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
af1568e6
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
af1568e6
编写于
9月 27, 2020
作者:
Y
yukavio
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
fix generate_proposals and affine grid error info
上级
6b727e08
变更
2
隐藏空白更改
内联
并排
Showing
2 changed file
with
18 addition
and
10 deletion
+18
-10
paddle/fluid/operators/affine_grid_cudnn_op.cu.cc
paddle/fluid/operators/affine_grid_cudnn_op.cu.cc
+18
-8
paddle/fluid/operators/detection/generate_proposals_op.cu
paddle/fluid/operators/detection/generate_proposals_op.cu
+0
-2
未找到文件。
paddle/fluid/operators/affine_grid_cudnn_op.cu.cc
浏览文件 @
af1568e6
...
@@ -26,8 +26,11 @@ template <typename T>
...
@@ -26,8 +26,11 @@ template <typename T>
class
CUDNNAffineGridOpKernel
:
public
framework
::
OpKernel
<
T
>
{
class
CUDNNAffineGridOpKernel
:
public
framework
::
OpKernel
<
T
>
{
public:
public:
void
Compute
(
const
framework
::
ExecutionContext
&
ctx
)
const
override
{
void
Compute
(
const
framework
::
ExecutionContext
&
ctx
)
const
override
{
PADDLE_ENFORCE
(
platform
::
is_gpu_place
(
ctx
.
GetPlace
()),
PADDLE_ENFORCE_EQ
(
platform
::
is_gpu_place
(
ctx
.
GetPlace
()),
true
,
"It must use CUDAPlace."
);
platform
::
errors
::
InvalidArgument
(
"Only "
"support for CUDAPlace.Please switch "
"your context from CPUPlace to "
"CUDAPlace or update your cudnn."
);
auto
&
dev_ctx
=
ctx
.
template
device_context
<
platform
::
CUDADeviceContext
>();
auto
&
dev_ctx
=
ctx
.
template
device_context
<
platform
::
CUDADeviceContext
>();
auto
handle
=
dev_ctx
.
cudnn_handle
();
auto
handle
=
dev_ctx
.
cudnn_handle
();
auto
*
theta
=
ctx
.
Input
<
Tensor
>
(
"Theta"
);
auto
*
theta
=
ctx
.
Input
<
Tensor
>
(
"Theta"
);
...
@@ -56,8 +59,10 @@ class CUDNNAffineGridOpKernel : public framework::OpKernel<T> {
...
@@ -56,8 +59,10 @@ class CUDNNAffineGridOpKernel : public framework::OpKernel<T> {
cudnnSpatialTransformerDescriptor_t
cudnn_st_desc
=
cudnnSpatialTransformerDescriptor_t
cudnn_st_desc
=
st_desc
.
descriptor
<
T
>
(
4
,
h_size_data
);
st_desc
.
descriptor
<
T
>
(
4
,
h_size_data
);
PADDLE_ENFORCE
(
platform
::
dynload
::
cudnnSpatialTfGridGeneratorForward
(
PADDLE_ENFORCE_EQ
(
platform
::
dynload
::
cudnnSpatialTfGridGeneratorForward
(
handle
,
cudnn_st_desc
,
theta_data
,
output_data
));
handle
,
cudnn_st_desc
,
theta_data
,
output_data
),
true
,
platform
::
errors
::
Fatal
(
"Some errors has occurred "
"during forward computation in cudnn."
));
}
}
};
};
...
@@ -65,8 +70,11 @@ template <typename T>
...
@@ -65,8 +70,11 @@ template <typename T>
class
CUDNNAffineGridGradOpKernel
:
public
framework
::
OpKernel
<
T
>
{
class
CUDNNAffineGridGradOpKernel
:
public
framework
::
OpKernel
<
T
>
{
public:
public:
void
Compute
(
const
framework
::
ExecutionContext
&
ctx
)
const
override
{
void
Compute
(
const
framework
::
ExecutionContext
&
ctx
)
const
override
{
PADDLE_ENFORCE
(
platform
::
is_gpu_place
(
ctx
.
GetPlace
()),
PADDLE_ENFORCE_EQ
(
platform
::
is_gpu_place
(
ctx
.
GetPlace
()),
true
,
"It must use CUDAPlace."
);
platform
::
errors
::
InvalidArgument
(
"Only "
"support for CUDAPlace. Please switch "
"your context from CPUPlace to "
"CUDAPlace or update your cudnn."
);
auto
&
dev_ctx
=
ctx
.
template
device_context
<
platform
::
CUDADeviceContext
>();
auto
&
dev_ctx
=
ctx
.
template
device_context
<
platform
::
CUDADeviceContext
>();
auto
handle
=
dev_ctx
.
cudnn_handle
();
auto
handle
=
dev_ctx
.
cudnn_handle
();
auto
output_grad
=
ctx
.
Input
<
Tensor
>
(
framework
::
GradVarName
(
"Output"
));
auto
output_grad
=
ctx
.
Input
<
Tensor
>
(
framework
::
GradVarName
(
"Output"
));
...
@@ -95,8 +103,10 @@ class CUDNNAffineGridGradOpKernel : public framework::OpKernel<T> {
...
@@ -95,8 +103,10 @@ class CUDNNAffineGridGradOpKernel : public framework::OpKernel<T> {
const
T
*
output_grad_data
=
output_grad
->
data
<
T
>
();
const
T
*
output_grad_data
=
output_grad
->
data
<
T
>
();
T
*
theta_grad_data
=
theta_grad
->
mutable_data
<
T
>
(
ctx
.
GetPlace
());
T
*
theta_grad_data
=
theta_grad
->
mutable_data
<
T
>
(
ctx
.
GetPlace
());
PADDLE_ENFORCE
(
platform
::
dynload
::
cudnnSpatialTfGridGeneratorBackward
(
PADDLE_ENFORCE_EQ
(
platform
::
dynload
::
cudnnSpatialTfGridGeneratorBackward
(
handle
,
cudnn_st_desc
,
output_grad_data
,
theta_grad_data
));
handle
,
cudnn_st_desc
,
output_grad_data
,
theta_grad_data
),
true
,
"Some errors "
"has occurred during forward computation in cudnn;"
);
}
}
};
};
...
...
paddle/fluid/operators/detection/generate_proposals_op.cu
浏览文件 @
af1568e6
...
@@ -247,8 +247,6 @@ static void NMS(const platform::CUDADeviceContext &ctx, const Tensor &proposals,
...
@@ -247,8 +247,6 @@ static void NMS(const platform::CUDADeviceContext &ctx, const Tensor &proposals,
const
Tensor
&
sorted_indices
,
const
T
nms_threshold
,
const
Tensor
&
sorted_indices
,
const
T
nms_threshold
,
Tensor
*
keep_out
)
{
Tensor
*
keep_out
)
{
int
boxes_num
=
proposals
.
dims
()[
0
];
int
boxes_num
=
proposals
.
dims
()[
0
];
PADDLE_ENFORCE_EQ
(
boxes_num
,
sorted_indices
.
dims
()[
0
]);
const
int
col_blocks
=
DIVUP
(
boxes_num
,
kThreadsPerBlock
);
const
int
col_blocks
=
DIVUP
(
boxes_num
,
kThreadsPerBlock
);
dim3
blocks
(
DIVUP
(
boxes_num
,
kThreadsPerBlock
),
dim3
blocks
(
DIVUP
(
boxes_num
,
kThreadsPerBlock
),
DIVUP
(
boxes_num
,
kThreadsPerBlock
));
DIVUP
(
boxes_num
,
kThreadsPerBlock
));
...
...
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