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
BaiXuePrincess
Paddle
提交
690cd1f7
P
Paddle
项目概览
BaiXuePrincess
/
Paddle
与 Fork 源项目一致
Fork自
PaddlePaddle / Paddle
通知
1
Star
1
Fork
0
代码
文件
提交
分支
Tags
贡献者
分支图
Diff
Issue
0
列表
看板
标记
里程碑
合并请求
0
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
0
Issue
0
列表
看板
标记
里程碑
合并请求
0
合并请求
0
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
提交
690cd1f7
编写于
4月 14, 2018
作者:
C
chengduoZH
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
refine gather and broadcast
上级
494c262a
变更
2
隐藏空白更改
内联
并排
Showing
2 changed file
with
17 addition
and
12 deletion
+17
-12
paddle/fluid/framework/details/broadcast_op_handle.cc
paddle/fluid/framework/details/broadcast_op_handle.cc
+5
-3
paddle/fluid/framework/details/gather_op_handle.cc
paddle/fluid/framework/details/gather_op_handle.cc
+12
-9
未找到文件。
paddle/fluid/framework/details/broadcast_op_handle.cc
浏览文件 @
690cd1f7
...
...
@@ -99,9 +99,11 @@ void BroadcastOpHandle::RunImpl() {
PADDLE_THROW
(
"Var should be LoDTensor or SelectedRows."
);
}
Tensor
*
out_tensor
=
GetTensorFromVar
(
out_var
);
paddle
::
framework
::
TensorCopy
(
*
in_tensor
,
out_p
,
*
(
dev_ctxes_
[
in_place
]),
out_tensor
);
auto
dev_ctx
=
dev_ctxes_
[
out_p
];
RunAndRecordEvent
(
out_p
,
[
in_tensor
,
out_var
,
dev_ctx
,
out_p
]
{
Tensor
*
out_tensor
=
GetTensorFromVar
(
out_var
);
paddle
::
framework
::
TensorCopy
(
*
in_tensor
,
out_p
,
*
(
dev_ctx
),
out_tensor
);
});
}
}
...
...
paddle/fluid/framework/details/gather_op_handle.cc
浏览文件 @
690cd1f7
...
...
@@ -84,7 +84,7 @@ void GatherOpHandle::RunImpl() {
"The type of input is not consistent."
);
PADDLE_ENFORCE_EQ
(
pre_in
.
height
(),
in_sr
.
height
(),
"The height of inputs is not consistent."
);
PADDLE_ENFORCE_EQ
(
pre_in
.
GetCompleteDims
(),
in_sr
.
GetCompleteDims
(),
,
PADDLE_ENFORCE_EQ
(
pre_in
.
GetCompleteDims
(),
in_sr
.
GetCompleteDims
(),
"The dims of inputs is not consistent."
);
auto
in_sr_rows
=
in_sr
.
rows
();
...
...
@@ -110,14 +110,17 @@ void GatherOpHandle::RunImpl() {
Tensor
*
out_tensor
=
out
->
mutable_value
();
// copy
int
s
=
0
,
e
=
0
;
for
(
size_t
j
=
0
;
j
<
in_tensors
.
size
();
++
j
)
{
e
+=
in_tensors
[
j
].
dims
()[
0
];
auto
sub_out
=
out_tensor
->
Slice
(
s
,
e
);
paddle
::
framework
::
TensorCopy
(
in_tensors
[
j
],
out_place
,
*
(
dev_ctxes_
[
in_places
[
j
]]),
&
sub_out
);
s
=
e
;
}
auto
dev_ctx
=
dev_ctxes_
[
out_place
];
RunAndRecordEvent
(
out_place
,
[
in_tensors
,
out_var
,
dev_ctx
,
out_place
]
{
int
s
=
0
,
e
=
0
;
for
(
size_t
j
=
0
;
j
<
in_tensors
.
size
();
++
j
)
{
e
+=
in_tensors
[
j
].
dims
()[
0
];
auto
sub_out
=
out_tensor
->
Slice
(
s
,
e
);
paddle
::
framework
::
TensorCopy
(
in_tensors
[
j
],
out_place
,
*
(
dev_ctx
),
&
sub_out
);
s
=
e
;
}
});
}
std
::
string
GatherOpHandle
::
Name
()
const
{
return
"gather"
;
}
...
...
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