Rename GPU device array helpers from cuda_ to gpu_

Common utility helpers between CUDA and ROCm are renamed:

- core/kernels/cuda_device_array.h -> core/kernels/gpu_device_array.h
- core/kernels/cuda_device_array_gpu.h -> core/kernels/gpu_device_array_gpu.h

In subsequent PRs, ROCm-specific terminology changes would be introduced.

tensorflow/core/kernels/BUILD

@@ -232,8 +232,8 @@ tf_kernel_library(
         "concat_lib_gpu_impl.cu.cc",
         "concat_lib.h",
         "concat_lib_gpu.h",
-        "cuda_device_array.h",
-        "cuda_device_array_gpu.h",
+        "gpu_device_array.h",
+        "gpu_device_array_gpu.h",
     ],
     deps = [
         ":bounds_check",
@@ -632,7 +632,7 @@ tf_kernel_library(
         "split_lib_gpu.h",
     ],
     deps = [
-        ":cuda_device_array",
+        ":gpu_device_array",
         "//tensorflow/core:framework",
         "//third_party/eigen3",
     ],
@@ -684,10 +684,10 @@ cc_header_only_library(
 )
 
 cc_library(
-    name = "cuda_device_array",
+    name = "gpu_device_array",
     hdrs = [
-        "cuda_device_array.h",
-        "cuda_device_array_gpu.h",
+        "gpu_device_array.h",
+        "gpu_device_array_gpu.h",
     ],
     visibility = ["//tensorflow:__subpackages__"],
     deps = [
@@ -1110,14 +1110,14 @@ tf_kernel_library(
 
 tf_kernel_library(
     name = "split_op",
-    gpu_srcs = ["cuda_device_array.h"],
+    gpu_srcs = ["gpu_device_array.h"],
     prefix = "split_op",
     deps = ARRAY_DEPS + [":split_lib"],
 )
 
 tf_kernel_library(
     name = "split_v_op",
-    gpu_srcs = ["cuda_device_array.h"],
+    gpu_srcs = ["gpu_device_array.h"],
     prefix = "split_v_op",
     deps = ARRAY_DEPS + [":split_lib"],
 )
@@ -2152,8 +2152,8 @@ tf_kernel_library(
 tf_kernel_library(
     name = "dynamic_stitch_op",
     gpu_srcs = [
-        "cuda_device_array.h",
-        "cuda_device_array_gpu.h",
+        "gpu_device_array.h",
+        "gpu_device_array_gpu.h",
     ],
     prefix = "dynamic_stitch_op",
     deps = DYNAMIC_DEPS,
@@ -3301,7 +3301,7 @@ tf_kernel_library(
 
 tf_kernel_library(
     name = "bucketize_op",
-    gpu_srcs = ["cuda_device_array.h"],
+    gpu_srcs = ["gpu_device_array.h"],
     prefix = "bucketize_op",
     deps = ARRAY_DEPS,
 )

tensorflow/core/kernels/bucketize_op_gpu.cu.cc

@@ -24,7 +24,7 @@ limitations under the License.
 #include "tensorflow/core/framework/tensor.h"
 #include "tensorflow/core/framework/tensor_shape.h"
 #include "tensorflow/core/kernels/bucketize_op.h"
-#include "tensorflow/core/kernels/cuda_device_array.h"
+#include "tensorflow/core/kernels/gpu_device_array.h"
 #include "tensorflow/core/platform/logging.h"
 #include "tensorflow/core/platform/types.h"
 #include "tensorflow/core/util/cuda_kernel_helper.h"

tensorflow/core/kernels/concat_lib_gpu.cc

@@ -27,7 +27,7 @@ limitations under the License.
 #if GOOGLE_CUDA
 
 #include "tensorflow/core/kernels/concat_lib_gpu.h"
-#include "tensorflow/core/kernels/cuda_device_array.h"
+#include "tensorflow/core/kernels/gpu_device_array.h"
 
