1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
|
/* Copyright 2015 Google Inc. 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.
==============================================================================*/
#if !GOOGLE_CUDA
#error This file must only be included when building with Cuda support
#endif
#ifndef TENSORFLOW_COMMON_RUNTIME_GPU_GPU_DEVICE_H_
#define TENSORFLOW_COMMON_RUNTIME_GPU_GPU_DEVICE_H_
#include "tensorflow/stream_executor/stream.h"
#include "third_party/eigen3/unsupported/Eigen/CXX11/Tensor"
#include "tensorflow/core/common_runtime/device_factory.h"
#include "tensorflow/core/common_runtime/gpu/gpu_event_mgr.h"
#include "tensorflow/core/common_runtime/gpu_device_context.h"
#include "tensorflow/core/common_runtime/local_device.h"
#include "tensorflow/core/framework/allocator.h"
#include "tensorflow/core/framework/device_base.h"
#include "tensorflow/core/framework/op_kernel.h"
#include "tensorflow/core/platform/port.h"
#include "tensorflow/core/public/session_options.h"
#include "tensorflow/core/public/status.h"
#include "tensorflow/core/public/tensor.h"
namespace tensorflow {
class EigenAllocator;
class BaseGPUDevice : public LocalDevice {
public:
BaseGPUDevice(const SessionOptions& options, const string& name,
Bytes memory_limit, BusAdjacency bus_adjacency, int gpu_id,
const string& physical_device_desc, Allocator* gpu_allocator,
Allocator* cpu_allocator);
~BaseGPUDevice() override;
// GPU devices require the Op Compute method to save a reference to
// any temporary tensors that are allocated until the Op execution
// completes.
bool SaveTemporaryTensors() const override { return true; }
Status FillContextMap(const Graph* graph,
DeviceContextMap* device_context_map);
void Compute(OpKernel* op_kernel, OpKernelContext* context) override;
Status Sync() override;
void ComputeAsync(AsyncOpKernel* op_kernel, OpKernelContext* context,
AsyncOpKernel::DoneCallback done) override;
Status MakeTensorFromProto(const TensorProto& tensor_proto,
const AllocatorAttributes alloc_attrs,
Tensor* tensor) override;
// The caller owns the returned device.
const PerOpGpuDevice* MakeGpuDevice(DeviceContext* dc,
Allocator* allocator) override;
protected:
Allocator* gpu_allocator_; // not owned
Allocator* cpu_allocator_; // not owned
private:
std::vector<gpu::Stream*> streams_;
std::vector<GPUDeviceContext*> device_contexts_;
GpuDeviceInfo* gpu_device_info_ = nullptr;
mutex trace_mu_;
int gpu_id_ = -1;
std::unique_ptr<EventMgr> em_;
const PerOpGpuDevice* NewDevice(int stream_id, Allocator* allocator);
};
class BaseGPUDeviceFactory : public DeviceFactory {
public:
void CreateDevices(const SessionOptions& options, const string& name_prefix,
std::vector<Device*>* devices) override;
private:
LocalDevice* CreateGPUDevice(const SessionOptions& options,
const string& name, int gpu_id);
virtual LocalDevice* CreateGPUDevice(const SessionOptions& options,
const string& name, Bytes memory_limit,
BusAdjacency bus_adjacency, int gpu_id,
const string& physical_device_desc,
Allocator* gpu_allocator,
Allocator* cpu_allocator) = 0;
void GetValidDeviceIds(std::vector<int>* ids);
};
} // namespace tensorflow
#endif // TENSORFLOW_COMMON_RUNTIME_GPU_GPU_DEVICE_H_
|
