[phi decoupling] move platform/transform to phi (#50498)

* move platform::transform to phi

* fix bugs

* move transform_test to phi

* fix cmake

* update namespace

* fix cmake

cmake/inference_lib.cmake

@@ -448,10 +448,8 @@ add_dependencies(fluid_lib_dist ${platform_lib_deps})
 copy(
   fluid_lib_dist
   SRCS ${src_dir}/${module}/*.h ${src_dir}/${module}/dynload/*.h
-       ${src_dir}/${module}/details/*.h
        ${PADDLE_BINARY_DIR}/paddle/phi/api/profiler/*.pb.h
-  DSTS ${dst_dir}/${module} ${dst_dir}/${module}/dynload
-       ${dst_dir}/${module}/details ${dst_dir}/${module})
+  DSTS ${dst_dir}/${module} ${dst_dir}/${module}/dynload ${dst_dir}/${module})
 
 set(module "string")
 copy(

paddle/fluid/framework/data_transform.h

@@ -24,7 +24,7 @@ limitations under the License. */
 #include "paddle/fluid/framework/variable.h"
 #include "paddle/fluid/platform/device_context.h"
 #include "paddle/fluid/platform/macros.h"
-#include "paddle/fluid/platform/transform.h"
+#include "paddle/phi/common/transform.h"
 #include "paddle/phi/kernels/funcs/math_function.h"
 
 namespace paddle {

paddle/fluid/framework/data_type_transform.cc

@@ -16,7 +16,7 @@ limitations under the License. */
 
 #include "paddle/fluid/framework/convert_utils.h"
 #include "paddle/fluid/framework/selected_rows_utils.h"
-#include "paddle/fluid/platform/transform.h"
+#include "paddle/phi/common/transform.h"
 
 #if defined(PADDLE_WITH_XPU)
 #include "paddle/fluid/platform/device/device_wrapper.h"
@@ -94,7 +94,7 @@ struct CastDataType {
     auto* out_begin = out_->mutable_data<OutType>(in_.place());
 
     if (platform::is_cpu_place(in_.place())) {
-      platform::Transform<phi::CPUContext> trans;
+      phi::Transform<phi::CPUContext> trans;
       auto* context = static_cast<const phi::CPUContext*>(ctx_);
       trans(*context,
             in_begin,
@@ -103,7 +103,7 @@ struct CastDataType {
             CastDataTypeFunctor<InType, OutType>());
 #if defined(__NVCC__) || defined(__HIPCC__)
     } else if (platform::is_gpu_place(in_.place())) {
-      platform::Transform<phi::GPUContext> trans;
+      phi::Transform<phi::GPUContext> trans;
       auto* context = static_cast<const phi::GPUContext*>(ctx_);
       trans(*context,
             in_begin,
@@ -114,7 +114,7 @@ struct CastDataType {
 #endif
 #if defined(PADDLE_WITH_IPU)
     } else if (platform::is_ipu_place(in_.place())) {
-      platform::Transform<phi::CPUContext> trans;
+      phi::Transform<phi::CPUContext> trans;
       auto* context = static_cast<const phi::CPUContext*>(ctx_);
       trans(*context,
             in_begin,

paddle/fluid/operators/cast_op.h

@@ -17,7 +17,7 @@ limitations under the License. */
 #include "paddle/fluid/framework/data_type.h"
 #include "paddle/fluid/framework/op_registry.h"
 #include "paddle/fluid/framework/phi_utils.h"
-#include "paddle/fluid/platform/transform.h"
+#include "paddle/phi/common/transform.h"
 #include "paddle/phi/kernels/cast_kernel.h"
 
 namespace paddle {
@@ -44,7 +44,7 @@ struct CastOpFunctor {
     auto numel = in_->numel();
     auto* in_end = in_begin + numel;
     auto* out_begin = out_->mutable_data<OutT>(ctx_.GetPlace());
-    platform::Transform<DeviceContext> trans;
+    phi::Transform<DeviceContext> trans;
     trans(
         ctx_, in_begin, in_end, out_begin, CastOpTransformFunctor<InT, OutT>());
   }

paddle/fluid/operators/center_loss_op.h

@@ -20,7 +20,7 @@ limitations under the License. */
 
 #include "paddle/fluid/framework/eigen.h"
 #include "paddle/fluid/framework/op_registry.h"
-#include "paddle/fluid/platform/transform.h"
+#include "paddle/phi/common/transform.h"
 #include "paddle/phi/kernels/funcs/blas/blas.h"
 
