【new ir】modify builtin_split bug (#56463)

* [prim][newir] add basic framework for primitive

* support desctensor in new ir

* add vjp interface

* support vjp in new ir

* support vjp in new ir

* polish vjp interface

* fix stop_gradients set

* fix vjp dispatch

* add comment

* add vjp test for new ir

* add test for tanh vjp

* [prim][newir] add basic framework for primitive

* support desctensor in new ir

* support vjp in new ir

* support vjp in new ir

* polish vjp interface

* fix stop_gradients set

* fix vjp dispatch

* add comment

* add vjp test for new ir

* add test for tanh vjp

* add eager and static backend for warp lower level api

* support call_vjp pybind

* polish code and add test for vjp

* remove useless code

* polish code

* remove useless code

* support mean vjp

* backward origin code

* add test for mean vjp and support has_vjp function

* fix call_vjp

* polish code

* add attrs and dtype interface

* add primitive ops set for backend

* fix compile bugs

* fix some bugs

* fix windows bugs

* add vjp test for tanh_

* fix inference CI

* fix inference ci

* modify fluid cmake

* origin test of tanh and mean passed

* fix conflict

* modify stop_gradient

* remove useless deps

* add cmake

* modify block.ops

* modify test

* fix conflict

* reply review comments

* reply review comments

* pulish code

* fix comment

* fix test

* polish code

* modify backward stop_gradients

* modify static_backend.cc

* refactor grad_op

* support add and add_inplace vjp

* remove useless code

* remove useless code

* remove cout

* modify add_n

* modify add_n with add_vjp test

* modify add_n with add_vjp test

* fix conflict and concat call_vjp

* modify backward test

* Add more gen api

* modify split kernel pass

* modify concat api

* modify builtin split bug

* delete vlog

---------
Co-authored-by: Ncxxly <chenxx_id@163.com>
Co-authored-by: NCharles-hit <wanghao107@baidu.com>
Co-authored-by: Nzhangbo9674 <zhangbo54@baidu.com>
Co-authored-by: NYuanRisheng <yuanrisheng@baidu.com>
Co-authored-by: N0x45f <wangzhen45@baidu.com>

paddle/fluid/ir/transforms/pd_op_to_kernel_pass.cc

@@ -592,7 +592,7 @@ std::unique_ptr<ir::Program> PdOpLowerToKernelPass(ir::Program* prog,
     }
 
     if (op_item->name() == "builtin.split") {
-      phi::Place out_place = place;
+      std::vector<phi::Place> out_places(op_item->num_results());
       // Copy op inputs
       std::vector<ir::OpResult> vec_inputs;
       if (op_item->num_operands() > 0) {
@@ -613,10 +613,12 @@ std::unique_ptr<ir::Program> PdOpLowerToKernelPass(ir::Program* prog,
 
           if (new_in.type().isa<ir::VectorType>()) {
             auto vec_types = new_in.type().dyn_cast<ir::VectorType>().data();
-            out_place =
-                vec_types[0]
-                    .dyn_cast<paddle::dialect::AllocatedDenseTensorType>()
-                    .place();
+            for (uint64_t idx = 0; idx < vec_types.size(); idx++) {
+              out_places[idx] =
+                  vec_types[idx]
+                      .dyn_cast<paddle::dialect::AllocatedDenseTensorType>()
+                      .place();
+            }
           } else {
             PADDLE_THROW(
                 phi::errors::Unimplemented("only support vector type for now"));
@@ -634,7 +636,7 @@ std::unique_ptr<ir::Program> PdOpLowerToKernelPass(ir::Program* prog,
             auto allocated_dense_tensor_dtype =
                 paddle::dialect::AllocatedDenseTensorType::get(
                     ctx,
-                    out_place,
+                    out_places[i],
                     result_type.dyn_cast<dialect::DenseTensorType>());
             op_output_types.push_back(allocated_dense_tensor_dtype);
           } else {

paddle/ir/core/builtin_op.cc

@@ -162,6 +162,17 @@ void CombineOp::Verify() const {
 }
 
 const char *SliceOp::attributes_name[attributes_num] = {"index"};  // NOLINT
+
+void SliceOp::Build(Builder &builder,
+                    OperationArgument &argument,
+                    const ir::OpResult &input,
+                    int index) {
+  argument.inputs = {input};
+  argument.output_types.emplace_back(input.type()
+                                         .dyn_cast<ir::VectorType>()
+                                         .data()[static_cast<size_t>(index)]);
+}
+
 void SliceOp::Verify() const {
   // inputs.size() == 1
   auto input_size = num_operands();
@@ -207,6 +218,17 @@ void SliceOp::Verify() const {
       output_type);
 }
 
+void SplitOp::Build(Builder &builder,
+                    OperationArgument &argument,
+                    const ir::OpResult &input) {
+  argument.inputs = {input};
+  for (size_t idx = 0; idx < input.type().dyn_cast<ir::VectorType>().size();
+       ++idx) {
+    argument.output_types.emplace_back(
+        input.type().dyn_cast<ir::VectorType>().data()[idx]);
+  }
+}
+
 void SplitOp::Verify() const {
   // inputs.size() == 1
   IR_ENFORCE(num_operands() == 1u, "The size of inputs must be equal to 1.");
@@ -235,18 +257,6 @@ void SplitOp::Verify() const {
   }
 }
 
-void SplitOp::Build(Builder &builder,
-                    OperationArgument &argument,
-                    const ir::OpResult &input) {
-  argument.inputs = {input};
-  std::vector<ir::Type> outputs_types;
-  for (size_t idx = 0; idx < input.type().dyn_cast<ir::VectorType>().size();
-       ++idx) {
-    argument.output_types.emplace_back(
-        input.type().dyn_cast<ir::VectorType>()[idx]);
-  }
-}
-
 const char *ConstantOp::attributes_name[attributes_num] = {"value"};  // NOLINT
 
 void ConstantOp::Build(Builder &builder,

paddle/ir/core/builtin_op.h

@@ -118,7 +118,8 @@ class IR_API SliceOp : public ir::Op<SliceOp> {
 
   static void Build(Builder &builder,             // NOLINT
                     OperationArgument &argument,  // NOLINT
-                    const ir::OpResult &input);
+                    const ir::OpResult &input,
+                    int index);
 
   void Verify() const;
   ir::Value input() { return operand_source(0); }

python/paddle/tensor/manipulation.py

@@ -1121,7 +1121,7 @@ def concat(x, axis=0, name=None):
         return _C_ops.concat(input, axis)
     else:
         if paddle.ir.core._use_new_ir_api():
-            if not isinstance(input, Variable):
+            if not isinstance(input, paddle.ir.Value):
                 input = [t for t in input if t.shape.count(0) == 0]
             return paddle._ir_ops.concat(input, axis)
         check_type(input, 'input', (list, tuple, Variable), 'concat')