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
f9681459
P
Paddle
项目概览
PaddlePaddle
/
Paddle
12 个月 前同步成功
通知
2292
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
体验新版 GitCode,发现更多精彩内容 >>
提交
f9681459
编写于
10月 17, 2017
作者:
Q
qijun
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
fix gpu build error
上级
ab8cc401
变更
5
隐藏空白更改
内联
并排
Showing
5 changed file
with
23 addition
and
8 deletion
+23
-8
paddle/operators/sgd_op.cc
paddle/operators/sgd_op.cc
+1
-1
paddle/operators/sgd_op.cu
paddle/operators/sgd_op.cu
+3
-3
paddle/operators/sgd_op.h
paddle/operators/sgd_op.h
+1
-1
paddle/pybind/pybind.cc
paddle/pybind/pybind.cc
+9
-1
python/paddle/v2/framework/tests/test_sgd_op.py
python/paddle/v2/framework/tests/test_sgd_op.py
+9
-2
未找到文件。
paddle/operators/sgd_op.cc
浏览文件 @
f9681459
...
...
@@ -61,7 +61,7 @@ param_out = param - learning_rate * grad;
template
<
typename
T
>
struct
SparseSGDFunctor
<
platform
::
CPUPlace
,
T
>
{
void
operator
()(
const
platform
::
DeviceContext
&
c
tx
,
void
operator
()(
const
platform
::
DeviceContext
&
c
ontext
,
const
framework
::
SelectedRows
&
input
,
const
framework
::
Tensor
&
learning_rate
,
framework
::
Tensor
*
output
)
{
...
...
paddle/operators/sgd_op.cu
浏览文件 @
f9681459
...
...
@@ -34,15 +34,15 @@ __global__ void SparseSGDFunctorKernel(const T* selected_rows,
for
(
int
index
=
tid
;
index
<
row_numel
;
index
+=
block_size
)
{
// Since index in rows of SelectedRows can be duplicate, we have to use
// Atomic Operation to avoid concurrent write error.
paddle
::
platform
::
CudaAtomic
Sub
(
tensor_out
+
index
,
learning_rate
[
0
]
*
selected_rows
[
index
]);
paddle
::
platform
::
CudaAtomic
Add
(
tensor_out
+
index
,
-
1.0
*
learning_rate
[
0
]
*
selected_rows
[
index
]);
}
}
}
// namespace
template
<
typename
T
>
struct
SparseSGDFunctor
<
platform
::
GPUPlace
,
T
>
{
void
operator
()(
const
platform
::
DeviceContext
&
c
tx
,
void
operator
()(
const
platform
::
DeviceContext
&
c
ontext
,
const
framework
::
SelectedRows
&
input
,
const
framework
::
Tensor
&
learning_rate
,
framework
::
Tensor
*
output
)
{
...
...
paddle/operators/sgd_op.h
浏览文件 @
f9681459
...
...
@@ -22,7 +22,7 @@ namespace operators {
template
<
typename
Place
,
typename
T
>
struct
SparseSGDFunctor
{
void
operator
()(
const
platform
::
DeviceContext
&
c
tx
,
void
operator
()(
const
platform
::
DeviceContext
&
c
ontext
,
const
framework
::
SelectedRows
&
input
,
const
framework
::
Tensor
&
learning_rate
,
framework
::
Tensor
*
output
);
...
...
paddle/pybind/pybind.cc
浏览文件 @
f9681459
...
...
@@ -153,7 +153,15 @@ PYBIND11_PLUGIN(core) {
py
::
return_value_policy
::
reference
)
.
def
(
"set_height"
,
&
SelectedRows
::
set_height
)
.
def
(
"height"
,
&
SelectedRows
::
height
)
.
def
(
"set_rows"
,
&
SelectedRows
::
set_rows
)
.
def
(
"set_rows"
,
[](
SelectedRows
&
self
,
std
::
vector
<
int64_t
>
rows
)
{
#ifndef PADDLE_WITH_CUDA
self
.
set_rows
(
rows
);
#else
Vector
<
int64_t
>
new_rows
(
rows
);
self
.
set_rows
(
new_rows
);
#endif
})
.
def
(
"rows"
,
[](
SelectedRows
&
self
)
{
#ifndef PADDLE_WITH_CUDA
return
self
.
rows
();
...
...
python/paddle/v2/framework/tests/test_sgd_op.py
浏览文件 @
f9681459
...
...
@@ -20,11 +20,10 @@ class TestSGDOp(OpTest):
class
TestSparseSGDOp
(
unittest
.
TestCase
):
def
test_sparse_sgd
(
self
):
def
check_with_place
(
self
,
place
):
scope
=
core
.
Scope
()
# create and initialize Grad Variable
place
=
core
.
CPUPlace
()
height
=
10
rows
=
[
0
,
4
,
7
]
row_numel
=
12
...
...
@@ -35,6 +34,7 @@ class TestSparseSGDOp(unittest.TestCase):
np_array
=
np
.
ones
((
len
(
rows
),
row_numel
)).
astype
(
"float32"
)
np_array
[
0
,
0
]
=
2.0
np_array
[
2
,
8
]
=
4.0
grad_tensor
=
grad_selected_rows
.
get_tensor
()
grad_tensor
.
set
(
np_array
,
place
)
...
...
@@ -76,6 +76,13 @@ class TestSparseSGDOp(unittest.TestCase):
# rows[2] = 7, 5.0 - 2.0 * 4.0
self
.
assertAlmostEqual
(
-
3.0
,
result_array
[
rows
[
2
],
8
])
def
test_sparse_sgd
(
self
):
places
=
[
core
.
CPUPlace
()]
if
core
.
is_compile_gpu
():
places
.
append
(
core
.
GPUPlace
(
0
))
for
place
in
places
:
self
.
check_with_place
(
place
)
if
__name__
==
"__main__"
:
unittest
.
main
()
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