 namespace tensorflow {
 namespace {

tensorflow/core/kernels/concat_lib_gpu_impl.cu.cc

@@ -24,7 +24,7 @@ limitations under the License.
 #include "tensorflow/core/framework/register_types.h"
 #include "tensorflow/core/framework/tensor_types.h"
 #include "tensorflow/core/kernels/concat_lib_gpu.h"
-#include "tensorflow/core/kernels/cuda_device_array_gpu.h"
+#include "tensorflow/core/kernels/gpu_device_array_gpu.h"
 #include "tensorflow/core/util/cuda_kernel_helper.h"
 
 namespace tensorflow {

tensorflow/core/kernels/dynamic_stitch_op.cc

@@ -22,7 +22,7 @@ limitations under the License.
 #include "tensorflow/core/lib/core/threadpool.h"
 
 #ifdef GOOGLE_CUDA
-#include "tensorflow/core/kernels/cuda_device_array.h"
+#include "tensorflow/core/kernels/gpu_device_array.h"
 #endif  // GOOGLE_CUDA
 
 namespace tensorflow {

tensorflow/core/kernels/dynamic_stitch_op_gpu.cu.cc

@@ -19,7 +19,7 @@ limitations under the License.
 
 #include "tensorflow/core/framework/register_types.h"
 #include "tensorflow/core/framework/tensor_types.h"
-#include "tensorflow/core/kernels/cuda_device_array_gpu.h"
+#include "tensorflow/core/kernels/gpu_device_array_gpu.h"
 #include "tensorflow/core/util/cuda_kernel_helper.h"
 
 namespace tensorflow {

tensorflow/core/kernels/gpu_device_array.h

@@ -19,7 +19,7 @@ limitations under the License.
 
 #include "tensorflow/core/common_runtime/gpu/gpu_event_mgr.h"
 #include "tensorflow/core/framework/op_kernel.h"
-#include "tensorflow/core/kernels/cuda_device_array_gpu.h"
+#include "tensorflow/core/kernels/gpu_device_array_gpu.h"
 
 namespace tensorflow {
 

tensorflow/core/kernels/split_lib_gpu.cu.cc

@@ -23,7 +23,7 @@ limitations under the License.
 
 #include "tensorflow/core/framework/register_types.h"
 #include "tensorflow/core/framework/tensor_types.h"
-#include "tensorflow/core/kernels/cuda_device_array_gpu.h"
+#include "tensorflow/core/kernels/gpu_device_array_gpu.h"
 #include "tensorflow/core/kernels/split_lib_gpu.h"
 #include "tensorflow/core/util/cuda_kernel_helper.h"
 

tensorflow/core/kernels/split_op.cc

@@ -29,7 +29,7 @@ limitations under the License.
 #include "tensorflow/core/util/work_sharder.h"
 #if GOOGLE_CUDA
 #include "tensorflow/core/common_runtime/gpu/gpu_event_mgr.h"
-#include "tensorflow/core/kernels/cuda_device_array.h"
+#include "tensorflow/core/kernels/gpu_device_array.h"
 #include "tensorflow/core/kernels/split_lib_gpu.h"
 #include "tensorflow/core/platform/stream_executor.h"
 #endif  // GOOGLE_CUDA

tensorflow/core/kernels/split_v_op.cc

@@ -35,7 +35,7 @@ limitations under the License.
 #include "tensorflow/core/util/work_sharder.h"
 #if GOOGLE_CUDA
 #include "tensorflow/core/common_runtime/gpu/gpu_event_mgr.h"
-#include "tensorflow/core/kernels/cuda_device_array.h"
+#include "tensorflow/core/kernels/gpu_device_array.h"
 #include "tensorflow/core/kernels/split_lib_gpu.h"
 #include "tensorflow/core/platform/stream_executor.h"
 #endif  // GOOGLE_CUDA