1 | /* Copyright 2017 The TensorFlow Authors. All Rights Reserved. |
2 | |
3 | Licensed under the Apache License, Version 2.0 (the "License"); |
4 | you may not use this file except in compliance with the License. |
5 | You may obtain a copy of the License at |
6 | |
7 | http://www.apache.org/licenses/LICENSE-2.0 |
8 | |
9 | Unless required by applicable law or agreed to in writing, software |
10 | distributed under the License is distributed on an "AS IS" BASIS, |
11 | WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
12 | See the License for the specific language governing permissions and |
13 | limitations under the License. |
14 | ==============================================================================*/ |
15 | |
16 | // See docs in ../ops/array_ops.cc. |
17 | #include "tensorflow/core/kernels/snapshot_op.h" |
18 | |
19 | #include "tensorflow/core/framework/op_kernel.h" |
20 | #include "tensorflow/core/framework/register_types.h" |
21 | #include "tensorflow/core/framework/types.h" |
22 | |
23 | namespace tensorflow { |
24 | typedef Eigen::ThreadPoolDevice CPUDevice; |
25 | typedef Eigen::GpuDevice GPUDevice; |
26 | |
27 | template <typename Device, typename Scalar> |
28 | class SnapshotOp : public OpKernel { |
29 | public: |
30 | explicit SnapshotOp(OpKernelConstruction* context) : OpKernel(context) {} |
31 | |
32 | void Compute(OpKernelContext* context) override { |
33 | const Tensor& input = context->input(0); |
34 | Tensor* output = nullptr; |
35 | // Try to use buffer forwarding to avoid an explicit copy. |
36 | OP_REQUIRES_OK(context, context->forward_input_or_allocate_output( |
37 | {0}, 0, input.shape(), &output)); |
38 | if (!output->SharesBufferWith(input)) { |
39 | functor::Snapshot<Device, Scalar> functor; |
40 | functor(context->eigen_device<Device>(), input.flat<Scalar>(), |
41 | output->flat<Scalar>()); |
42 | } |
43 | } |
44 | }; |
45 | |
46 | #define REGISTER_KERNEL(TYPE) \ |
47 | REGISTER_KERNEL_BUILDER( \ |
48 | Name("Snapshot").Device(DEVICE_CPU).TypeConstraint<TYPE>("T"), \ |
49 | SnapshotOp<CPUDevice, TYPE>); |
50 | |
51 | TF_CALL_POD_TYPES(REGISTER_KERNEL); |
52 | #undef REGISTER_KERNEL |
53 | |
54 | #if GOOGLE_CUDA || TENSORFLOW_USE_ROCM |
55 | #define REGISTER_KERNEL(TYPE) \ |
56 | REGISTER_KERNEL_BUILDER( \ |
57 | Name("Snapshot").Device(DEVICE_GPU).TypeConstraint<TYPE>("T"), \ |
58 | SnapshotOp<GPUDevice, TYPE>); |
59 | |
60 | TF_CALL_POD_TYPES(REGISTER_KERNEL); |
61 | #undef REGISTER_KERNEL |
62 | #endif |
63 | |
64 | |
65 | } // namespace tensorflow |
66 | |