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-Lite
提交
ff8c95d8
P
Paddle-Lite
项目概览
PaddlePaddle
/
Paddle-Lite
通知
331
Star
4
Fork
1
代码
文件
提交
分支
Tags
贡献者
分支图
Diff
Issue
271
列表
看板
标记
里程碑
合并请求
78
Wiki
0
Wiki
分析
仓库
DevOps
项目成员
Pages
P
Paddle-Lite
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
271
Issue
271
列表
看板
标记
里程碑
合并请求
78
合并请求
78
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
ff8c95d8
编写于
8月 07, 2020
作者:
C
chenjiaoAngel
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
fxi foormat test=develop
上级
7b282a0a
变更
4
展开全部
隐藏空白更改
内联
并排
Showing
4 changed file
with
774 addition
and
774 deletion
+774
-774
lite/backends/arm/math/conv3x3s1p01_depthwise_fp32.cc
lite/backends/arm/math/conv3x3s1p01_depthwise_fp32.cc
+2
-2
lite/backends/arm/math/conv3x3s1p01_depthwise_fp32_relu.cc
lite/backends/arm/math/conv3x3s1p01_depthwise_fp32_relu.cc
+322
-319
lite/backends/arm/math/conv3x3s2p01_depthwise_fp32.cc
lite/backends/arm/math/conv3x3s2p01_depthwise_fp32.cc
+78
-79
lite/backends/arm/math/conv3x3s2p01_depthwise_fp32_relu.cc
lite/backends/arm/math/conv3x3s2p01_depthwise_fp32_relu.cc
+372
-374
未找到文件。
lite/backends/arm/math/conv3x3s1p01_depthwise_fp32.cc
浏览文件 @
ff8c95d8
...
...
@@ -2631,7 +2631,7 @@ void conv_depthwise_3x3s1p1_bias_leakyRelu(float *dout,
int
cnt
=
cnt_col
;
asm
volatile
(
INIT_S1
LEFT_COMPUTE_S1
LEFT_RESULT_S1_LEAKY_RELU
MID_COMPUTE_S1
MID_RESULT_S1_LEAKY_RELU
RIGHT_COMPUTE_S1
RIGHT_RESULT_S1_LEAKY_RELU
RIGHT_RESULT_S1_LEAKY_RELU
:
[
cnt
]
"+r"
(
cnt
),
[
din_ptr0
]
"+r"
(
din_ptr0
),
[
din_ptr1
]
"+r"
(
din_ptr1
),
...
...
@@ -2729,7 +2729,7 @@ void conv_depthwise_3x3s1p1_bias_leakyRelu(float *dout,
unsigned
int
*
vmask_ptr
=
vmask
;
asm
volatile
(
INIT_S1
LEFT_COMPUTE_S1
LEFT_RESULT_S1_LEAKY_RELU
MID_COMPUTE_S1
MID_RESULT_S1_LEAKY_RELU
RIGHT_COMPUTE_S1
RIGHT_RESULT_S1_LEAKY_RELU
RIGHT_RESULT_S1_LEAKY_RELU
:
[
dout_ptr1
]
"+r"
(
doutr0
),
[
dout_ptr2
]
"+r"
(
doutr1
),
[
din0_ptr
]
"+r"
(
din_ptr0
),
...
...
lite/backends/arm/math/conv3x3s1p01_depthwise_fp32_relu.cc
浏览文件 @
ff8c95d8
此差异已折叠。
点击以展开。
lite/backends/arm/math/conv3x3s2p01_depthwise_fp32.cc
浏览文件 @
ff8c95d8
...
...
@@ -1747,53 +1747,52 @@ void conv_depthwise_3x3s2p1_bias_leakyRelu(float* dout,
doutr1_ptr
=
write_ptr
;
}
int
cnt
=
cnt_col
;
asm
volatile
(
INIT_S2
LEFT_COMPUTE_S2
LEFT_RESULT_S2_LEAKY_RELU
MID_COMPUTE_S2
MID_RESULT_S2_LEAKY_RELU
RIGHT_COMPUTE_S2
RIGHT_RESULT_S2_LEAKY_RELU
:
[
inptr0
]
"+r"
(
din0_ptr
),
[
inptr1
]
"+r"
(
din1_ptr
),
[
inptr2
]
"+r"
(
din2_ptr
),
[
inptr3
]
"+r"
(
din3_ptr
),
[
inptr4
]
"+r"
(
din4_ptr
),
[
outptr0
]
"+r"
(
doutr0_ptr
),
[
outptr1
]
"+r"
(
doutr1_ptr
),
[
cnt
]
"+r"
(
cnt
)
:
[
vzero
]
"w"
(
vzero
),
[
w0
]
"w"
(
wr0
),
[
w1
]
"w"
(
wr1
),
[
w2
]
"w"
(
wr2
),
[
remain
]
"r"
(
cnt_remain
),
[
scale_ptr
]
"r"
(
scale
),
[
mask1
]
"w"
(
vmask_rp1
),
[
mask2
]
"w"
(
vmask_rp2
),
[
wmask
]
"w"
(
wmask
),
[
vbias
]
"w"
(
wbias
)
:
"cc"
,
"memory"
,
"v0"
,
"v1"
,
"v2"
,
"v3"
,
"v4"
,
"v5"
,
"v6"