 namespace paddle {
@@ -94,7 +94,7 @@ class CenterLossKernel : public framework::OpKernel<T> {
     T *center_out_index;
     T *center_loss_diff_index;
     T *acc_index;
-    platform::Transform<DeviceContext> trans;
+    phi::Transform<DeviceContext> trans;
 
     for (int i = 0; i < batch_size; ++i) {
       tLabel = label_data[i];

paddle/fluid/operators/clip_by_norm_op.h

@@ -17,7 +17,7 @@ limitations under the License. */
 #include "paddle/fluid/framework/eigen.h"
 #include "paddle/fluid/framework/op_registry.h"
 #include "paddle/fluid/framework/selected_rows_utils.h"
-#include "paddle/fluid/platform/transform.h"
+#include "paddle/phi/common/transform.h"
 #include "paddle/phi/kernels/funcs/selected_rows_functor.h"
 
 namespace paddle {

paddle/fluid/operators/detection/anchor_generator_op.h

@@ -17,7 +17,7 @@ limitations under the License. */
 #include <vector>
 
 #include "paddle/fluid/framework/op_registry.h"
-#include "paddle/fluid/platform/transform.h"
+#include "paddle/phi/common/transform.h"
 #include "paddle/phi/kernels/funcs/math_function.h"
 
 namespace paddle {

paddle/fluid/operators/detection/prior_box_op.h

@@ -17,7 +17,7 @@ limitations under the License. */
 #include <vector>
 
 #include "paddle/fluid/framework/op_registry.h"
-#include "paddle/fluid/platform/transform.h"
+#include "paddle/phi/common/transform.h"
 #include "paddle/phi/core/visit_type.h"
 #include "paddle/phi/kernels/funcs/math_function.h"
 

paddle/fluid/operators/elementwise/elementwise_op_function.h

@@ -28,7 +28,8 @@ limitations under the License. */
 #include "paddle/fluid/memory/malloc.h"
 #include "paddle/fluid/operators/elementwise/elementwise_functor.h"
 #include "paddle/fluid/platform/device/gpu/gpu_info.h"
-#include "paddle/fluid/platform/transform.h"
+#include "paddle/phi/api/lib/utils/tensor_utils.h"
+#include "paddle/phi/common/transform.h"
 #include "paddle/phi/kernels/cpu/elementwise.h"
 #include "paddle/phi/kernels/cpu/elementwise_grad.h"
 

paddle/fluid/operators/fake_quantize_op.cc

@@ -19,7 +19,7 @@ limitations under the License. */
 
 #include "paddle/fluid/framework/eigen.h"
 #include "paddle/fluid/framework/op_version_registry.h"
-#include "paddle/fluid/platform/transform.h"
+#include "paddle/phi/common/transform.h"
 #include "paddle/phi/kernels/impl/clip_kernel_impl.h"
 
 namespace paddle {
@@ -98,7 +98,7 @@ struct ClipAndFakeQuantFunctor<phi::CPUContext, T> {
                   phi::DenseTensor *out) {
     T s = scale.data<T>()[0];
     T inv_s = inverse(s);
-    platform::Transform<phi::CPUContext> trans;
+    phi::Transform<phi::CPUContext> trans;
     if (round_type == 0) {
       trans(ctx,
             in.data<T>(),
@@ -130,7 +130,7 @@ struct ClipAndFakeQuantDequantFunctor<phi::CPUContext, T> {
     T s = scale.data<T>()[0];
     T inv_s = inverse(s);
 
-    platform::Transform<phi::CPUContext> trans;
+    phi::Transform<phi::CPUContext> trans;
     if (round_type == 0) {
       trans(ctx,
             in.data<T>(),
@@ -175,7 +175,7 @@ struct ChannelClipAndFakeQuantFunctor<phi::CPUContext, T> {
     auto *out_data = out->mutable_data<T>(ctx.GetPlace());
     auto in_dims = in.dims();
     const int64_t channel = in_dims[quant_axis];
-    platform::Transform<phi::CPUContext> trans;
+    phi::Transform<phi::CPUContext> trans;
     if (quant_axis == 0) {
       const int64_t channel_size = in.numel() / channel;
       for (int64_t i = 0; i < channel; i++) {
@@ -256,7 +256,7 @@ struct ChannelClipFakeQuantDequantFunctor<phi::CPUContext, T> {
     auto *out_data = out->mutable_data<T>(ctx.GetPlace());
     auto in_dims = in.dims();
     const int64_t channel = in_dims[quant_axis];
-    platform::Transform<phi::CPUContext> trans;
+    phi::Transform<phi::CPUContext> trans;
     if (quant_axis == 0) {
       const int64_t channel_size = in.numel() / channel;
       for (int i = 0; i < channel; i++) {

