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
urhero
Paddle
提交
1afa9492
P
Paddle
项目概览
urhero
/
Paddle
与 Fork 源项目一致
Fork自
PaddlePaddle / Paddle
通知
1
Star
0
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
体验新版 GitCode,发现更多精彩内容 >>
提交
1afa9492
编写于
11月 26, 2018
作者:
P
peizhilin
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
Recover the profiler
上级
445fff24
变更
7
显示空白变更内容
内联
并排
Showing
7 changed file
with
10 addition
and
39 deletion
+10
-39
cmake/cuda.cmake
cmake/cuda.cmake
+5
-2
paddle/fluid/framework/lod_tensor.cc
paddle/fluid/framework/lod_tensor.cc
+1
-16
paddle/fluid/framework/lod_tensor_test.cc
paddle/fluid/framework/lod_tensor_test.cc
+0
-2
paddle/fluid/framework/operator.cc
paddle/fluid/framework/operator.cc
+4
-7
paddle/fluid/inference/api/analysis_predictor.cc
paddle/fluid/inference/api/analysis_predictor.cc
+0
-4
paddle/fluid/inference/api/api_impl.cc
paddle/fluid/inference/api/api_impl.cc
+0
-4
paddle/fluid/inference/tests/api/tester_helper.h
paddle/fluid/inference/tests/api/tester_helper.h
+0
-4
未找到文件。
cmake/cuda.cmake
浏览文件 @
1afa9492
...
...
@@ -199,10 +199,13 @@ elseif(CMAKE_BUILD_TYPE STREQUAL "MinSizeRel")
list
(
APPEND CUDA_NVCC_FLAGS
${
CMAKE_CXX_FLAGS_RELEASE
}
)
endif
()
else
(
NOT WIN32
)
list
(
APPEND CUDA_NVCC_FLAGS
"--compiler-options;/bigobj"
)
if
(
CMAKE_BUILD_TYPE STREQUAL
"Debug"
)
list
(
APPEND CUDA_NVCC_FLAGS
"-g -G --compiler-options;/bigobj"
)
list
(
APPEND CUDA_NVCC_FLAGS
"-g -G"
)
# match the cl's _ITERATOR_DEBUG_LEVEL
list
(
APPEND CUDA_NVCC_FLAGS
"-D_DEBUG"
)
elseif
(
CMAKE_BUILD_TYPE STREQUAL
"Release"
)
list
(
APPEND CUDA_NVCC_FLAGS
"-O3 -DNDEBUG
--compiler-options;/bigobj
"
)
list
(
APPEND CUDA_NVCC_FLAGS
"-O3 -DNDEBUG"
)
else
()
message
(
FATAL
"Windows only support Release or Debug build now. Please set visual studio build type to Release/Debug, x64 build."
)
endif
()
...
...
paddle/fluid/framework/lod_tensor.cc
浏览文件 @
1afa9492
...
...
@@ -26,10 +26,8 @@ limitations under the License. */
#include "paddle/fluid/memory/memcpy.h"
#include "paddle/fluid/memory/memory.h"
#if !defined(_WIN32)
#include "paddle/fluid/recordio/scanner.h"
#include "paddle/fluid/recordio/writer.h"
#endif // _WIN32
namespace
paddle
{
namespace
framework
{
...
...
@@ -305,7 +303,6 @@ void DeserializeFromStream(std::istream &is, LoDTensor *tensor,
TensorFromStream
(
is
,
static_cast
<
Tensor
*>
(
tensor
),
dev_ctx
);
}
#if !defined(_WIN32)
void
WriteToRecordIO
(
recordio
::
Writer
*
writer
,
const
std
::
vector
<
LoDTensor
>
&
tensor
,
const
platform
::
DeviceContext
&
dev_ctx
)
{
...
...
@@ -335,19 +332,7 @@ bool ReadFromRecordIO(recordio::Scanner *scanner,
return
true
;
}
#else
class
Writer
{};
class
Scanner
{};
void
WriteToRecordIO
(
recordio
::
Writer
*
writer
,
const
std
::
vector
<
LoDTensor
>
&
tensor
,
const
platform
::
DeviceContext
&
dev_ctx
)
{}
bool
ReadFromRecordIO
(
recordio
::
Scanner
*
scanner
,
const
platform
::
DeviceContext
&
dev_ctx
,
std
::
vector
<
LoDTensor
>
*
result_ptr
)
{
PADDLE_ENFORCE
(
"windows didn't supported recordio!."
);
return
true
;
}
#endif // _WIN32
std
::
vector
<
LoDTensor
>
LoDTensor
::
SplitLoDTensor
(
const
std
::
vector
<
platform
::
Place
>
places
)
const
{
check_memory_size
();
...
...
paddle/fluid/framework/lod_tensor_test.cc
浏览文件 @
1afa9492
...
...
@@ -274,7 +274,6 @@ TEST(LoD, ConvertToOffsetBasedLoD) {
EXPECT_EQ
(
offset_lod
,
expected
);
}
#if !defined(_WIN32)
template
<
typename
T
>
static
void
TestRecordIO
()
{
LoDTensor
tensor
;
...
...
@@ -321,7 +320,6 @@ TEST(LoDTensor, RecordIO) {
TestRecordIO
<
float
>
();
TestRecordIO
<
double
>
();
}
#endif // !defined(_WIN32)
}
// namespace framework
}
// namespace paddle
paddle/fluid/framework/operator.cc
浏览文件 @
1afa9492
...
