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
|
#ifndef CAFFE2_OPERATORS_ENSURE_CPU_OUTPUT_OP_H_
#define CAFFE2_OPERATORS_ENSURE_CPU_OUTPUT_OP_H_
#include "caffe2/core/context.h"
#include "caffe2/core/operator.h"
#include "caffe2/utils/math.h"
namespace caffe2 {
template <class Context>
class EnsureCPUOutputOp : public Operator<Context> {
public:
USE_OPERATOR_CONTEXT_FUNCTIONS;
template <class... Args>
explicit EnsureCPUOutputOp(Args&&... args)
: Operator<Context>(std::forward<Args>(args)...) {}
bool RunOnDevice() override {
if (this->InputIsTensorType(0, CPU)) {
return CopyWithContext<CPUContext>();
} else if (this->InputIsTensorType(0, Context::GetDeviceType())) {
// CUDA Context will go this branch
return CopyWithContext<Context>();
} else {
CAFFE_THROW(
"Unexpected Input Blob: ",
OperatorBase::Inputs().at(0)->meta().name());
}
return true;
}
private:
template <class InputContext>
bool CopyWithContext() {
// Output is always on CPU
auto* output = this->template Output<Tensor>(0, CPU);
auto& input =
this->template Input<Tensor>(0, InputContext::GetDeviceType());
output->ResizeLike(input);
context_.CopyItemsToCPU(
input.dtype(),
input.numel(),
input.raw_data(),
output->raw_mutable_data(input.dtype()));
context_.FinishDeviceComputation();
return true;
}
};
} // namespace caffe2
#endif // CAFFE2_OPERATORS_ENSURE_CPU_OUTPUT_OP_H_
|
