Skip to content
体验新版
项目
组织
正在加载...
登录
切换导航
打开侧边栏
PaddlePaddle
Paddle
提交
3daf5185
P
Paddle
项目概览
PaddlePaddle
/
Paddle
大约 1 年 前同步成功
通知
2299
Star
20931
Fork
5422
代码
文件
提交
分支
Tags
贡献者
分支图
Diff
Issue
1423
列表
看板
标记
里程碑
合并请求
543
Wiki
0
Wiki
分析
仓库
DevOps
项目成员
Pages
P
Paddle
项目概览
项目概览
详情
发布
仓库
仓库
文件
提交
分支
标签
贡献者
分支图
比较
Issue
1,423
Issue
1,423
列表
看板
标记
里程碑
合并请求
543
合并请求
543
Pages
分析
分析
仓库分析
DevOps
Wiki
0
Wiki
成员
成员
收起侧边栏
关闭侧边栏
动态
分支图
创建新Issue
提交
Issue看板
未验证
提交
3daf5185
编写于
11月 23, 2022
作者:
W
Wilber
提交者:
GitHub
11月 23, 2022
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
add map_depthwise_conv_to_conv pass (#47955)
上级
67204c18
变更
4
隐藏空白更改
内联
并排
Showing
4 changed file
with
102 addition
and
2 deletion
+102
-2
paddle/fluid/framework/ir/CMakeLists.txt
paddle/fluid/framework/ir/CMakeLists.txt
+1
-0
paddle/fluid/framework/ir/map_depthwise_conv_to_conv_pass.cc
paddle/fluid/framework/ir/map_depthwise_conv_to_conv_pass.cc
+61
-0
paddle/fluid/framework/ir/map_depthwise_conv_to_conv_pass.h
paddle/fluid/framework/ir/map_depthwise_conv_to_conv_pass.h
+36
-0
paddle/fluid/inference/api/paddle_pass_builder.cc
paddle/fluid/inference/api/paddle_pass_builder.cc
+4
-2
未找到文件。
paddle/fluid/framework/ir/CMakeLists.txt
浏览文件 @
3daf5185
...
@@ -76,6 +76,7 @@ pass_library(embedding_fc_lstm_fuse_pass inference)
...
@@ -76,6 +76,7 @@ pass_library(embedding_fc_lstm_fuse_pass inference)
pass_library
(
fc_gru_fuse_pass inference
)
pass_library
(
fc_gru_fuse_pass inference
)
pass_library
(
seq_concat_fc_fuse_pass inference
)
pass_library
(
seq_concat_fc_fuse_pass inference
)
pass_library
(
multi_batch_merge_pass base
)
pass_library
(
multi_batch_merge_pass base
)
pass_library
(
map_depthwise_conv_to_conv_pass inference
)
pass_library
(
conv_bn_fuse_pass inference
)
pass_library
(
conv_bn_fuse_pass inference
)
pass_library
(
seqconv_eltadd_relu_fuse_pass inference
)
pass_library
(
seqconv_eltadd_relu_fuse_pass inference
)
pass_library
(
seqpool_concat_fuse_pass inference
)
pass_library
(
seqpool_concat_fuse_pass inference
)
...
...
paddle/fluid/framework/ir/map_depthwise_conv_to_conv_pass.cc
0 → 100644
浏览文件 @
3daf5185
// Copyright (c) 2018 PaddlePaddle Authors. 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.
#include "paddle/fluid/framework/ir/map_depthwise_conv_to_conv_pass.h"
#include <string>
#include "paddle/fluid/framework/op_version_registry.h"
#include "paddle/fluid/platform/enforce.h"
namespace
paddle
{
namespace
framework
{
namespace
ir
{
void
MapDepthwiseConv2ConvPass
::
ApplyImpl
(
ir
::
Graph
*
graph
)
const
{
PADDLE_ENFORCE_NOT_NULL
(
graph
,
platform
::
errors
::
InvalidArgument
(
"Graph cannot be nullptr."
