Skip to content

P
Paddle

PaddlePaddle / Paddle
11 个月 前同步成功

通知 2289
Star 20931
Fork 5422
P
Paddle

体验新版 GitCode，发现更多精彩内容 >>

提交 25ad6884 编写于 作者: Y Yi Wang
浏览文件
操作

Merge branch 'develop' of http://github.com/paddlepaddle/paddle into cpplint-memory-detail

上级 e831bd43 c1b6692f
隐藏空白更改
内联 并排
Showing with 639 addition and 497 deletion
cmake/external/warpctc.cmake
浏览文件 @ 25ad6884
......@@ -62,7 +62,8 @@ ExternalProject_Add(
)
MESSAGE(STATUS "warp-ctc library: ${WARPCTC_LIBRARIES}")
INCLUDE_DIRECTORIES(${WARPCTC_INCLUDE_DIR})
INCLUDE_DIRECTORIES(${WARPCTC_INCLUDE_DIR}) # For warpctc code to include its headers.
INCLUDE_DIRECTORIES(${THIRD_PARTY_PATH}/install) # For Paddle code to include warpctc headers.
ADD_LIBRARY(warpctc SHARED IMPORTED GLOBAL)
SET_PROPERTY(TARGET warpctc PROPERTY IMPORTED_LOCATION ${WARPCTC_LIBRARIES})
......
doc/design/file_manager/README.md 已删除 100644 → 0
浏览文件 @ e831bd43
# FileManager设计文档
## 目标
在本文档中，我们设计说明了名为FileManager系统，方便用户上传自己的训练数据以进行分布式训练
主要功能包括：
- 提供常用的命令行管理命令管理文件和目录
- 支持大文件的断点上传、下载
## 名词解释
- PFS：是`Paddlepaddle cloud File System`的缩写，是对用户文件存储空间的抽象，与之相对的是local filesystem。目前我们用CephFS来搭建。
- [CephFS](http://docs.ceph.com/docs/master/cephfs/)：一个POSIX兼容的文件系统。
- Chunk：逻辑划上文件分块的单位。
## 模块
### 架构图
<image src=./src/filemanager.png width=900>
### PFSClient
- 功能： 详细设计[link](./pfs/pfsclient.md)
- 提供用户管理文件的命令
- 需要可以跨平台执行
- 双向验证
PFSClient需要和Ingress之间做双向验证<sup>[tls](#tls)</sup>，所以用户需要首先在`cloud.paddlepaddle.org`上注册一下，申请用户空间，并且把系统生成的CA(certificate authority)、Key、CRT(CA signed certificate)下载到本地，然后才能使用PFSClient。
### [Ingress](https://kubernetes.io/docs/concepts/services-networking/ingress/)
- 功能：
提供七层协议的反向代理、基于粘性会话的负载均衡功能。
- 透传用户身份的办法
Ingress需要把PFSClient的身份信息传给PFSServer，配置的方法参考[link](http://www.integralist.co.uk/posts/clientcertauth.html#3)
### PFSServer
PFSServer提供RESTful API接口，接收处理PFSClient端的文件管理请求，并且把结果返回PFSClient端。
RESTful API
- /api/v1/files
- `GET /api/v1/files`: Get metadata of files or directories.
- `POST /api/v1/files`: Create files or directories.
- `PATCH /api/v1/files`: Update files or directories.
- `DELETE /api/v1/files`: Delete files or directories.
- /api/v1/file/chunks
- `GET /api/v1/storage/file/chunks`: Get chunks's metadata of a file.
- /api/v1/storage/files
- `GET /api/v1/storage/files`: Download files or directories.
- `POST /api/v1/storage/files`: Upload files or directories.
- /api/v1/storage/file/chunks
- `GET /api/v1/storage/file/chunks`: Download chunks's data.
- `POST /api/v1/storage/file/chunks`: Upload chunks's data.
## 文件传输优化
### 分块文件传输
用户文件可能是比较大的，上传到Cloud或者下载到本地的时间可能比较长，而且在传输的过程中也可能出现网络不稳定的情况。为了应对以上的问题，我们提出了Chunk的概念，一个Chunk由所在的文件偏移、数据、数据长度及校验值组成。文件的上传和下载都是通过对Chunk的操作来实现的。由于Chunk比较小（默认256K），完成一个传输动作完成的时间也比较短，不容易出错。PFSClient需要在传输完毕最后一个Chunk的时候检查destination文件的MD5值是否和source文件一致。
一个典型的Chunk如下所示：
```
type Chunk struct {
fileOffset int64
checksum uint32
len uint32
data []byte
}
```
### 生成sparse文件
当destination文件不存在或者大小和source文件不一致时，可以用[Fallocate](https://Go.org/pkg/syscall/#Fallocate)生成sparse文件，然后就可以并发写入多个Chunk。
### 覆盖不一致的部分
文件传输的的关键在于需要PFSClient端对比source和destination的文件Chunks的checksum是否保持一致，不一致的由PFSClient下载或者传输Chunk完成。这样已经传输成功的部分就不用重新传输了。
## 用户使用流程
参考[link](https://github.com/PaddlePaddle/Paddle/blob/develop/doc/design/cluster_train/data_dispatch.md)
## 框架生成
用[swagger](https://github.com/swagger-api/swagger-codegen)生成PFSClient和PFSServer的框架部分，以便我们可以把更多的精力放到逻辑本身上。
## 参考文档
- <a name=tls></a>[TLS complete guide](https://github.com/k8sp/tls/blob/master/tls.md)
- [aws.s3](http://docs.aws.amazon.com/cli/latest/reference/s3/)
- [linux man document](https://linux.die.net/man/)
doc/design/file_manager/pfs/pfsclient.md 已删除 100644 → 0
浏览文件 @ e831bd43
# PFSClient
## Description
The `pfs` command is a Command Line Interface to manage your files on PaddlePaddle Cloud
## Synopsis
```
paddle [options] pfs <subcommand> [parameters]
```
## Options
```
--profile (string)
Use a specific profile from your credential file.
