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
76a6b88d
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
未验证
提交
76a6b88d
编写于
2月 24, 2022
作者:
C
Chen Weihang
提交者:
GitHub
2月 24, 2022
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
[PHi] Skip kernel declare for cuda only kernel on rocm (#39869)
* skip kernel declare for cuda only kernel on rocm * fix error
上级
2457a7d1
变更
1
显示空白变更内容
内联
并排
Showing
1 changed file
with
9 addition
and
3 deletion
+9
-3
cmake/pten.cmake
cmake/pten.cmake
+9
-3
未找到文件。
cmake/pten.cmake
浏览文件 @
76a6b88d
...
@@ -58,15 +58,21 @@ endfunction()
...
@@ -58,15 +58,21 @@ endfunction()
function
(
kernel_declare TARGET_LIST
)
function
(
kernel_declare TARGET_LIST
)
foreach
(
kernel_path
${
TARGET_LIST
}
)
foreach
(
kernel_path
${
TARGET_LIST
}
)
file
(
READ
${
kernel_path
}
kernel_impl
)
file
(
READ
${
kernel_path
}
kernel_impl
)
# TODO(chenweihang): rename PD_REGISTER_KERNEL to PD_REGISTER_KERNEL
string
(
REGEX MATCH
"(PD_REGISTER_KERNEL|PD_REGISTER_GENERAL_KERNEL)
\\
([
\t\r\n
]*[a-z0-9_]*,[
\t\r\n
\/]*[a-z0-9_]*"
first_registry
"
${
kernel_impl
}
"
)
# NOTE(chenweihang): now we don't recommend to use digit in kernel name
string
(
REGEX MATCH
"(PD_REGISTER_KERNEL|PD_REGISTER_GENERAL_KERNEL)
\\
([
\t\r\n
]*[a-z0-9_]*,"
first_registry
"
${
kernel_impl
}
"
)
if
(
NOT first_registry STREQUAL
""
)
if
(
NOT first_registry STREQUAL
""
)
# some gpu kernel only can run on cuda, not support rocm, so we add this branch
if
(
WITH_ROCM
)
string
(
FIND
"
${
first_registry
}
"
"cuda_only"
pos
)
if
(
pos GREATER 1
)
continue
()
endif
()
endif
()
# parse the first kernel name
# parse the first kernel name
string
(
REPLACE
"PD_REGISTER_KERNEL("
""
kernel_name
"
${
first_registry
}
"
)
string
(
REPLACE
"PD_REGISTER_KERNEL("
""
kernel_name
"
${
first_registry
}
"
)
string
(
REPLACE
"PD_REGISTER_GENERAL_KERNEL("
""
kernel_name
"
${
kernel_name
}
"
)
string
(
REPLACE
"PD_REGISTER_GENERAL_KERNEL("
""
kernel_name
"
${
kernel_name
}
"
)
string
(
REPLACE
","
""
kernel_name
"
${
kernel_name
}
"
)
string
(
REPLACE
","
""
kernel_name
"
${
kernel_name
}
"
)
string
(
REGEX REPLACE
"[
\t\r\n
]+"
""
kernel_name
"
${
kernel_name
}
"
)
string
(
REGEX REPLACE
"[
\t\r\n
]+"
""
kernel_name
"
${
kernel_name
}
"
)
string
(
REGEX REPLACE
"//cuda_only"
""
kernel_name
"
${
kernel_name
}
"
)
# append kernel declare into declarations.h
# append kernel declare into declarations.h
# TODO(chenweihang): default declare ALL_LAYOUT for each kernel
# TODO(chenweihang): default declare ALL_LAYOUT for each kernel
if
(
${
kernel_path
}
MATCHES
"./cpu\/"
)
if
(
${
kernel_path
}
MATCHES
"./cpu\/"
)
...
...
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