...
@@ -153,17 +153,14 @@ void OperatorBase::Run(const Scope& scope, const platform::Place& place) {
#endif
}
// The profile has a process-wide mutex, results in serious performance issue
// in concurrency scenerio. Here use an `if` to fix this issue.
// Please not remove the `if`, ask @Superjomn if there are any concern.
#ifndef _WIN32
// The profile has a process-wide mutex, results in serious performance issue
// in concurrency scenerio. Here use an `if` to fix this issue.
// Please not remove the `if`, ask @Superjomn if there are any concern.
if
(
platform
::
IsProfileEnabled
())
{
platform
::
DeviceContextPool
&
pool
=
platform
::
DeviceContextPool
::
Instance
();
platform
::
RecordEvent
record_event
(
Type
(),
pool
.
Get
(
place
));
RunImpl
(
scope
,
place
);
}
else
#endif
{
}
else
{
RunImpl
(
scope
,
place
);
}
VLOG
(
30
)
<<
place
<<
" "
<<
DebugStringEx
(
&
scope
);
...
...
paddle/fluid/inference/api/analysis_predictor.cc
浏览文件 @
1afa9492
...
...
@@ -56,7 +56,6 @@ bool AnalysisPredictor::Init(
const
std
::
shared_ptr
<
framework
::
Scope
>
&
parent_scope
,
const
std
::
shared_ptr
<
framework
::
ProgramDesc
>
&
program
)
{
VLOG
(
30
)
<<
"Predictor::init()"
;
#if !defined(_WIN32)
if
(
FLAGS_profile
)
{
LOG
(
WARNING
)
<<
"Profiler is actived, might affect the performance"
;
LOG
(
INFO
)
<<
"You can turn off by set gflags '-profile false'"
;
...
...
@@ -64,7 +63,6 @@ bool AnalysisPredictor::Init(
:
platform
::
ProfilerState
::
kCPU
;
platform
::
EnableProfiler
(
tracking_device
);
}
#endif
// no matter with or without MKLDNN
paddle
::
platform
::
SetNumThreads
(
FLAGS_paddle_num_threads
);
...
...
@@ -501,12 +499,10 @@ bool AnalysisPredictor::LoadParameters() {
}
AnalysisPredictor
::~
AnalysisPredictor
()
{
#if !defined(_WIN32)
if
(
FLAGS_profile
)
{
platform
::
DisableProfiler
(
platform
::
EventSortingKey
::
kTotal
,
"./profile.log"
);
}
#endif
if
(
sub_scope_
)
{
scope_
->
DeleteScope
(
sub_scope_
);
}
...
...
paddle/fluid/inference/api/api_impl.cc
浏览文件 @
1afa9492
...
...
@@ -64,7 +64,6 @@ void NativePaddlePredictor::PrepareFeedFetch() {
bool
NativePaddlePredictor
::
Init
(
std
::
shared_ptr
<
framework
::
Scope
>
parent_scope
)
{
VLOG
(
3
)
<<
"Predictor::init()"
;
#if !defined(_WIN32)
if
(
FLAGS_profile
)
{
LOG
(
WARNING
)
<<
"Profiler is actived, might affect the performance"
;
LOG
(
INFO
)
<<
"You can turn off by set gflags '-profile false'"
;
...
...
@@ -73,7 +72,6 @@ bool NativePaddlePredictor::Init(
:
platform
::
ProfilerState
::
kCPU
;
platform
::
EnableProfiler
(
tracking_device
);
}
#endif
// no matter with or without MKLDNN
paddle
::
platform
::
SetNumThreads
(
FLAGS_paddle_num_threads
);
...
...
@@ -121,12 +119,10 @@ bool NativePaddlePredictor::Init(
}
NativePaddlePredictor
::~
NativePaddlePredictor
()
{
#if !defined(_WIN32)
if
(
FLAGS_profile
)
{
platform
::
DisableProfiler
(
platform
::
EventSortingKey
::
kTotal
,
"./profile.log"
);
}
#endif
if
(
sub_scope_
)
{
scope_
->
DeleteScope
(
sub_scope_
);
}
...
...
paddle/fluid/inference/tests/api/tester_helper.h
浏览文件 @
1afa9492
...
...
@@ -177,11 +177,9 @@ void TestOneThreadPrediction(
warmup_timer
.
tic
();
predictor
->
Run
(
inputs
[
0
],
outputs
,
batch_size
);
PrintTime
(
batch_size
,
1
,
1
,
0
,
warmup_timer
.
toc
(),
1
);
#if !defined(_WIN32)
if
(
FLAGS_profile
)
{
paddle
::
platform
::
ResetProfiler
();
}
#endif
}
LOG
(
INFO
)
<<
"Run "
<<
num_times
<<
" times..."
;
...
...
@@ -230,11 +228,9 @@ void TestMultiThreadPrediction(
warmup_timer
.
tic
();
predictor
->
Run
(
inputs
[
0
],
outputs
,
batch_size
);
PrintTime
(
batch_size
,
1
,
num_threads
,
tid
,
warmup_timer
.
toc
(),
1
);
#if !defined(_WIN32)
if
(
FLAGS_profile
)
{
paddle
::
platform
::
ResetProfiler
();
}
#endif
}
LOG
(
INFO
)
<<
"Thread "
<<
tid
<<
" run "
<<
num_times
<<
" times..."
;
...
...
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