--help (string)
Display more information about command
--version
Output version information and exit
--debug
Show detailed debugging log
--only-show-errors (boolean)
Only errors and warnings are displayed. All other output is suppressed.
```
## Path Arguments
When using a command, we need to specify path arguments. There are two path argument type: `localpath` and `pfspath`.
A `pfspath` begin with `/pfs`, eg: `/pfs/$DATACENTER/home/$USER/folder`.
[Here](https://github.com/PaddlePaddle/Paddle/blob/develop/doc/design/cluster_train/data_dispatch.md#上传训练文件) is how to config datacenters.
## order of Path Arguments
Commonly, if there are two path arguments, the first is the source, and the second is the destination.
## Subcommonds
- rm - remove files or directories
```
Synopsis:
rm [-r] [-v] <PFSPath> ...
Options:
-r
Remove directories and their contents recursively
-v
Cause rm to be verbose, showing files after they are removed.
Examples:
paddle pfs rm /pfs/$DATACENTER/home/$USER/file
paddle pfs rm -r /pfs/$DATACENTER/home/$USER/folder
```
- mv - move (rename) files
```
Synopsis:
mv [-f | -n] [-v] <LocalPath> <PFSPath>
mv [-f | -n] [-v] <LocalPath> ... <PFSPath>
mv [-f | -n] [-v] <PFSPath> <LocalPath>
mv [-f | -n] [-v] <PFSPath> ... <LocalPath>
mv [-f | -n] [-v] <PFSPath> <PFSPath>
mv [-f | -n] [-v] <PFSPath> ... <PFSPath>
Options:
-f
Do not prompt for confirmation before overwriting the destination path. (The -f option overrides previous -n options.)
-n
Do not overwrite an existing file. (The -n option overrides previous -f options.)
-v
Cause mv to be verbose, showing files after they are moved.
Examples:
paddle pfs mv ./text1.txt /pfs/$DATACENTER/home/$USER/text1.txt
```
- cp - copy files or directories
```
Synopsis:
cp [-r] [-f | -n] [-v] [--preserve--links] <LocalPath> <PFSPath>
cp [-r] [-f | -n] [-v] [--preserve--links] <LocalPath> ... <PFSPath>
cp [-r] [-f | -n] [-v] [--preserve--links] <PFSPath> <LocalPath>
cp [-r] [-f | -n] [-v] [--preserve--links] <PFSPath> ... <LocalPath>
cp [-r] [-f | -n] [-v] [--preserve--links] <PFSPath> <PFSPath>
cp [-r] [-f | -n] [-v] [--preserve--links] <PFSPath> ... <PFSPath>
Options:
-r
Copy directories recursively
-f
Do not prompt for confirmation before overwriting the destination path. (The -f option overrides previous -n options.)
-n
Do not overwrite an existing file. (The -n option overrides previous -f options.)
-v
Cause cp to be verbose, showing files after they are copied.
--preserve--links
Reserve links when copy links
Examples:
paddle pfs cp ./file /pfs/$DATACENTER/home/$USER/file
paddle pfs cp /pfs/$DATACENTER/home/$USER/file ./file
```
- ls- list files
```
Synopsis:
ls [-r] <PFSPath> ...
Options:
-R
List directory(ies) recursively
Examples:
paddle pfs ls /pfs/$DATACENTER/home/$USER/file
paddle pfs ls /pfs/$DATACENTER/home/$USER/folder
```
- mkdir - mkdir directory(ies)
Create intermediate directory(ies) as required.
```
Synopsis:
mkdir <PFSPath> ...