paddle/fluid/operators/fake_quantize_op.h

@@ -20,7 +20,7 @@ limitations under the License. */
 #include "paddle/fluid/framework/op_registry.h"
 #include "paddle/fluid/framework/tensor_util.h"
 #include "paddle/fluid/memory/malloc.h"
-#include "paddle/fluid/platform/transform.h"
+#include "paddle/phi/common/transform.h"
 #include "paddle/phi/core/hostdevice.h"
 #include "paddle/phi/kernels/funcs/blas/blas.h"
 

paddle/fluid/operators/isfinite_op.h

@@ -20,7 +20,7 @@
 #include "paddle/fluid/framework/op_registry.h"
 #include "paddle/fluid/framework/tensor_util.h"
 #include "paddle/fluid/platform/float16.h"
-#include "paddle/fluid/platform/transform.h"
+#include "paddle/phi/common/transform.h"
 #include "paddle/phi/kernels/isfinite_kernel.h"
 #include "paddle/phi/kernels/reduce_all_kernel.h"
 #include "paddle/phi/kernels/reduce_any_kernel.h"

paddle/fluid/operators/lstmp_op.h

@@ -20,7 +20,7 @@ limitations under the License. */
 #include "paddle/fluid/framework/op_registry.h"
 #include "paddle/fluid/operators/activation_op.h"
 #include "paddle/fluid/platform/place.h"
-#include "paddle/fluid/platform/transform.h"
+#include "paddle/phi/common/transform.h"
 #include "paddle/phi/kernels/funcs/blas/blas.h"
 #include "paddle/phi/kernels/funcs/detail/activation_functions.h"
 #include "paddle/phi/kernels/funcs/lstm_compute.h"
@@ -29,7 +29,7 @@ limitations under the License. */
 namespace paddle {
 namespace operators {
 
-using platform::Transform;
+using phi::Transform;
 
 template <typename T,
           int MajorType = Eigen::RowMajor,

paddle/fluid/operators/quantize_linear_op.cc

@@ -17,7 +17,7 @@ limitations under the License. */
 
 #include "paddle/fluid/framework/eigen.h"
 #include "paddle/fluid/framework/op_version_registry.h"
-#include "paddle/fluid/platform/transform.h"
+#include "paddle/phi/common/transform.h"
 #include "paddle/phi/kernels/funcs/blas/blas.h"
 #include "paddle/phi/kernels/impl/clip_kernel_impl.h"
 

paddle/fluid/operators/quantize_linear_op.h

@@ -18,8 +18,8 @@ limitations under the License. */
 #include "paddle/fluid/framework/tensor_util.h"
 #include "paddle/fluid/memory/malloc.h"
 #include "paddle/fluid/operators/fake_quantize_op.h"
-#include "paddle/fluid/platform/transform.h"
 #include "paddle/phi/common/data_type.h"
+#include "paddle/phi/common/transform.h"
 #include "paddle/phi/core/ddim.h"
 #include "paddle/phi/core/hostdevice.h"
 #include "paddle/phi/kernels/cast_kernel.h"

paddle/fluid/platform/CMakeLists.txt

@@ -248,10 +248,6 @@ if(WITH_GPU)
     device_context_test_cuda_graph
     SRCS device_context_test_cuda_graph.cu
     DEPS device_context gpu_info cuda_graph_with_memory_pool)
-  nv_test(
-    transform_test
-    SRCS transform_test.cu
-    DEPS memory place device_context)
 endif()
 
 if(WITH_ROCM)
@@ -277,10 +273,6 @@ if(WITH_ROCM)
     device_context_test
     SRCS device_context_test.cu
     DEPS device_context gpu_info)
-  hip_test(
-    transform_test
-    SRCS transform_test.cu
-    DEPS memory place device_context)
 endif()
 
 cc_library(timer SRCS timer.cc)

