Clean paddle/fluid/platform/for_range.h [fluid_ops](PaddlePaddle#66142)

Author: co63oc

paddle/fluid/imperative/reducer.h

@@ -30,7 +30,7 @@
 #include "paddle/fluid/framework/tensor.h"
 #include "paddle/fluid/framework/tensor_util.h"
 #include "paddle/fluid/framework/variable.h"
-#include "paddle/fluid/platform/for_range.h"
+#include "paddle/phi/kernels/funcs/for_range.h"
 #include "paddle/phi/kernels/funcs/math_function.h"
 
 namespace paddle {
@@ -72,8 +72,8 @@ struct DivNRanksForAllReduce {
   template <typename T>
   void apply() const {
     T* data = in_->mutable_data<T>(ctx_.GetPlace());
-    platform::ForRange<Dex> for_range(static_cast<const Dex&>(ctx_),
-                                      static_cast<size_t>(in_->numel()));
+    phi::funcs::ForRange<Dex> for_range(static_cast<const Dex&>(ctx_),
+                                        static_cast<size_t>(in_->numel()));
     DivNRanksFunctor<T> functor(nranks_, data);
     for_range(functor);
   }

paddle/fluid/platform/for_range.h

@@ -22,7 +22,7 @@
 #include "paddle/fluid/memory/allocation/best_fit_allocator.h"
 #include "paddle/fluid/memory/allocation/cuda_allocator.h"
 #include "paddle/fluid/memory/memcpy.h"
-#include "paddle/fluid/platform/for_range.h"
+#include "paddle/phi/kernels/funcs/for_range.h"
 namespace paddle {
 namespace memory {
 namespace allocation {
@@ -62,7 +62,7 @@ TEST(BestFitAllocator, concurrent_cuda) {
       size_t* data = reinterpret_cast<size_t*>(allocation->ptr());
 
       ForEachFill fill(data);
-      platform::ForRange<phi::GPUContext> for_range(dev_ctx, allocate_size);
+      phi::funcs::ForRange<phi::GPUContext> for_range(dev_ctx, allocate_size);
       for_range(fill);
 
       memory::Copy(phi::CPUPlace(),

test/cpp/fluid/memory/best_fit_allocator_test.cu

@@ -19,7 +19,7 @@
 
 #include "gtest/gtest.h"
 #include "paddle/fluid/operators/activation_op.h"
-#include "paddle/fluid/platform/for_range.h"
+#include "paddle/phi/kernels/funcs/for_range.h"
 
 namespace paddle {
 namespace operators {
@@ -98,13 +98,13 @@ static bool TestLeakyReluGradGradMain(const phi::DDim &dim,
   if (phi::is_gpu_place(place)) {
     auto &cuda_dev_ctx = dynamic_cast<phi::GPUContext &>(dev_ctx);
     functor(cuda_dev_ctx, &x, out, &ddx, &ddout, dout, dx);
-    platform::ForRange<phi::GPUContext> for_range(cuda_dev_ctx, limit);
+    phi::funcs::ForRange<phi::GPUContext> for_range(cuda_dev_ctx, limit);
     for_range(actual_functor);
   } else {
 #endif
     auto &cpu_dev_ctx = dynamic_cast<phi::CPUContext &>(dev_ctx);
     functor(cpu_dev_ctx, &x, out, &ddx, &ddout, dout, dx);
-    platform::ForRange<phi::CPUContext> for_range(cpu_dev_ctx, limit);
+    phi::funcs::ForRange<phi::CPUContext> for_range(cpu_dev_ctx, limit);
     for_range(actual_functor);
 #if defined(__NVCC__) || defined(__HIPCC__)
   }

test/cpp/fluid/test_leaky_relu_grad_grad_functor.h

@@ -18,7 +18,7 @@
 #include "paddle/fluid/framework/data_type.h"
 #include "paddle/fluid/framework/tensor.h"
 #include "paddle/fluid/platform/device_context.h"
-#include "paddle/fluid/platform/for_range.h"
+#include "paddle/phi/kernels/funcs/for_range.h"
 
 namespace custom_raw_op {
 
@@ -50,12 +50,12 @@ struct ReluFunctor {
 
     const auto &dev_ctx = *phi::DeviceContextPool::Instance().Get(place);
 
-#define LAUNCH_RELU_KERNEL(DevCtxT)                              \
-  do {                                                           \
-    auto &__dev_ctx = dynamic_cast<const DevCtxT &>(dev_ctx);    \
-    paddle::platform::ForRange<DevCtxT> for_range(__dev_ctx, n); \
-    Impl<T> functor(x_data, y_data);                             \
-    for_range(functor);                                          \
+#define LAUNCH_RELU_KERNEL(DevCtxT)                           \
+  do {                                                        \
+    auto &__dev_ctx = dynamic_cast<const DevCtxT &>(dev_ctx); \
+    phi::funcs::ForRange<DevCtxT> for_range(__dev_ctx, n);    \
+    Impl<T> functor(x_data, y_data);                          \
+    for_range(functor);                                       \
   } while (0)
 
 #if defined(__NVCC__) || defined(__HIPCC__)