Examples:
paddle pfs mkdir /pfs/$DATACENTER/home/$USER/folder
```
doc/fluid/dev/index_cn.rst
浏览文件 @ 25ad6884
......@@ -9,5 +9,5 @@
use_eigen_cn.md
name_convention.md
support_new_device.md
releasing_process.md
releasing_process_cn.md
op_markdown_format.md
doc/fluid/dev/index_en.rst
浏览文件 @ 25ad6884
......@@ -9,5 +9,5 @@ Development
use_eigen_en.md
name_convention.md
support_new_device.md
releasing_process.md
releasing_process_en.md
op_markdown_format.md
doc/fluid/dev/releasing_process.md → doc/fluid/dev/releasing_process_cn.md
浏览文件 @ 25ad6884
......@@ -10,19 +10,10 @@ PaddlePaddle每次发新的版本，遵循以下流程:
* 使用Regression Test List作为检查列表，测试本次release的正确性。
* 如果失败，记录下所有失败的例子，在这个`release/版本号`分支中，修复所有bug后，Patch号加一，到第二步
* 修改`python/setup.py.in`中的版本信息,并将`istaged`字段设为`True`。
* 编译这个版本的python wheel包，并发布到pypi。
* 由于pypi.python.org目前遵循[严格的命名规范PEP 513](https://www.python.org/dev/peps/pep-0513)，在使用twine上传之前，需要重命名wheel包中platform相关的后缀，比如将`linux_x86_64`修改成`manylinux1_x86_64`。
* pypi上的package名称为paddlepaddle和paddlepaddle_gpu，如果要上传GPU版本的包，需要修改build/python/setup.py中，name: "paddlepaddle_gpu"并重新打包wheel包：`python setup.py bdist_wheel`。
* 上传方法：
```
cd build/python
pip install twine
twine upload dist/[package to upload]
```
* 编译这个版本的Docker发行镜像，发布到dockerhub。如果失败，修复Docker编译镜像问题，Patch号加一，返回第二步
1. 第三步完成后，将`release/版本号`分支合入master分支，并删除`release/版本号`分支。将master分支的合入commit打上tag，tag为`版本号`。同时再将`master`分支合入`develop`分支。最后删除`release/版本号`分支。
1. 协同完成Release Note的书写
* 将这个版本的python wheel包发布到pypi。
* 更新Docker镜像（参考后面的操作细节）。
1. 第三步完成后，将`release/版本号`分支合入master分支，将master分支的合入commit打上tag，tag为`版本号`。同时再将`master`分支合入`develop`分支。
1. 协同完成Release Note的书写。
需要注意的是:
......@@ -31,13 +22,18 @@ PaddlePaddle每次发新的版本，遵循以下流程:
## 发布wheel包到pypi
使用[PaddlePaddle CI](https://paddleci.ngrok.io/project.html?projectId=Manylinux1&tab=projectOverview)
1. 使用[PaddlePaddle CI](https://paddleci.ngrok.io/project.html?projectId=Manylinux1&tab=projectOverview)
完成自动化二进制编译，参考下图，选择需要发布的版本（通常包含一个CPU版本和一个GPU版本），点击"run"右侧的"..."按钮，可以
弹出下面的选择框，在第二个tab (Changes)里选择需要发布的分支，这里选择0.11.0，然后点击"Run Build"按钮。等待编译完成后
可以在此页面的"Artifacts"下拉框中找到生成的3个二进制文件，分别对应CAPI，`cp27m`和`cp27mu`的版本。然后按照上述的方法
使用`twine`工具上传即可。
<img src="https://raw.githubusercontent.com/PaddlePaddle/Paddle/develop/doc/fluid/images/ci_build_whl.png">
弹出下面的选择框，在第二个tab (Changes)里选择需要发布的分支，这里选择0.11.0，然后点击"Run Build"按钮。
<img src="https://raw.githubusercontent.com/PaddlePaddle/Paddle/develop/doc/fluid/images/ci_build_whl.png">
1. 等待编译完成后可以在此页面的"Artifacts"下拉框中找到生成的3个二进制文件，分别对应CAPI，`cp27m`和`cp27mu`的版本。
1. 由于pypi.python.org目前遵循[严格的命名规范PEP 513](https://www.python.org/dev/peps/pep-0513)，在使用twine上传之前，需要重命名wheel包中platform相关的后缀，比如将`linux_x86_64`修改成`manylinux1_x86_64`。
1. 上传：
```
cd build/python
pip install twine
twine upload dist/[package to upload]
```
* 注：CI环境使用 https://github.com/PaddlePaddle/buildtools 这里的DockerImage作为编译环境以支持更多的Linux
发型版，如果需要手动编译，也可以使用这些镜像。这些镜像也可以从 https://hub.docker.com/r/paddlepaddle/paddle_manylinux_devel/tags/ 下载得到。
......@@ -48,10 +44,20 @@ PaddlePaddle每次发新的版本，遵循以下流程:
上述PaddlePaddle CI编译wheel完成后会自动将Docker镜像push到DockerHub，所以，发布Docker镜像只需要对自动push的镜像打上
版本号对应的tag即可：
1. 进入 https://hub.docker.com/r/paddlepaddle/paddle/tags/ 查看latest tag的更新时间是否在上述编译wheel包完成后是否最新。
1. 执行 `docker pull paddlepaddle/paddle:[latest tag]`，latest tag可以是latest或latest-gpu等。
1. 执行 `docker tag paddlepaddle/paddle:[latest tag] paddlepaddle/paddle:[version]`
1. 执行 `docker push paddlepaddle/paddle:[version]`
```
docker pull [镜像]:latest
docker tag [镜像]:latest [镜像]:[version]
docker push [镜像]:[version]
```
需要更新的镜像tag包括：
* `[version]`: CPU版本
* `[version]-openblas`: openblas版本
* `[version]-gpu`: GPU版本（CUDA 8.0 cudnn 5）
* `[version]-gpu-[cudaver]-[cudnnver]`: 不同cuda, cudnn版本的镜像
之后可进入 https://hub.docker.com/r/paddlepaddle/paddle/tags/ 查看是否发布成功。
## PaddlePaddle 分支规范
......@@ -76,7 +82,7 @@ PaddlePaddle开发过程使用[git-flow](http://nvie.com/posts/a-successful-git-
### PaddlePaddle Book中所有章节
PaddlePaddle每次发版本首先要保证PaddlePaddle Book中所有章节功能的正确性。功能的正确性包括验证PaddlePaddle目前的`paddle_trainer`训练和纯使用`Python`训练模型正确性。
PaddlePaddle每次发版本首先要保证PaddlePaddle Book中所有章节功能的正确性。功能的正确性包括验证PaddlePaddle目前的`paddle_trainer`训练和纯使用`Python`训练（V2和Fluid）模型正确性。
<table>
<thead>
......