paddle/fluid/platform/details/cuda_transform_iterator_cast.h

-/* Copyright (c) 2016 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
-
-#if !defined(__NVCC__) && !defined(__HIPCC__)
-#error device_ptr_cast must be include by .cu file
-#endif
-
-#include <type_traits>  // For std::remove_pointer and std::is_pointer.
-
-#include "thrust/device_ptr.h"
-
-namespace paddle {
-namespace platform {
-namespace details {
-
-// PointerToThrustDevicePtr has two speicalizations, one casts a (CUDA
-// device) pointer into thrust::device_ptr, the other keeps rest types
-// un-casted.
-template <typename T, bool is_ptr>
-struct PointerToThrustDevicePtr;
-
-template <typename T>
-struct PointerToThrustDevicePtr<T, true> {
-  using ELEM = typename std::remove_pointer<T>::type;
-  using RTYPE = thrust::device_ptr<ELEM>;
-
-  inline thrust::device_ptr<ELEM> operator()(ELEM* ele) const {
-    return thrust::device_pointer_cast(ele);
-  }
-};
-
-template <typename T>
-struct PointerToThrustDevicePtr<T, false> {
-  using RTYPE = T;
-  inline RTYPE operator()(RTYPE it) const { return it; }
-};
-
-// CastToCUDATransformIterator casts a pointer to thrust::device_ptr
-// so it could be used as the iterator of thrust::transform.  It
-// doesn't cast other types.
-//
-// We need CastToCUDATransformIterator because it is often that we
-// want to use device memory pointers as transform iterators, e.g., to
-// transform a block of float32 to float16.  In this case, we want
-// CastToCUDATransformIterator to cast float16/32 pointers to
-// thrust::device_ptr, otherwise they cannot work as the iterator
-// required by thrust::transform.  At the same time, we don't want to
-// cast thrust::device_ptr to thrust::device_ptr repeatedly.
-template <typename T>
-auto CastToCUDATransformIterator(T t) ->
-    typename PointerToThrustDevicePtr<T, std::is_pointer<T>::value>::RTYPE {
-  PointerToThrustDevicePtr<T, std::is_pointer<T>::value> cast;
-  return cast(t);
-}
-
-}  // namespace details
-}  // namespace platform
-}  // namespace paddle

paddle/fluid/platform/transform.h → paddle/phi/common/transform.h

-/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved.
+/* 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.
@@ -17,20 +17,17 @@ limitations under the License. */
 #include <algorithm>
 #include <type_traits>
 
-#include "paddle/fluid/platform/device_context.h"
-#include "paddle/fluid/platform/enforce.h"
-#include "paddle/fluid/platform/place.h"
+#include "paddle/phi/backends/all_context.h"
+#include "paddle/phi/core/enforce.h"
 #include "paddle/phi/core/hostdevice.h"
 
 #if defined(__NVCC__) || defined(__HIPCC__)
 #include <thrust/execution_policy.h>
 #include <thrust/transform.h>
-
-#include "paddle/fluid/platform/details/cuda_transform_iterator_cast.h"
+#include "thrust/device_ptr.h"
 #endif
 
-namespace paddle {
-namespace platform {
+namespace phi {
 
 // Transform applys a unary or a binary functor on each element in a
 // range defined by a pair of iterators.
@@ -40,16 +37,16 @@ namespace platform {
 //
 // NOTE: We need to define InputIter and OutputIter defined as
 //       different types, because the InputIter points op's inputs and
-//       OutputIter pints to op's outputs.
+//       OutputIter points to op's outputs.
 //
 // NOTE: We don't assume that InputIter to be const InputType* and
 //       OutputIter to be OutputType*, because we might use a iterator
 //       class, paddle::fluid::operators::RowwiseTRansformIterator.
-template <typename DeviceContext>
+template <typename Context>
 struct Transform {
   // The unary version.
   template <typename InputIter, typename OutputIter, typename UnaryOperation>
-  void operator()(const DeviceContext& context,
+  void operator()(const Context& context,
                   InputIter first,
                   InputIter last,
                   OutputIter result,
@@ -60,7 +57,7 @@ struct Transform {
             typename InputIter2,
             typename OutputIter,
             typename BinaryOperation>
-  void operator()(const DeviceContext& context,
+  void operator()(const Context& context,
                   InputIter1 first1,
                   InputIter1 last1,
                   InputIter2 first2,
@@ -97,6 +94,46 @@ struct Transform<phi::CPUContext> {
 
 #if defined(__NVCC__) || defined(__HIPCC__)
 