,
"v7"
,
"v8"
,
"v9"
,
"v10"
,
"v11"
,
"v12"
,
"v13"
,
"v14"
,
"v15"
,
"v16"
,
"v17"
,
"v18"
,
"v19"
,
"v20"
,
"v21"
,
"v22"
);
asm
volatile
(
INIT_S2
LEFT_COMPUTE_S2
LEFT_RESULT_S2_LEAKY_RELU
MID_COMPUTE_S2
MID_RESULT_S2_LEAKY_RELU
RIGHT_COMPUTE_S2
RIGHT_RESULT_S2_LEAKY_RELU
:
[
inptr0
]
"+r"
(
din0_ptr
),
[
inptr1
]
"+r"
(
din1_ptr
),
[
inptr2
]
"+r"
(
din2_ptr
),
[
inptr3
]
"+r"
(
din3_ptr
),
[
inptr4
]
"+r"
(
din4_ptr
),
[
outptr0
]
"+r"
(
doutr0_ptr
),
[
outptr1
]
"+r"
(
doutr1_ptr
),
[
cnt
]
"+r"
(
cnt
)
:
[
vzero
]
"w"
(
vzero
),
[
w0
]
"w"
(
wr0
),
[
w1
]
"w"
(
wr1
),
[
w2
]
"w"
(
wr2
),
[
remain
]
"r"
(
cnt_remain
),
[
scale_ptr
]
"r"
(
scale
),
[
mask1
]
"w"
(
vmask_rp1
),
[
mask2
]
"w"
(
vmask_rp2
),
[
wmask
]
"w"
(
wmask
),
[
vbias
]
"w"
(
wbias
)
:
"cc"
,
"memory"
,
"v0"
,
"v1"
,
"v2"
,
"v3"
,
"v4"
,
"v5"
,
"v6"
,
"v7"
,
"v8"
,
"v9"
,
"v10"
,
"v11"
,
"v12"
,
"v13"
,
"v14"
,
"v15"
,
"v16"
,
"v17"
,
"v18"
,
"v19"
,
"v20"
,
"v21"
,
"v22"
);
doutr0
=
doutr0
+
2
*
w_out
;
}
#else
...
...
@@ -1830,36 +1829,36 @@ void conv_depthwise_3x3s2p1_bias_leakyRelu(float* dout,
}
int
cnt
=
cnt_col
;
unsigned
int
*
mask_ptr
=
dmask
;
asm
volatile
(
INIT_S2
LEFT_COMPUTE_S2
LEFT_RESULT_S2_LEAKY_RELU
MID_COMPUTE_S2
MID_RESULT_S2_LEAKY_RELU
RIGHT_COMPUTE_S2
RIGHT_RESULT_S2_LEAKY_RELU
:
[
din0_ptr
]
"+r"
(
din0_ptr
),
[
din1_ptr
]
"+r"
(
din1_ptr
),
[
din2_ptr
]
"+r"
(
din2_ptr
),
[
outptr
]
"+r"
(
doutr0_ptr
),
[
cnt
]
"+r"
(
cnt
),
[
mask_ptr
]
"+r"
(
mask_ptr
)
:
[
remain
]
"r"
(
cnt_remain
),
[
wr0
]
"w"
(
wr0
),
[
wr1
]
"w"
(
wr1
),
[
wr2
]
"w"
(
wr2
),
[
scale_ptr
]
"r"
(
scale
),
[
bias
]
"r"
(
bias_c
)
:
"cc"
,
"memory"
,
"q3"
,
"q4"
,
"q5"
,
"q6"
,
"q7"
,
"q8"
,
"q9"
,
"q10"
,
"q11"
,
"q12"
,
"q13"
,
"q14"
,
"q15"
);
asm
volatile
(
INIT_S2
LEFT_COMPUTE_S2
LEFT_RESULT_S2_LEAKY_RELU
MID_COMPUTE_S2
MID_RESULT_S2_LEAKY_RELU
RIGHT_COMPUTE_S2
RIGHT_RESULT_S2_LEAKY_RELU
:
[
din0_ptr
]
"+r"
(
din0_ptr
),
[
din1_ptr
]
"+r"
(
din1_ptr
),
[
din2_ptr
]
"+r"
(
din2_ptr
),
[
outptr
]
"+r"
(
doutr0_ptr
),
[
cnt
]
"+r"
(
cnt
),
[
mask_ptr
]
"+r"
(
mask_ptr
)
:
[
remain
]
"r"
(
cnt_remain
),
[
wr0
]
"w"
(
wr0
),
[
wr1
]
"w"
(
wr1
),
[
wr2
]
"w"
(
wr2
),
[
scale_ptr
]
"r"
(
scale
),
[
bias
]
"r"
(
bias_c
)
:
"cc"
,
"memory"
,
"q3"
,
"q4"
,
"q5"
,
"q6"
,
"q7"
,
"q8"
,
"q9"
,
"q10"
,
"q11"
,
"q12"
,
"q13"
,
"q14"
,
"q15"
);
doutr0
=
doutr0
+
w_out
;
}
#endif
...
...
@@ -2349,8 +2348,8 @@ void conv_depthwise_3x3s2p0_bias_relu6(float* dout,
}
int
cnt
=
tile_w
;
unsigned
int
*
mask_ptr
=
dmask
;
asm
volatile
(
INIT_S2
MID_COMPUTE_S2
MID_RESULT_S2_RELU6
RIGHT_
COMPUTE_S2
RIGHT_
RESULT_S2_RELU6
asm
volatile
(
INIT_S2
MID_COMPUTE_S2
MID_RESULT_S2_RELU6
RIGHT_COMPUTE_S2
RIGHT_RESULT_S2_RELU6
:
[
din0_ptr
]
"+r"
(
din0_ptr
),
[
din1_ptr
]
"+r"
(
din1_ptr
),
[
din2_ptr
]
"+r"
(
din2_ptr
),
...
...
lite/backends/arm/math/conv3x3s2p01_depthwise_fp32_relu.cc
浏览文件 @
ff8c95d8
此差异已折叠。
点击以展开。
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