doc/fluid/dev/releasing_process_en.md 0 → 100644
浏览文件 @ 25ad6884
# PaddlePaddle Releasing Process
PaddlePaddle manages its branches using "git-flow branching model", and [Semantic Versioning](http://semver.org/) as it's version number semantics.
Each time we release a new PaddlePaddle version, we should follow the below steps:
1. Fork a new branch from `develop` named `release/[version]`, e.g. `release/0.10.0`.
1. Push a new tag on the release branch, the tag name should be like `[version]rc.patch`. The
first tag should be `0.10.0rc1`, and the second should be `0.10.0.rc2` and so on.
1. After that, we should do:
* Run all regression test on the Regression Test List (see PaddlePaddle TeamCity CI), to confirm
that this release has no major bugs.
* If regression test fails, we must fix those bugs and create a new `release/[version]`
branch from previous release branch.
* Modify `python/setup.py.in`, change the version number and change `ISTAGED` to `True`.
* Publish PaddlePaddle release wheel packages to pypi (see below instructions for detail).
* Update the Docker images (see below instructions for detail).
1. After above step, merge `release/[version]` branch to master and push a tag on the master commit,
then merge `master` to `develop`.
1. Update the Release Note.
***NOTE:***
* Do ***NOT*** merge commits from develop branch to release branches to keep the release branch contain
features only for current release, so that we can test on that version.
* If we want to fix bugs on release branches, we must merge the fix to master, develop and release branch.
## Publish Wheel Packages to pypi
1. Use our [CI tool](https://paddleci.ngrok.io/project.html?projectId=Manylinux1&tab=projectOverview)
to build all wheel packages needed to publish. As shown in the following picture, choose a build
version, click "..." button on the right side of "Run" button, and switch to the second tab in the
pop-up box, choose the current release branch and click "Run Build" button. You may repeat this
step to start different versions of builds.
<img src="https://raw.githubusercontent.com/PaddlePaddle/Paddle/develop/doc/fluid/images/ci_build_whl.png">
1. After the build succeeds, download the outputs under "Artifacts" including capi, `cp27m` and `cp27mu`.
1. Since pypi.python.org follows [PEP 513](https://www.python.org/dev/peps/pep-0513), before we
upload the package using `twine`, we need to rename the package from `linux_x86_64` to
`manylinux1_x86_64`.
1. Start the upload:
```
cd build/python
pip install twine
twine upload dist/[package to upload]
```
* NOTE: We use a special Docker image to build our releases to support more Linux distributions, you can
download it from https://hub.docker.com/r/paddlepaddle/paddle_manylinux_devel/tags/, or build it using
scripts under `tools/manylinux1`.
* pypi does not allow overwrite the already uploaded version of wheel package, even if you delete the
old version. you must change the version number before upload a new one.
## Publish Docker Images
Our CI tool will push latest images to DockerHub, so we only need to push a version tag like:
```
docker pull [image]:latest
docker tag [image]:latest [image]:[version]
docker push [image]:[version]
```
Tags that need to be updated are:
* `[version]`: CPU only version image
* `[version]-openblas`: openblas version image
* `[version]-gpu`: GPU version（using CUDA 8.0 cudnn 5）
* `[version]-gpu-[cudaver]-[cudnnver]`: tag for different cuda, cudnn versions
You can then checkout the latest pushed tags at https://hub.docker.com/r/paddlepaddle/paddle/tags/.
## Branching Model
We use [git-flow](http://nvie.com/posts/a-successful-git-branching-model/) as our branching model,
with some modifications:
* `master` branch is the stable branch. Each version on the master branch is tested and guaranteed.
* `develop` branch is for development. Each commit on develop branch has passed CI unit test, but no
regression tests are run.
* `release/[version]` branch is used to publish each release. Latest release version branches have
bugfix only for that version, but no feature updates.
* Developer forks are not required to follow
[git-flow](http://nvie.com/posts/a-successful-git-branching-model/)
branching model, all forks is like a feature branch.
* Advise: developer fork's develop branch is used to sync up with main repo's develop branch.
* Advise: developer use it's fork's develop branch to for new branch to start developing.
* Use that branch on developer's fork to create pull requests and start reviews.
* developer can push new commits to that branch when the pull request is open.
* Bug fixes are also started from developers forked repo. And, bug fixes branch can merge to
`master`, `develop` and `releases`.
## PaddlePaddle Regression Test List
### All Chapters of PaddlePaddle Book
We need to guarantee that all the chapters of PaddlePaddle Book can run correctly. Including
V1 (`paddle_trainer` training) and V2 training and Fluid training.