+// PointerToThrustDevicePtr has two speicalizations, one casts a (CUDA
+// device) pointer into thrust::device_ptr, the other keeps rest types
+// un-casted.
+template <typename T, bool is_ptr>
+struct PointerToThrustDevicePtr;
+
+template <typename T>
+struct PointerToThrustDevicePtr<T, true> {
+  using ELEM = typename std::remove_pointer<T>::type;
+  using RTYPE = thrust::device_ptr<ELEM>;
+
+  inline thrust::device_ptr<ELEM> operator()(ELEM* ele) const {
+    return thrust::device_pointer_cast(ele);
+  }
+};
+
+template <typename T>
+struct PointerToThrustDevicePtr<T, false> {
+  using RTYPE = T;
+  inline RTYPE operator()(RTYPE it) const { return it; }
+};
+
+// CastToCUDATransformIterator casts a pointer to thrust::device_ptr
+// so it could be used as the iterator of thrust::transform.  It
+// doesn't cast other types.
+//
+// We need CastToCUDATransformIterator because it is often that we
+// want to use device memory pointers as transform iterators, e.g., to
+// transform a block of float32 to float16.  In this case, we want
+// CastToCUDATransformIterator to cast float16/32 pointers to
+// thrust::device_ptr, otherwise they cannot work as the iterator
+// required by thrust::transform.  At the same time, we don't want to
+// cast thrust::device_ptr to thrust::device_ptr repeatedly.
+template <typename T>
+auto CastToCUDATransformIterator(T t) ->
+    typename PointerToThrustDevicePtr<T, std::is_pointer<T>::value>::RTYPE {
+  PointerToThrustDevicePtr<T, std::is_pointer<T>::value> cast;
+  return cast(t);
+}
+
 template <>
 struct Transform<phi::GPUContext> {
   template <typename InputIter, typename OutputIter, typename UnaryOperation>
@@ -106,21 +143,21 @@ struct Transform<phi::GPUContext> {
                   OutputIter result,
                   UnaryOperation op) {
     auto place = context.GetPlace();
-    PADDLE_ENFORCE_EQ(is_gpu_place(place),
+    PADDLE_ENFORCE_EQ(place.GetType() == phi::AllocationType::GPU,
                       true,
-                      platform::errors::PreconditionNotMet(
+                      phi::errors::PreconditionNotMet(
                           "The CUDA Transform must be used in GPU place."));
 #ifdef __HIPCC__
     thrust::transform(thrust::hip::par.on(context.stream()),
-                      details::CastToCUDATransformIterator(first),
-                      details::CastToCUDATransformIterator(last),
-                      details::CastToCUDATransformIterator(result),
+                      CastToCUDATransformIterator(first),
+                      CastToCUDATransformIterator(last),
+                      CastToCUDATransformIterator(result),
                       op);
 #else
     thrust::transform(thrust::cuda::par.on(context.stream()),
-                      details::CastToCUDATransformIterator(first),
-                      details::CastToCUDATransformIterator(last),
-                      details::CastToCUDATransformIterator(result),
+                      CastToCUDATransformIterator(first),
+                      CastToCUDATransformIterator(last),
+                      CastToCUDATransformIterator(result),
                       op);
 #endif
   }
@@ -136,28 +173,27 @@ struct Transform<phi::GPUContext> {
                   OutputIter result,
                   BinaryOperation op) {
     auto place = context.GetPlace();
-    PADDLE_ENFORCE_EQ(is_gpu_place(place),
+    PADDLE_ENFORCE_EQ(place.GetType() == phi::AllocationType::GPU,
                       true,
-                      platform::errors::PreconditionNotMet(
+                      phi::errors::PreconditionNotMet(
                           "The CUDA Transform must be used in GPU place."));
 #ifdef __HIPCC__
     thrust::transform(thrust::hip::par.on(context.stream()),
-                      details::CastToCUDATransformIterator(first1),
-                      details::CastToCUDATransformIterator(last1),
-                      details::CastToCUDATransformIterator(first2),
-                      details::CastToCUDATransformIterator(result),
+                      CastToCUDATransformIterator(first1),
+                      CastToCUDATransformIterator(last1),
+                      CastToCUDATransformIterator(first2),
+                      CastToCUDATransformIterator(result),
                       op);
 #else
     thrust::transform(thrust::cuda::par.on(context.stream()),
-                      details::CastToCUDATransformIterator(first1),
-                      details::CastToCUDATransformIterator(last1),
-                      details::CastToCUDATransformIterator(first2),
-                      details::CastToCUDATransformIterator(result),
+                      CastToCUDATransformIterator(first1),
+                      CastToCUDATransformIterator(last1),
+                      CastToCUDATransformIterator(first2),
+                      CastToCUDATransformIterator(result),
                       op);
 #endif
   }
 };
 #endif
 
