Support setting infershape function for custom grad op (PaddlePaddle#…

…38776) * unify infer_shape func calling * support set grad infer shape fn for custom op * unify infershape in new executor and eager * remove todo comment * revert infershape in operator
b3602sss · Jan 10, 2022 · 046553c · 046553c
1 parent cd2855b
commit 046553c
Show file tree

Hide file tree

Showing 10 changed files with 236 additions and 160 deletions.
diff --git a/paddle/fluid/eager/legacy/prepared_operator.cc b/paddle/fluid/eager/legacy/prepared_operator.cc
@@ -174,8 +174,7 @@ static void PreparedOpRunImpl(
 
   EagerInferShapeContext infer_shape_ctx(&ins, &outs, &attrs, &default_attrs,
                                          op.Type());
-  static_cast<const paddle::framework::OperatorWithKernel&>(op).InferShape(
-      &infer_shape_ctx);
+  op.Info().infer_shape_(&infer_shape_ctx);
 
   func(EagerExecutionContext(op, scope, *dev_ctx, ctx, ins, outs, attrs,
                              default_attrs));

diff --git a/paddle/fluid/framework/custom_operator.cc b/paddle/fluid/framework/custom_operator.cc
diff --git a/paddle/fluid/framework/new_executor/data_transfer.cc b/paddle/fluid/framework/new_executor/data_transfer.cc
@@ -94,8 +94,7 @@ void DataTranferHelper::RunAndConstructOpFuncNode(
 
   // 2. Execute infer shape and choose kernel
   auto& all_op_kernels = OperatorWithKernel::AllOpKernels();
-  static_cast<const framework::OperatorWithKernel*>(op.get())->InferShape(
-      &infer_shape_ctx);
+  op.get()->Info().infer_shape_(&infer_shape_ctx);
   auto kernels_iter = all_op_kernels.find(op_type);
   PADDLE_ENFORCE_NE(kernels_iter, all_op_kernels.end(),
                     platform::errors::Unavailable(

diff --git a/paddle/fluid/framework/new_executor/interpretercore_util.cc b/paddle/fluid/framework/new_executor/interpretercore_util.cc
@@ -355,7 +355,7 @@ void build_op_func_list(const platform::Place& place,
         // TODO(Aurelius84): In case of control flow ops, they are NOT
         // inheritted
         // from OperatorWithKernel.
-        op_with_kernel->InferShape(&infer_shape_ctx);
+        op_with_kernel->Info().infer_shape_(&infer_shape_ctx);
       }
 
       auto kernels_iter = all_op_kernels.find(op->Type());

diff --git a/paddle/fluid/framework/operator.cc b/paddle/fluid/framework/operator.cc
@@ -1090,7 +1090,7 @@ void OperatorWithKernel::RuntimeInferShape(const Scope& scope,
                                            const platform::Place& place,
                                            const RuntimeContext& ctx) const {
   RuntimeInferShapeContext infer_shape_ctx(*this, ctx);
-  this->InferShape(&infer_shape_ctx);
+  this->Info().infer_shape_(&infer_shape_ctx);
 }
 
 void OperatorWithKernel::RunImpl(const Scope& scope,
@@ -1178,6 +1178,8 @@ void OperatorWithKernel::RunImpl(const Scope& scope,
     platform::RecordEvent record_event("infer_shape",
                                        platform::EventRole::kInnerOp);
     RuntimeInferShapeContext infer_shape_ctx(*this, *runtime_ctx);
+    // TODO(chenweihang): replace this after removing `this->IsMKLDNNType()`
+    // in some mkldnn infershape functions, such conv2d infershape
     this->InferShape(&infer_shape_ctx);
   }
 

diff --git a/paddle/fluid/imperative/prepared_operator.cc b/paddle/fluid/imperative/prepared_operator.cc
@@ -491,8 +491,7 @@ static void PreparedOpRunImpl(
 
   DygraphInferShapeContext<VarType> infer_shape_ctx(&ins, &outs, &attrs,
                                                     &default_attrs, op.Type());
-  static_cast<const framework::OperatorWithKernel&>(op).InferShape(
-      &infer_shape_ctx);
+  op.Info().infer_shape_(&infer_shape_ctx);
 
   func(DygraphExecutionContext<VarType>(op, scope, *dev_ctx, ctx, ins, outs,
                                         attrs, default_attrs));
@@ -537,8 +536,7 @@ static void PreparedOpRunPtImpl(
     const framework::AttributeMap& default_attrs) {
   DygraphInferShapeContext<VarType> infer_shape_ctx(&ins, &outs, &attrs,
                                                     &default_attrs, op.Type());
-  static_cast<const framework::OperatorWithKernel&>(op).InferShape(
-      &infer_shape_ctx);
+  op.Info().infer_shape_(&infer_shape_ctx);
 
   BuildDygraphPtenKernelContext<VarType>(pt_kernel_signature, pt_kernel, ins,
                                          outs, attrs, default_attrs, dev_ctx,

diff --git a/paddle/pten/api/lib/op_meta_info.cc b/paddle/pten/api/lib/op_meta_info.cc
@@ -122,13 +122,6 @@ OpMetaInfoBuilder& OpMetaInfoBuilder::SetKernelFn(KernelFunc func) {
 }
 
