsyPadding)
+
+create_test_padding_VALID_class(TestAsyPadding)
+create_test_padding_VALID_class(TestWithPad_AsyPadding)
+create_test_padding_VALID_class(TestWithStride_AsyPadding)
+create_test_padding_VALID_class(TestWithGroup_AsyPadding)
+create_test_padding_VALID_class(TestWithInput1x1Filter1x1_AsyPadding)
+
 if __name__ == '__main__':
     unittest.main()