-}  // namespace platform
-}  // namespace paddle
+}  // namespace phi

paddle/phi/kernels/cpu/bitwise_kernel.cc

@@ -20,7 +20,7 @@ limitations under the License. */
 #include "paddle/phi/kernels/funcs/elementwise_base.h"
 
 // See Note [ Why still include the fluid headers? ]
-#include "paddle/fluid/platform/transform.h"
+#include "paddle/phi/common/transform.h"
 
 namespace phi {
 
@@ -48,7 +48,7 @@ void BitwiseNotKernel(const Context& dev_ctx,
   T* out_data = dev_ctx.template Alloc<T>(out);
   size_t numel = x.numel();
   funcs::BitwiseNotFunctor<T> func;
-  paddle::platform::Transform<Context> trans;
+  phi::Transform<Context> trans;
   trans(dev_ctx, x_data, x_data + numel, out_data, func);
 }
 

paddle/phi/kernels/cpu/cast_impl.h

@@ -17,7 +17,7 @@
 #include "paddle/phi/backends/cpu/cpu_context.h"
 
 // See Note [ Why still include the fluid headers? ]
-#include "paddle/fluid/platform/transform.h"
+#include "paddle/phi/common/transform.h"
 
 namespace phi {
 
@@ -36,7 +36,7 @@ void CastKernelImpl(const CPUContext& dev_ctx,
 
   auto* out_begin = dev_ctx.Alloc<OutT>(out);
 
-  paddle::platform::Transform<CPUContext> trans;
+  phi::Transform<CPUContext> trans;
   trans(dev_ctx,
         in_begin,
         in_end,

paddle/phi/kernels/cpu/hsigmoid_loss_kernel.cc

@@ -14,8 +14,8 @@
 
 #include "paddle/phi/kernels/hsigmoid_loss_kernel.h"
 
-#include "paddle/fluid/platform/transform.h"
 #include "paddle/phi/backends/cpu/cpu_context.h"
+#include "paddle/phi/common/transform.h"
 #include "paddle/phi/core/kernel_registry.h"
 #include "paddle/phi/kernels/funcs/eigen/common.h"
 #include "paddle/phi/kernels/funcs/eigen/eigen_function.h"
@@ -82,7 +82,7 @@ void HSigmoidLossKernel(const Context& ctx,
   }
   bit_code->Mul(pre_out, w, x);
   // clip to [-40, 40]
-  paddle::platform::Transform<Context> trans;
+  phi::Transform<Context> trans;
   trans(ctx,
         pre_out_data,
         pre_out_data + pre_out->numel(),

paddle/phi/kernels/cpu/logical_kernel.cc

@@ -20,7 +20,7 @@
 #include "paddle/phi/kernels/funcs/logical_functor.h"
 
 // See Note [ Why still include the fluid headers? ]
-#include "paddle/fluid/platform/transform.h"
+#include "paddle/phi/common/transform.h"
 
 namespace phi {
 
@@ -47,7 +47,7 @@ void LogicalNotKernel(const Context& dev_ctx,
   auto* out_ptr = dev_ctx.template Alloc<bool>(out);
   funcs::LogicalNotFunctor<T> unary_func;
 
-  paddle::platform::Transform<Context> trans;
+  phi::Transform<Context> trans;
   trans(dev_ctx, x.data<T>(), x.data<T>() + x.numel(), out_ptr, unary_func);
 }
 

paddle/phi/kernels/funcs/elementwise_base.h

@@ -14,8 +14,8 @@ limitations under the License. */
 
 #pragma once
 
-#include "paddle/fluid/platform/transform.h"
 #include "paddle/phi/backends/all_context.h"
+#include "paddle/phi/common/transform.h"
 #include "paddle/phi/core/dense_tensor.h"
 #include "paddle/phi/kernels/empty_kernel.h"
 #include "paddle/phi/kernels/funcs/common_shape.h"
@@ -220,12 +220,12 @@ class TransformFunctor {
   }
 
   inline void Run() const {
-    paddle::platform::Transform<DeviceContext> trans;
+    phi::Transform<DeviceContext> trans;
     trans(ctx_, x_, x_ + nx_, y_, z_, func_);
   }
 
