[CINN Support 0D-Tensor] CINN supports 0D-Tensor with trick temporari…

…ly (#53382)

* [CINN Support 0D-Tensor] CINN supports 0D-Tensor with trick temporarily

* Add unittest

paddle/fluid/framework/details/CMakeLists.txt

@@ -382,6 +382,7 @@ set(IR_PASS_DEPS
 
 if(WITH_CINN)
   set(IR_PASS_DEPS ${IR_PASS_DEPS} build_cinn_pass)
+  set(IR_PASS_DEPS ${IR_PASS_DEPS} cinn_zero_tensor_trick_pass)
 endif()
 
 if(NOT APPLE

paddle/fluid/framework/details/build_strategy.cc

@@ -57,6 +57,9 @@ class ParallelExecutorPassBuilder : public ir::PassBuilder {
 
 #ifdef PADDLE_WITH_CINN
     if (FLAGS_use_cinn || strategy.build_cinn_pass_) {
+      // Note: This is a trick to support 0D-Tensor for CINN. This pass will be
+      // removed in the near future.
+      AppendPass("cinn_zero_tensor_trick_pass");
       // Note: This pass is used to enable cinn.
       AppendPass("build_cinn_pass");
       AppendPrintGraphPass("graph_viz_pass", "_build_cinn_graph");
@@ -532,6 +535,7 @@ USE_PASS(fused_attention_pass);
 USE_PASS(fuse_adamw_op_pass);
 #endif
 #ifdef PADDLE_WITH_CINN
+USE_PASS(cinn_zero_tensor_trick_pass);
 USE_PASS(build_cinn_pass);
 #endif
 #ifdef PADDLE_WITH_CUDA

paddle/fluid/framework/paddle2cinn/CMakeLists.txt

@@ -8,6 +8,8 @@ pass_library(
   errors
   enforce)
 
+pass_library(cinn_zero_tensor_trick_pass base)
+
 cc_library(
   transform_desc
   SRCS transform_desc.cc
@@ -62,6 +64,20 @@ if(WITH_TESTING)
   set_tests_properties(build_cinn_pass_test PROPERTIES LABELS "RUN_TYPE=CINN")
   target_link_libraries(build_cinn_pass_test ${PYTHON_LIBRARIES})
 
+  cc_test_old(
+    cinn_zero_tensor_trick_pass_test
+    SRCS
+    cinn_zero_tensor_trick_pass_test.cc
+    DEPS
+    build_cinn_pass
+    cinn_compiler
+    op_registry
+    elementwise_add_op
+    generated_op)
+  set_tests_properties(cinn_zero_tensor_trick_pass_test
+                       PROPERTIES LABELS "RUN_TYPE=CINN")
+  target_link_libraries(cinn_zero_tensor_trick_pass_test ${PYTHON_LIBRARIES})
+
   cc_test_old(transform_desc_test SRCS transform_desc_test.cc DEPS
               transform_desc)
   set_tests_properties(transform_desc_test PROPERTIES LABELS "RUN_TYPE=CINN")