));
FusePassBase
::
Init
(
"map_depthwise_conv_to_conv_pass"
,
graph
);
int
found_count
=
0
;
std
::
unordered_map
<
std
::
string
,
std
::
string
>
replaced_map
{
{
"depthwise_conv2d"
,
"conv2d"
},
};
auto
nodes
=
graph
->
Nodes
();
for
(
auto
&
node
:
nodes
)
{
if
(
!
node
->
IsOp
())
continue
;
auto
*
op_desc
=
node
->
Op
();
std
::
string
op_type
=
op_desc
->
Type
();
if
(
!
replaced_map
.
count
(
op_type
))
continue
;
op_desc
->
SetType
(
replaced_map
[
op_type
]);
op_desc
->
Flush
();
++
found_count
;
}
AddStatis
(
found_count
);
}
}
// namespace ir
}
// namespace framework
}
// namespace paddle
REGISTER_PASS
(
map_depthwise_conv_to_conv_pass
,
paddle
::
framework
::
ir
::
MapDepthwiseConv2ConvPass
);
REGISTER_PASS_CAPABILITY
(
map_depthwise_conv_to_conv_pass
)
.
AddCombination
(
paddle
::
framework
::
compatible
::
OpVersionComparatorCombination
()
.
LE
(
"depthwise_conv2d"
,
1
)
.
LE
(
"conv2d"
,
1
));
paddle/fluid/framework/ir/map_depthwise_conv_to_conv_pass.h
0 → 100644
浏览文件 @
3daf5185
// Copyright (c) 2018 PaddlePaddle Authors. 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.
#pragma once
#include <string>
#include "paddle/fluid/framework/ir/fuse_pass_base.h"
namespace
paddle
{
namespace
framework
{
namespace
ir
{
class
MapDepthwiseConv2ConvPass
:
public
FusePassBase
{
public:
MapDepthwiseConv2ConvPass
()
=
default
;
virtual
~
MapDepthwiseConv2ConvPass
()
=
default
;
protected:
void
ApplyImpl
(
Graph
*
graph
)
const
override
;
};
}
// namespace ir
}
// namespace framework
}
// namespace paddle
paddle/fluid/inference/api/paddle_pass_builder.cc
100755 → 100644
浏览文件 @
3daf5185
...
@@ -165,6 +165,7 @@ const std::vector<std::string> kLiteSubgraphPasses({
...
@@ -165,6 +165,7 @@ const std::vector<std::string> kLiteSubgraphPasses({
// running errors. After fusion operator supports low precision, delete this.
// running errors. After fusion operator supports low precision, delete this.
const
std
::
vector
<
std
::
string
>
kGpuLowerPrecisionPasses
{
const
std
::
vector
<
std
::
string
>
kGpuLowerPrecisionPasses
{
"simplify_with_basic_ops_pass"
,
"simplify_with_basic_ops_pass"
,
"map_depthwise_conv_to_conv_pass"
,
"conv_bn_fuse_pass"
,
"conv_bn_fuse_pass"
,
"conv_eltwiseadd_bn_fuse_pass"
,
"conv_eltwiseadd_bn_fuse_pass"
,
"conv_elementwise_add_act_fuse_pass"
,
"conv_elementwise_add_act_fuse_pass"
,
...
@@ -202,8 +203,9 @@ const std::vector<std::string> kTrtLowerPrecisionPasses{
...
@@ -202,8 +203,9 @@ const std::vector<std::string> kTrtLowerPrecisionPasses{
GpuPassStrategy
::
GpuPassStrategy
()
:
PassStrategy
({})
{
GpuPassStrategy
::
GpuPassStrategy
()
:
PassStrategy
({})
{
passes_
.
assign
({
passes_
.
assign
({
// "identity_scale_op_clean_pass", //
// "identity_scale_op_clean_pass", //
"is_test_pass"
,
//
"is_test_pass"
,
//
"simplify_with_basic_ops_pass"
,
//
"simplify_with_basic_ops_pass"
,
//
"map_depthwise_conv_to_conv_pass"
,
"conv_bn_fuse_pass"
,
//
"conv_bn_fuse_pass"
,
//
"conv_eltwiseadd_bn_fuse_pass"
,
//
"conv_eltwiseadd_bn_fuse_pass"
,
//
"embedding_eltwise_layernorm_fuse_pass"
,
//
"embedding_eltwise_layernorm_fuse_pass"
,
//
...
...
编辑
预览
Markdown
is supported
0%
请重试
或
添加新附件
.
添加附件
取消
You are about to add
0
people
to the discussion. Proceed with caution.
先完成此消息的编辑!
取消
想要评论请
注册
或
登录