   inline void RunRowWise(int n, int pre) const {
-    paddle::platform::Transform<DeviceContext> trans;
+    phi::Transform<DeviceContext> trans;
     if (is_xsize_larger_) {
       trans(ctx_,
             x_,
@@ -244,7 +244,7 @@ class TransformFunctor {
   }
 
   inline void RunMidWise(int n, int pre, int post) const {
-    paddle::platform::Transform<DeviceContext> trans;
+    phi::Transform<DeviceContext> trans;
     if (is_xsize_larger_) {
       trans(ctx_,
             x_,

paddle/phi/kernels/impl/clip_grad_kernel_impl.h

@@ -15,8 +15,8 @@
 #pragma once
 
 #include "paddle/fluid/platform/device_context.h"
-#include "paddle/fluid/platform/transform.h"
 #include "paddle/phi/backends/all_context.h"
+#include "paddle/phi/common/transform.h"
 #include "paddle/phi/core/kernel_registry.h"
 #include "paddle/phi/kernels/clip_kernel.h"
 #if defined(__NVCC__) || defined(__HIPCC__)
@@ -59,7 +59,7 @@ void ClipGradKernel(const Context& dev_ctx,
   auto* d_x_data = dev_ctx.template Alloc<T>(x_grad);
   const T* d_out_data = out_grad.data<T>();
   const T* x_data = x.data<T>();
-  paddle::platform::Transform<Context> trans;
+  phi::Transform<Context> trans;
   trans(dev_ctx,
         d_out_data,
         d_out_data + numel,

paddle/phi/kernels/impl/clip_kernel_impl.h

@@ -14,8 +14,8 @@
 
 #pragma once
 
-#include "paddle/fluid/platform/transform.h"
 #include "paddle/phi/backends/all_context.h"
+#include "paddle/phi/common/transform.h"
 #include "paddle/phi/core/kernel_registry.h"
 #include "paddle/phi/kernels/clip_kernel.h"
 #if defined(__NVCC__) || defined(__HIPCC__)
@@ -67,7 +67,7 @@ void ClipKernel(const Context& dev_ctx,
     phi::funcs::ElementwiseKernel<T>(dev_ctx, ins, &outs, functor);
 #endif
   } else {
-    paddle::platform::Transform<Context> trans;
+    phi::Transform<Context> trans;
     trans(
         dev_ctx, x_data, x_data + numel, out_data, ClipFunctor<T>(min_, max_));
   }

paddle/phi/kernels/impl/isfinite_kernel_impl.h

@@ -18,7 +18,7 @@
 #include "paddle/phi/kernels/isfinite_kernel.h"
 
 // See Note [ Why still include the fluid headers? ]
-#include "paddle/fluid/platform/transform.h"
+#include "paddle/phi/common/transform.h"
 
 namespace phi {
 
@@ -28,7 +28,7 @@ namespace phi {
       const Context& ctx, const DenseTensor& x, DenseTensor* out) {        \
     auto* out_ptr = ctx.template Alloc<bool>(out);                         \
     funcs::functor<T> unary_func;                                          \
-    paddle::platform::Transform<Context> trans;                            \
+    phi::Transform<Context> trans;                                         \
     trans(ctx, x.data<T>(), x.data<T>() + x.numel(), out_ptr, unary_func); \
   }
 

paddle/phi/kernels/selected_rows/impl/clip_kernel_impl.h

@@ -50,7 +50,7 @@ void ClipSparseKernel(const Context& dev_ctx,
   auto* out_tensor = out->mutable_value();
   auto* out_data = out_tensor->data<T>();
   int64_t numel = out_tensor->numel();
-  paddle::platform::Transform<Context> trans;
+  phi::Transform<Context> trans;
   trans(dev_ctx,
         out_data,
         out_data + numel,

paddle/phi/tests/common/CMakeLists.txt

@@ -23,10 +23,18 @@ if(WITH_GPU)
     phi_test_scalar
     SRCS test_scalar.cu
     DEPS scalar api_scalar)
+  nv_test(
+    transform_test
+    SRCS transform_test.cu
+    DEPS memory place device_context)
 endif()
 if(WITH_ROCM)
   hip_test(
     phi_test_scalar
     SRCS test_scalar.cu
     DEPS scalar api_scalar)
+  hip_test(
+    transform_test
+    SRCS transform_test.cu
+    DEPS memory place device_context)
 endif()

paddle/fluid/platform/transform_test.cu → paddle/phi/tests/common/transform_test.cu