<table>
<thead>
<tr>
<th></th>
<th>Linear Regression</th>
<th>Recognize Digits</th>
<th>Image Classification</th>
<th>Word2Vec</th>
<th>Personalized Recommendation</th>
<th>Sentiment Analysis</th>
<th>Semantic Role Labeling</th>
<th>Machine Translation</th>
</tr>
</thead>
<tbody>
<tr>
<td>API.V2 + Docker + GPU </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
</tr>
<tr>
<td> API.V2 + Docker + CPU </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
</tr>
<tr>
<td>`paddle_trainer` + Docker + GPU </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
</tr>
<tr>
<td>`paddle_trainer` + Docker + CPU </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
</tr>
<tr>
<td> API.V2 + Ubuntu + GPU</td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
</tr>
<tr>
<td>API.V2 + Ubuntu + CPU </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
</tr>
<tr>
<td> `paddle_trainer` + Ubuntu + GPU</td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
</tr>
<tr>
<td> `paddle_trainer` + Ubuntu + CPU</td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
<td> </td>
</tr>
</tbody>
</table>
paddle/fluid/framework/details/CMakeLists.txt
浏览文件 @ 25ad6884
......@@ -16,6 +16,6 @@ else()
endif()
cc_library(multi_devices_graph_builder SRCS multi_devices_graph_builder.cc DEPS ssa_graph_builder computation_op_handle
scale_loss_grad_op_handle ${multi_devices_graph_builder_deps})
cc_library(ssa_graph_executor SRCS ssa_graph_executor.cc DEPS ssa_graph)
cc_library(ssa_graph_executor SRCS ssa_graph_executor.cc DEPS ssa_graph framework_proto)
cc_library(threaded_ssa_graph_executor SRCS threaded_ssa_graph_executor.cc DEPS fetch_op_handle ssa_graph_executor scope
simple_threadpool device_context)
paddle/fluid/framework/scope.cc
浏览文件 @ 25ad6884
......@@ -15,7 +15,6 @@ limitations under the License. */
#include "paddle/fluid/framework/scope.h"
#include <memory> // for unique_ptr
#include <mutex> // for call_once
#include <set>
#include "glog/logging.h"
#include "paddle/fluid/framework/threadpool.h"
......@@ -39,6 +38,7 @@ Scope::~Scope() {
}
Scope& Scope::NewScope() const {
std::unique_lock<std::mutex> lock(mutex_);
kids_.push_back(new Scope(this));
return *kids_.back();
}
......@@ -92,6 +92,7 @@ std::vector<std::string> Scope::LocalVarNames() const {
}
void Scope::DeleteScope(Scope* scope) {
std::unique_lock<std::mutex> lock(mutex_);
auto it = std::find(this->kids_.begin(), this->kids_.end(), scope);
PADDLE_ENFORCE(it != this->kids_.end(), "Cannot find %p as kid scope", scope);
this->kids_.erase(it);
......@@ -103,7 +104,7 @@ void Scope::DeleteScope(Scope* scope) {
}
}
void Scope::EraseVars(std::vector<std::string>& var_names) {
void Scope::EraseVars(const std::vector<std::string>& var_names) {
std::set<std::string> var_set(var_names.begin(), var_names.end());
for (auto it = vars_.begin(); it != vars_.end();) {
if (var_set.find(it->first) != var_set.end()) {
......
paddle/fluid/framework/scope.h
浏览文件 @ 25ad6884
......@@ -15,6 +15,7 @@ limitations under the License. */
#pragma once
#include <list>
#include <mutex> // NOLINT
#include <string>
#include <unordered_map>
#include <vector>
......@@ -51,7 +52,7 @@ class Scope {
/// Create a variable with a scope-unique name.
Variable* Var(std::string* name = nullptr);
void EraseVars(std::vector<std::string>& var_names);
void EraseVars(const std::vector<std::string>& var_names);
/// Find a variable in the scope or any of its ancestors. Returns
/// nullptr if cannot find.
......@@ -88,6 +89,9 @@ class Scope {
Scope const* parent_{nullptr};
DISABLE_COPY_AND_ASSIGN(Scope);
private:
mutable std::mutex mutex_;
};
} // namespace framework
} // namespace paddle
paddle/fluid/operators/detail/CMakeLists.txt
浏览文件 @ 25ad6884
......@@ -5,5 +5,5 @@ if(WITH_DISTRIBUTE)
set_source_files_properties(serde_test.cc grpc_server_test.cc PROPERTIES COMPILE_FLAGS ${DISTRIBUTE_COMPILE_FLAGS})
cc_test(serde_test SRCS serde_test.cc variable_response.cc DEPS grpc++_unsecure grpc_unsecure gpr
cares zlib protobuf sendrecvop_grpc)
cc_test(grpc_server_test SRCS grpc_server_test.cc DEPS sendrecvop_grpc grpc++_unsecure grpc_unsecure gpr cares zlib protobuf)
cc_test(grpc_server_test SRCS grpc_server_test.cc DEPS sendrecvop_grpc grpc++_unsecure grpc_unsecure gpr cares zlib protobuf executor proto_desc lookup_table_op)
endif()
paddle/fluid/operators/detail/grpc_client.cc
浏览文件 @ 25ad6884
......@@ -138,7 +138,7 @@ bool RPCClient::AsyncPrefetchVariable(const std::string& ep,
auto* var = p_scope->FindVar(in_var_name_val);
::grpc::ByteBuffer req;
SerializeToByteBuffer(in_var_name_val, var, *p_ctx, &req);
SerializeToByteBuffer(in_var_name_val, var, *p_ctx, &req, out_var_name_val);
// var handle
VarHandle var_h;
......