 OpMetaInfoBuilder& OpMetaInfoBuilder::SetInferShapeFn(InferShapeFunc func) {
-  PADDLE_ENFORCE_EQ(
-      index_,
-      0UL,
-      platform::errors::Unimplemented(
-          "Currently, the InferShapeFn setting of Grad Op is not supported, "
-          "And backward Tensor `X@GRAD` will use the shape of forward Tensor "
-          "`X` by default."));
   info_ptr_->SetInferShapeFn(std::forward<InferShapeFunc>(func));
   return *this;
 }

diff --git a/python/paddle/fluid/tests/custom_op/custom_relu_op.cc b/python/paddle/fluid/tests/custom_op/custom_relu_op.cc
@@ -105,3 +105,49 @@ PD_BUILD_GRAD_OP(custom_relu)
     .Inputs({"X", "Out", paddle::Grad("Out")})
     .Outputs({paddle::Grad("X")})
     .SetKernelFn(PD_KERNEL(ReluBackward));
+
+std::vector<paddle::Tensor> relu_cpu_backward_without_x(
+    const paddle::Tensor& out, const paddle::Tensor& grad_out) {
+  auto grad_x = paddle::Tensor(paddle::PlaceType::kCPU, out.shape());
+
+  PD_DISPATCH_FLOATING_TYPES(out.type(), "relu_cpu_backward", ([&] {
+                               relu_cpu_backward_kernel<data_t>(
+                                   grad_out.data<data_t>(),
+                                   out.data<data_t>(),
+                                   grad_x.mutable_data<data_t>(out.place()),
+                                   out.size());
+                             }));
+
+  return {grad_x};
+}
+
+std::vector<paddle::Tensor> relu_cuda_backward_without_x(
+    const paddle::Tensor& out, const paddle::Tensor& grad_out);
+
+std::vector<paddle::Tensor> ReluBackwardWithoutX(
+    const paddle::Tensor& out, const paddle::Tensor& grad_out) {
+  if (out.place() == paddle::PlaceType::kCPU) {
+    return relu_cpu_backward_without_x(out, grad_out);
+  } else if (out.place() == paddle::PlaceType::kGPU) {
+    return relu_cuda_backward_without_x(out, grad_out);
+  } else {
+    PD_THROW("Not implemented.");
+  }
+}
+
+std::vector<std::vector<int64_t>> ReluBackwardWithoutXInferShape(
+    const std::vector<int64_t>& out_shape,
+    const std::vector<int64_t>& grad_out_shape) {
+  return {out_shape};
+}
+
+PD_BUILD_OP(custom_relu_no_x_in_backward)
+    .Inputs({"X"})
+    .Outputs({"Out"})
+    .SetKernelFn(PD_KERNEL(ReluForward));
+
+PD_BUILD_GRAD_OP(custom_relu_no_x_in_backward)
+    .Inputs({"Out", paddle::Grad("Out")})
+    .Outputs({paddle::Grad("X")})
+    .SetKernelFn(PD_KERNEL(ReluBackwardWithoutX))
+    .SetInferShapeFn(PD_INFER_SHAPE(ReluBackwardWithoutXInferShape));
diff --git a/python/paddle/fluid/tests/custom_op/custom_relu_op.cu b/python/paddle/fluid/tests/custom_op/custom_relu_op.cu
@@ -70,3 +70,22 @@ std::vector<paddle::Tensor> relu_cuda_backward(const paddle::Tensor& x,
 
   return {grad_x};
 }
+
+std::vector<paddle::Tensor> relu_cuda_backward_without_x(
+    const paddle::Tensor& out, const paddle::Tensor& grad_out) {
+  auto grad_x = paddle::Tensor(paddle::PlaceType::kGPU, out.shape());
+
+  int numel = out.size();
+  int block = 512;
+  int grid = (numel + block - 1) / block;
+  PD_DISPATCH_FLOATING_AND_HALF_TYPES(
+      out.type(), "relu_cuda_backward_kernel", ([&] {
+        relu_cuda_backward_kernel<data_t><<<grid, block, 0, out.stream()>>>(
+            grad_out.data<data_t>(),
+            out.data<data_t>(),
+            grad_x.mutable_data<data_t>(out.place()),
+            numel);
+      }));
+
+  return {grad_x};
+}
diff --git a/python/paddle/fluid/tests/custom_op/test_custom_relu_op_jit.py b/python/paddle/fluid/tests/custom_op/test_custom_relu_op_jit.py
@@ -49,7 +49,8 @@
 class TestJITLoad(unittest.TestCase):
     def setUp(self):
         self.custom_ops = [
-            custom_module.custom_relu, custom_module.custom_relu_dup
+            custom_module.custom_relu, custom_module.custom_relu_dup,
+            custom_module.custom_relu_no_x_in_backward
         ]
         self.dtypes = ['float32', 'float64']
         if paddle.is_compiled_with_cuda():