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
|
/**
* Copyright (c) 2016-present, Facebook, Inc.
*
* 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.
*/
#pragma once
#include "caffe2/operators/batch_matmul_op.h"
#include "caffe2/quantization/server/dnnlowp_op.h"
#include "fbgemm/Fbgemm.h"
namespace caffe2 {
template <typename T>
class BatchMatMulDNNLowPOp final
: public DNNLowPOp<T, BatchMatMulOp<CPUContext>> {
public:
BatchMatMulDNNLowPOp(const OperatorDef& operator_def, Workspace* ws);
bool RunOnDevice() override;
USE_OPERATOR_FUNCTIONS(CPUContext);
USE_DNNLOWP_OPERATOR_BASE_FUNCTIONS(T, BatchMatMulOp<CPUContext>);
private:
bool trans_a_;
bool trans_b_;
bool broadcast_{false};
bool is_B_constant_{false};
std::vector<std::int8_t> B_quantized_;
std::vector<std::unique_ptr<fbgemm::PackBMatrix<std::int8_t>>> Bq_packed_;
std::vector<std::uint8_t> A_pack_buf_;
std::vector<std::int32_t> row_offsets_, column_offsets_;
std::vector<dnnlowp::TensorQuantizationParams> B_qparams_;
std::vector<dnnlowp::RequantizationParams> requantization_params_;
std::vector<std::int32_t> Y_int32_;
bool first_invocation_{true};
}; // BatchMatMulDNNLowPOp
} // namespace caffe2
|