paddle/fluid/operators/detail/grpc_server.cc
浏览文件 @ 25ad6884
......@@ -138,39 +138,48 @@ class RequestPrefetch final : public RequestBase {
framework::Scope* scope,
const platform::DeviceContext* dev_ctx,
framework::Executor* executor,
framework::ProgramDesc* program, int blkid)
framework::ProgramDesc* program,
framework::ExecutorPrepareContext* prefetch_ctx)
: RequestBase(service, cq, dev_ctx),
responder_(&ctx_),
scope_(scope),
executor_(executor),
program_(program),
blkid_(blkid) {
prefetch_ctx_(prefetch_ctx) {
request_.reset(new VariableResponse(scope, dev_ctx_));
int method_id = static_cast<int>(detail::GrpcMethod::kPrefetchVariable);
service_->RequestAsyncUnary(method_id, &ctx_, &request_, &responder_, cq_,
cq_, this);
service_->RequestAsyncUnary(method_id, &ctx_, request_.get(), &responder_,
cq_, cq_, this);
}
virtual ~RequestPrefetch() {}
virtual std::string GetReqName() { return request_.varname(); }
virtual std::string GetReqName() { return request_->Varname(); }
virtual void Process() {
// prefetch process...
::grpc::ByteBuffer reply;
// TODO(Yancey1989): execute the Block which containers prefetch ops
VLOG(3) << "RequestPrefetch Process in";
std::string var_name = request_->OutVarname();
auto var_desc = program_->Block(0).FindVar(var_name);
framework::Scope* local_scope = &scope_->NewScope();
auto* var = local_scope->FindVar(var_name);
InitializeVariable(var, var_desc->GetType());
executor_->RunPreparedContext(prefetch_ctx_, scope_, false, false);
SerializeToByteBuffer(var_name, var, *dev_ctx_, &reply);
responder_.Finish(reply, ::grpc::Status::OK, this);
status_ = FINISH;
}
protected:
sendrecv::VariableMessage request_;
std::shared_ptr<VariableResponse> request_;
ServerAsyncResponseWriter<::grpc::ByteBuffer> responder_;
framework::Scope* scope_;
framework::Executor* executor_;
framework::ProgramDesc* program_;
framework::ExecutorPrepareContext* prefetch_ctx_;
int blkid_;
};
......@@ -268,7 +277,7 @@ void AsyncGRPCServer::TryToRegisterNewPrefetchOne() {
}
RequestPrefetch* prefetch =
new RequestPrefetch(&service_, cq_prefetch_.get(), scope_, dev_ctx_,
executor_, program_, prefetch_blk_id_);
executor_, program_, prefetch_ctx_);
VLOG(4) << "Create RequestPrefetch status:" << prefetch->Status();
}
......
paddle/fluid/operators/detail/grpc_server.h
浏览文件 @ 25ad6884
......@@ -63,6 +63,10 @@ class AsyncGRPCServer final {
void SetExecutor(framework::Executor *executor) { executor_ = executor; }
void SetPrefetchPreparedCtx(framework::ExecutorPrepareContext *prepared) {
prefetch_ctx_ = prepared;
}
int GetSelectedPort() { return selected_port_; }
const ReceivedMessage Get() { return this->var_recv_queue_.Pop(); }
......@@ -111,6 +115,7 @@ class AsyncGRPCServer final {
std::unique_ptr<std::thread> t_prefetch_;
int prefetch_blk_id_;
framework::ExecutorPrepareContext *prefetch_ctx_;
framework::ProgramDesc *program_;
framework::Executor *executor_;
int selected_port_;
......