paddle/fluid/framework/paddle2cinn/cinn_zero_tensor_trick_pass.cc

@@ -0,0 +1,78 @@
+/* Copyright (c) 2023 PaddlePaddle Authors. All Rights Reserved.
+
+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. */
+
+#include "paddle/fluid/framework/paddle2cinn/cinn_zero_tensor_trick_pass.h"
+
+#include <string>
+#include "glog/logging.h"
+
+#include "paddle/fluid/framework/op_proto_maker.h"
+#include "paddle/fluid/framework/program_desc.h"
+
+namespace paddle {
+namespace framework {
+namespace paddle2cinn {
+
+void CinnZeroTensorTrickPass::ApplyImpl(ir::Graph* graph) const {
+  // fix shape attr of these ops
+  const std::unordered_set<std::string> op_cases_fix_attr{"fill_constant",
+                                                          "uniform_random",
+                                                          "expand_v2",
+                                                          "assign_value",
+                                                          "gaussian_random",
+                                                          "set_value"};
+  for (const ir::Node* n : graph->Nodes()) {
+    if (n->IsOp() && op_cases_fix_attr.count(n->Op()->Type())) {
+      if (n->Op()->HasAttr("shape")) {
+        auto attr_type = n->Op()->GetAttrType("shape");
+        if (attr_type == paddle::framework::proto::INTS) {
+          auto shapes =
+              PADDLE_GET_CONST(std::vector<int32_t>, n->Op()->GetAttr("shape"));
+          if (shapes.empty()) {
+            shapes.push_back(1);
+            n->Op()->SetAttr("shape", shapes);
+            VLOG(4) << "op " << n->Op()->Type()
+                    << " shape attribute dims is empty, fix dim -> {1} ";
+          }
+        } else { /* attr_type == paddle::framework::proto::LONGS */
+          auto shapes =
+              PADDLE_GET_CONST(std::vector<int64_t>, n->Op()->GetAttr("shape"));
+          if (shapes.empty()) {
+            shapes.push_back(1);
+            n->Op()->SetAttr("shape", shapes);
+            VLOG(4) << "op " << n->Op()->Type()
+                    << " shape attribute dims is empty, fix dim -> {1} ";
+          }
+        }
+      }
+    }
+    if (n->IsVar()) {
+      if (n->Var() && n->Var()->GetType() == proto::VarType::LOD_TENSOR) {
+        std::vector<int64_t> shape = n->Var()->GetShape();
+        if (shape.empty()) {
+          shape.push_back(1);
+          n->Var()->SetShape(shape);
+          VLOG(4) << "var " << n->Name() << " dims is empty, fix dim -> {1} ";
+        }
+      }
+    }
+  }
+}
+
+}  // namespace paddle2cinn
+}  // namespace framework
+}  // namespace paddle
+
+REGISTER_PASS(cinn_zero_tensor_trick_pass,
+              paddle::framework::paddle2cinn::CinnZeroTensorTrickPass);

paddle/fluid/framework/paddle2cinn/cinn_zero_tensor_trick_pass.h

@@ -0,0 +1,33 @@
+/* Copyright (c) 2023 PaddlePaddle Authors. All Rights Reserved.
+
+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. */
+
+#pragma once
+
+#include "paddle/fluid/framework/ir/graph.h"
+#include "paddle/fluid/framework/ir/pass.h"
+
+namespace paddle {
+namespace framework {
+namespace paddle2cinn {
+
+class Graph;
+
+class CinnZeroTensorTrickPass : public framework::ir::Pass {
+ protected:
+  void ApplyImpl(ir::Graph* graph) const override;
+};
+
+}  // namespace paddle2cinn
+}  // namespace framework
+}  // namespace paddle

paddle/fluid/framework/paddle2cinn/cinn_zero_tensor_trick_pass_test.cc

@@ -0,0 +1,56 @@
+/* Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
+
+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. */
+
+#include <gtest/gtest.h>
+
+#include "paddle/fluid/framework/paddle2cinn/cinn_zero_tensor_trick_pass.h"
+
+#include "paddle/fluid/framework/ir/pass_tester_helper.h"
+#include "paddle/fluid/platform/enforce.h"
+
+namespace paddle {
+namespace framework {
+namespace paddle2cinn {
+
+TEST(CinnZeroTensorTrickPass, basic) {
+  ir::Layers layers;
+  auto* x = layers.data("x", {});
+  auto* y = layers.data("y", {3, 4});
+  auto* add_out_0 = layers.elementwise_add(x, y, nullptr, 0);
+  std::unique_ptr<ir::Graph> graph(new ir::Graph(layers.main_program()));
+  auto pass = ir::PassRegistry::Instance().Get("cinn_zero_tensor_trick_pass");
+  VLOG(3) << DebugString(graph);
+
+  graph.reset(pass->Apply(graph.release()));
+  VLOG(3) << DebugString(graph);
+
+  for (auto* n : graph->Nodes()) {
+    if (n->IsVar()) {
+      if (n->Var() && n->Var()->GetType() == proto::VarType::LOD_TENSOR) {
+        std::vector<int64_t> shape = n->Var()->GetShape();
+        PADDLE_ENFORCE_EQ(
+            shape.empty(),
+            false,
+            platform::errors::PreconditionNotMet(
+                "The shape of elementwise_add should not be empty after fuse"));
+      }
+    }
+  }
+}
+
+}  // namespace paddle2cinn
+}  // namespace framework
+}  // namespace paddle
+
+USE_PASS(cinn_zero_tensor_trick_pass);