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
wzpei9
Paddle
提交
7307b439
P
Paddle
项目概览
wzpei9
/
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
1
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
1
Issue
1
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
7307b439
编写于
8月 09, 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
a821fec1
变更
5
显示空白变更内容
内联
并排
Showing
5 changed file
with
22 addition
and
8 deletion
+22
-8
CMakeLists.txt
CMakeLists.txt
+2
-2
paddle/operators/math/math_function.cu
paddle/operators/math/math_function.cu
+4
-2
paddle/operators/math/math_function.h
paddle/operators/math/math_function.h
+15
-1
paddle/operators/mul_op.cu
paddle/operators/mul_op.cu
+1
-0
paddle/operators/mul_op.h
paddle/operators/mul_op.h
+0
-3
未找到文件。
CMakeLists.txt
浏览文件 @
7307b439
...
...
@@ -36,8 +36,8 @@ include(simd)
################################ Configurations #######################################
option
(
WITH_GPU
"Compile PaddlePaddle with NVIDIA GPU"
${
CUDA_FOUND
}
)
option
(
WITH_AVX
"Compile PaddlePaddle with AVX intrinsics"
${
AVX_FOUND
}
)
option
(
WITH_MKLDNN
"Compile PaddlePaddle with mkl-dnn support."
${
AVX_FOUND
}
)
option
(
WITH_MKLML
"Compile PaddlePaddle with mklml package."
${
AVX_FOUND
}
)
option
(
WITH_MKLDNN
"Compile PaddlePaddle with mkl-dnn support."
OFF
)
option
(
WITH_MKLML
"Compile PaddlePaddle with mklml package."
OFF
)
option
(
WITH_DSO
"Compile PaddlePaddle with dynamic linked CUDA"
ON
)
option
(
WITH_TESTING
"Compile PaddlePaddle with unit testing"
ON
)
option
(
WITH_SWIG_PY
"Compile PaddlePaddle with inference api"
ON
)
...
...
paddle/operators/math/math_function.cu
浏览文件 @
7307b439
...
...
@@ -13,7 +13,6 @@ See the License for the specific language governing permissions and
limitations under the License. */
#include "paddle/operators/math/math_function.h"
namespace
paddle
{
namespace
operators
{
namespace
math
{
...
...
@@ -26,6 +25,8 @@ void gemm<platform::GPUPlace, float>(
platform
::
DeviceContext
*
context
)
{
// Note that cublas follows fortran order, so the order is different from
// the cblas convention.
int
lda
=
(
transA
==
CblasNoTrans
)
?
K
:
M
;
int
ldb
=
(
transB
==
CblasNoTrans
)
?
N
:
K
;
cublasOperation_t
cuTransA
=
(
transA
==
CblasNoTrans
)
?
CUBLAS_OP_N
:
CUBLAS_OP_T
;
cublasOperation_t
cuTransB
=
...
...
@@ -44,6 +45,8 @@ void gemm<platform::GPUPlace, double>(
const
int
ldc
,
platform
::
DeviceContext
*
context
)
{
// Note that cublas follows fortran order, so the order is different from
// the cblas convention.
lda
=
(
transA
==
CblasNoTrans
)
?
K
:
M
;
ldb
=
(
transB
==
CblasNoTrans
)
?
N
:
K
;
cublasOperation_t
cuTransA
=
(
transA
==
CblasNoTrans
)
?
CUBLAS_OP_N
:
CUBLAS_OP_T
;
cublasOperation_t
cuTransB
=
...
...
@@ -118,7 +121,6 @@ void matmul<platform::GPUPlace, double>(const framework::Tensor& in1,
in1
.
data
<
double
>
(),
K
,
in2
.
data
<
double
>
(),
N
,
beta
,
out
->
data
<
double
>
(),
N
,
context
);
}
}
// namespace math
}
// namespace operators
}
// namespace paddle
paddle/operators/math/math_function.h
浏览文件 @
7307b439
...
...
@@ -37,6 +37,20 @@ extern "C" {
#include <lapacke.h>
#endif
#ifndef LAPACK_FOUND
extern
"C"
{
#include <cblas.h>
int
LAPACKE_sgetrf
(
int
matrix_layout
,
int
m
,
int
n
,
float
*
a
,
int
lda
,
int
*
ipiv
);
int
LAPACKE_dgetrf
(
int
matrix_layout
,
int
m
,
int
n
,
double
*
a
,
int
lda
,
int
*
ipiv
);
int
LAPACKE_sgetri
(
int
matrix_layout
,
int
n
,
float
*
a
,
int
lda
,
const
int
*
ipiv
);
int
LAPACKE_dgetri
(
int
matrix_layout
,
int
n
,
double
*
a
,
int
lda
,
const
int
*
ipiv
);
}
#endif
#include <cmath>
#include "paddle/framework/tensor.h"
#include "paddle/platform/device_context.h"
...
...
@@ -61,7 +75,7 @@ void gemm(const CBLAS_TRANSPOSE transA,
const
int
ldc
,
platform
::
DeviceContext
*
context
);
// matrix multiply with continous memory
// matrix multiply with contin
u
ous memory
template
<
typename
Place
,
typename
T
>
void
matmul
(
const
framework
::
Tensor
&
in1
,
bool
in1_T
,
...
...
paddle/operators/mul_op.cu
浏览文件 @
7307b439
...
...
@@ -15,4 +15,5 @@
#define EIGEN_USE_GPU
#include "paddle/operators/mul_op.h"
namespace
ops
=
paddle
::
operators
;
REGISTER_OP_GPU_KERNEL
(
mul
,
ops
::
MulKernel
<
paddle
::
platform
::
GPUPlace
,
float
>
);
paddle/operators/mul_op.h
浏览文件 @
7307b439
...
...
@@ -31,9 +31,6 @@ template <typename Place, typename T>
class
MulKernel
:
public
framework
::
OpKernel
{
public:
void
Compute
(
const
framework
::
ExecutionContext
&
context
)
const
override
{
Eigen
::
array
<
Eigen
::
IndexPair
<
Eigen
::
DenseIndex
>
,
1
>
dim_pair
=
{
{
Eigen
::
IndexPair
<
Eigen
::
DenseIndex
>
(
1
,
0
)}};
auto
input0
=
context
.
Input
<
Tensor
>
(
"X"
);
auto
input1
=
context
.
Input
<
Tensor
>
(
"Y"
);
auto
output
=
context
.
Output
<
Tensor
>
(
0
);
...
...
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