paddle/fluid/operators/detail/grpc_server_test.cc
浏览文件 @ 25ad6884
......@@ -20,43 +20,121 @@ limitations under the License. */
#include "paddle/fluid/operators/detail/grpc_client.h"
#include "paddle/fluid/operators/detail/grpc_server.h"
#include "paddle/fluid/framework/block_desc.h"
#include "paddle/fluid/framework/op_registry.h"
#include "paddle/fluid/framework/operator.h"
namespace framework = paddle::framework;
namespace platform = paddle::platform;
namespace detail = paddle::operators::detail;
USE_OP(lookup_table);
std::unique_ptr<detail::AsyncGRPCServer> rpc_service_;
framework::BlockDesc* AppendPrefetchBlcok(framework::ProgramDesc* program) {
auto root_block = program->MutableBlock(0);
auto* block = program->AppendBlock(*root_block);
framework::VariableNameMap input({{"W", {"w"}}, {"Ids", {"ids"}}});
framework::VariableNameMap output({{"Output", {"out"}}});
auto op = block->AppendOp();
op->SetType("lookup_table");
op->SetInput("W", {"w"});
op->SetInput("Ids", {"ids"});
op->SetOutput("Out", {"out"});
auto& out = *root_block->Var("out");
out.SetType(framework::proto::VarType::SELECTED_ROWS);
out.SetShape({10, 10});
return block;
}
void CreateVarsOnScope(framework::Scope* scope, platform::CPUPlace* place) {
auto w_var = scope->Var("w");
w_var->GetMutable<framework::SelectedRows>();
auto out_var = scope->Var("out");
out_var->GetMutable<framework::SelectedRows>();
auto ids_var = scope->Var("ids");
ids_var->GetMutable<framework::SelectedRows>();
}
void InitTensorsOnClient(framework::Scope* scope, platform::CPUPlace* place,
int64_t rows_numel) {
CreateVarsOnScope(scope, place);
auto ids_var = scope->Var("ids")->GetMutable<framework::SelectedRows>();
auto rows = ids_var->mutable_rows();
for (int64_t i = 0; i < rows_numel; ++i) rows->push_back(i * 2);
ids_var->mutable_value()->Resize({rows_numel, 1});
ids_var->mutable_value()->mutable_data<float>(*place);
}
void InitTensorsOnServer(framework::Scope* scope, platform::CPUPlace* place,
int64_t rows_numel) {
CreateVarsOnScope(scope, place);
auto w = scope->Var("w")->GetMutable<framework::SelectedRows>();
auto rows = w->mutable_rows();
for (int64_t i = 0; i < rows_numel; ++i) rows->push_back(i);
auto w_value = w->mutable_value();
w_value->Resize({rows_numel, 10});
auto ptr = w_value->mutable_data<float>(*place);
for (int64_t i = 0; i < w_value->numel(); ++i) {
ptr[i] = static_cast<float>(i / 10);
}
}
void StartServer(const std::string& endpoint) {
rpc_service_.reset(new detail::AsyncGRPCServer(endpoint));
framework::ProgramDesc program;
framework::Scope scope;
platform::CPUPlace place;
framework::Executor exe(place);
platform::CPUDeviceContext ctx(place);
auto* block = AppendPrefetchBlcok(&program);
auto prepared = exe.Prepare(program, block->ID());
InitTensorsOnServer(&scope, &place, 10);
rpc_service_->SetProgram(&program);
rpc_service_->SetPrefetchPreparedCtx(prepared.get());
rpc_service_->SetDevCtx(&ctx);
rpc_service_->SetScope(&scope);
rpc_service_->SetExecutor(&exe);
rpc_service_->RunSyncUpdate();
}
TEST(PREFETCH, CPU) {
// start up a server instance backend
// TODO(Yancey1989): Need to start a server with optimize blocks and
// prefetch blocks.
std::thread server_thread(StartServer, "127.0.0.1:8889");
sleep(2);
framework::Scope scope;
platform::CPUPlace place;
platform::CPUDeviceContext ctx(place);
// create var on local scope
std::string in_var_name("in");
int64_t rows_numel = 5;
InitTensorsOnClient(&scope, &place, rows_numel);
std::string in_var_name("ids");
std::string out_var_name("out");
auto* in_var = scope.Var(in_var_name);
auto* in_tensor = in_var->GetMutable<framework::LoDTensor>();
in_tensor->Resize({10, 10});
VLOG(3) << "before mutable_data";
in_tensor->mutable_data<int>(place);
scope.Var(out_var_name);
VLOG(3) << "before fetch";
detail::RPCClient client;
client.AsyncPrefetchVariable("127.0.0.1:8889", ctx, scope, in_var_name,
out_var_name);
client.Wait();
auto var = scope.Var(out_var_name);
auto value = var->GetMutable<framework::SelectedRows>()->value();
auto ptr = value.mutable_data<float>(place);
rpc_service_->ShutDown();
server_thread.join();
rpc_service_.reset(nullptr);
for (int64_t i = 0; i < rows_numel; ++i) {
EXPECT_EQ(ptr[0 + i * value.dims()[1]], static_cast<float>(i * 2));
}
}
paddle/fluid/operators/detail/send_recv.proto
浏览文件 @ 25ad6884
......@@ -21,7 +21,7 @@ service SendRecvService {
rpc SendVariable(VariableMessage) returns (VoidMessage) {}
// Argument VariableMessage for GetVariable should only contain varname.
rpc GetVariable(VariableMessage) returns (VariableMessage) {}
// Prefetch variable by Ids
// pre-fetch variable by given variable name and Ids
rpc PrefetchVariable(VariableMessage) returns (VariableMessage) {}
}
......@@ -67,6 +67,8 @@ message VariableMessage {
bytes serialized = 8;
// selected_rows data
bytes rows = 9;
// Look up table block execution output variable name.
string out_varname = 10;
}
message VoidMessage {}
paddle/fluid/operators/detail/sendrecvop_utils.cc
浏览文件 @ 25ad6884
......@@ -30,7 +30,8 @@ namespace detail {
void SerializeToByteBuffer(const std::string& name, framework::Variable* var,
const platform::DeviceContext& ctx,
::grpc::ByteBuffer* msg) {
::grpc::ByteBuffer* msg,
const std::string& out_name) {
using VarMsg = sendrecv::VariableMessage;
sendrecv::VariableMessage request;
std::string header;
......@@ -52,6 +53,9 @@ void SerializeToByteBuffer(const std::string& name, framework::Variable* var,
e.WriteUint64(VarMsg::kTypeFieldNumber, 1);
}
if (!out_name.empty()) {
e.WriteString(VarMsg::kOutVarnameFieldNumber, out_name);
}
switch (framework::ToVarType(var->Type())) {
case framework::proto::VarType_Type_LOD_TENSOR: {
auto tensor = var->Get<framework::LoDTensor>();
......