@@ -14,10 +14,11 @@ limitations under the License. */
 
 #include <gtest/gtest.h>
 
-#include "paddle/fluid/memory/allocation/allocator_facade.h"
+#include "paddle/phi/common/transform.h"
+
 #include "paddle/fluid/memory/memcpy.h"
 #include "paddle/fluid/memory/memory.h"
-#include "paddle/fluid/platform/transform.h"
+#include "paddle/phi/backends/all_context.h"
 #include "paddle/phi/core/hostdevice.h"
 
 template <typename T>
@@ -44,7 +45,7 @@ using paddle::platform::CUDAPlace;
 using phi::CPUContext;
 using phi::GPUContext;
 
-using paddle::platform::Transform;
+using phi::Transform;
 
 TEST(Transform, CPUUnary) {
   CPUContext ctx;
@@ -58,19 +59,17 @@ TEST(Transform, CPUUnary) {
 
 TEST(Transform, GPUUnary) {
   CUDAPlace gpu0(0);
-  phi::GPUContext ctx(gpu0);
-  ctx.SetAllocator(paddle::memory::allocation::AllocatorFacade::Instance()
-                       .GetAllocator(gpu0, ctx.stream())
-                       .get());
-  ctx.PartialInitWithAllocator();
+  phi::DeviceContextPool& pool = phi::DeviceContextPool::Instance();
+  auto* ctx = reinterpret_cast<phi::GPUContext*>(pool.Get(phi::GPUPlace()));
+
   float cpu_buf[4] = {0.1, 0.2, 0.3, 0.4};
   auto gpu_allocation = Alloc(gpu0, sizeof(float) * 4);
   float* gpu_buf = static_cast<float*>(gpu_allocation->ptr());
-  Copy(gpu0, gpu_buf, CPUPlace(), cpu_buf, sizeof(cpu_buf), ctx.stream());
+  Copy(gpu0, gpu_buf, CPUPlace(), cpu_buf, sizeof(cpu_buf), ctx->stream());
   Transform<phi::GPUContext> trans;
-  trans(ctx, gpu_buf, gpu_buf + 4, gpu_buf, Scale<float>(10));
-  ctx.Wait();
-  Copy(CPUPlace(), cpu_buf, gpu0, gpu_buf, sizeof(cpu_buf), ctx.stream());
+  trans(*ctx, gpu_buf, gpu_buf + 4, gpu_buf, Scale<float>(10));
+  ctx->Wait();
+  Copy(CPUPlace(), cpu_buf, gpu0, gpu_buf, sizeof(cpu_buf), ctx->stream());
   for (int i = 0; i < 4; ++i) {
     ASSERT_NEAR(cpu_buf[i], static_cast<float>(i + 1), 1e-5);
   }
@@ -89,18 +88,16 @@ TEST(Transform, CPUBinary) {
 TEST(Transform, GPUBinary) {
   int buf[4] = {1, 2, 3, 4};
   CUDAPlace gpu0(0);
-  phi::GPUContext ctx(gpu0);
-  ctx.SetAllocator(paddle::memory::allocation::AllocatorFacade::Instance()
-                       .GetAllocator(gpu0, ctx.stream())
-                       .get());
-  ctx.PartialInitWithAllocator();
+  phi::DeviceContextPool& pool = phi::DeviceContextPool::Instance();
+  auto* ctx = reinterpret_cast<phi::GPUContext*>(pool.Get(phi::GPUPlace()));
+
   auto gpu_allocation = Alloc(gpu0, sizeof(buf));
   int* gpu_buf = static_cast<int*>(gpu_allocation->ptr());
-  Copy(gpu0, gpu_buf, CPUPlace(), buf, sizeof(buf), ctx.stream());
+  Copy(gpu0, gpu_buf, CPUPlace(), buf, sizeof(buf), ctx->stream());
   Transform<phi::GPUContext> trans;
-  trans(ctx, gpu_buf, gpu_buf + 4, gpu_buf, gpu_buf, Multiply<int>());
-  ctx.Wait();
-  Copy(CPUPlace(), buf, gpu0, gpu_buf, sizeof(buf), ctx.stream());
+  trans(*ctx, gpu_buf, gpu_buf + 4, gpu_buf, gpu_buf, Multiply<int>());
+  ctx->Wait();
+  Copy(CPUPlace(), buf, gpu0, gpu_buf, sizeof(buf), ctx->stream());
   for (int i = 0; i < 4; ++i) {
     ASSERT_EQ((i + 1) * (i + 1), buf[i]);
   }