paddle/fluid/operators/detail/sendrecvop_utils.h
浏览文件 @ 25ad6884
......@@ -46,7 +46,8 @@ typedef void (*DestroyCallback)(void*);
void SerializeToByteBuffer(const std::string& name, framework::Variable* var,
const platform::DeviceContext& ctx,
::grpc::ByteBuffer* msg);
::grpc::ByteBuffer* msg,
const std::string& out_varname = std::string());
void DeserializeFromByteBuffer(const ::grpc::ByteBuffer& msg,
const platform::DeviceContext& ctx,
......
paddle/fluid/operators/detail/variable_response.cc
浏览文件 @ 25ad6884
......@@ -416,6 +416,20 @@ int VariableResponse::Parse(Source* source) {
}
break;
}
case sendrecv::VariableMessage::kOutVarnameFieldNumber: {
uint32_t length;
if ((wt != WIRETYPE_LENGTH_DELIMITED) || !input.ReadVarint32(&length)) {
return tag;
}
std::string temp;
if (!input.ReadString(&temp, length)) {
return tag;
}
meta_.set_out_varname(temp);
break;
}
default: {
// Unknown tag, return unknown error.
......
paddle/fluid/operators/detail/variable_response.h
浏览文件 @ 25ad6884
......@@ -55,6 +55,7 @@ class VariableResponse {
int Parse(const ::grpc::ByteBuffer& byte_buffer);
inline std::string Varname() { return meta_.varname(); }
inline std::string OutVarname() { return meta_.out_varname(); }
// should call parse first.
framework::Variable* GetVar() { return scope_->FindVar(meta_.varname()); }
......
paddle/fluid/operators/reader/create_batch_reader_op.cc
浏览文件 @ 25ad6884
......@@ -39,10 +39,13 @@ class CreateBatchReaderOp : public framework::OperatorBase {
private:
void RunImpl(const framework::Scope& scope,
const platform::Place& dev_place) const override {
const auto& underlying_reader = scope.FindVar(Input("UnderlyingReader"))
->Get<framework::ReaderHolder>();
auto* out = scope.FindVar(Output("Out"))
->template GetMutable<framework::ReaderHolder>();
if (out->Get() != nullptr) {
return;
}
const auto& underlying_reader = scope.FindVar(Input("UnderlyingReader"))
->Get<framework::ReaderHolder>();
out->Reset(
new BatchReader(underlying_reader.Get(), Attr<int>("batch_size")));
}
......
paddle/fluid/operators/reader/create_double_buffer_reader_op.cc
浏览文件 @ 25ad6884
......@@ -99,10 +99,13 @@ class CreateDoubleBufferReaderOp : public framework::OperatorBase {
private:
void RunImpl(const framework::Scope& scope,
const platform::Place& dev_place) const override {
const auto& underlying_reader = scope.FindVar(Input("UnderlyingReader"))
->Get<framework::ReaderHolder>();
auto* out = scope.FindVar(Output("Out"))
->template GetMutable<framework::ReaderHolder>();
if (out->Get() != nullptr) {
return;
}
const auto& underlying_reader = scope.FindVar(Input("UnderlyingReader"))
->Get<framework::ReaderHolder>();
auto place_str = Attr<std::string>("place");
platform::Place place;
......
paddle/fluid/operators/reader/create_multi_pass_reader_op.cc
浏览文件 @ 25ad6884
......@@ -62,12 +62,15 @@ class CreateMultiPassReaderOp : public framework::OperatorBase {
private:
void RunImpl(const framework::Scope& scope,
const platform::Place& dev_place) const override {
auto* out = detail::Ref(scope.FindVar(Output("Out")))
.GetMutable<framework::ReaderHolder>();
if (out->Get() != nullptr) {
return;
}
const auto& underlying_reader = scope.FindVar(Input("UnderlyingReader"))
->Get<framework::ReaderHolder>();
auto& out = detail::Ref(scope.FindVar(Output("Out")));
int pass_num = Attr<int>("pass_num");
out.GetMutable<framework::ReaderHolder>()->Reset(
new MultiPassReader(underlying_reader.Get(), pass_num));
out->Reset(new MultiPassReader(underlying_reader.Get(), pass_num));
}
};
......
paddle/fluid/operators/reader/create_shuffle_reader_op.cc
浏览文件 @ 25ad6884
......@@ -80,10 +80,14 @@ class CreateShuffleReaderOp : public framework::OperatorBase {
private:
void RunImpl(const framework::Scope& scope,
const platform::Place& dev_place) const override {
auto* out = detail::Ref(scope.FindVar(Output("Out")))
.GetMutable<framework::ReaderHolder>();
if (out->Get() != nullptr) {
return;
}
const auto& underlying_reader = scope.FindVar(Input("UnderlyingReader"))
->Get<framework::ReaderHolder>();
auto& var = detail::Ref(scope.FindVar(Output("Out")));
var.GetMutable<framework::ReaderHolder>()->Reset(
out->Reset(
new ShuffleReader(underlying_reader.Get(),
static_cast<size_t>(Attr<int>("buffer_size"))));
}
......
paddle/fluid/platform/dynload/cublas.h
浏览文件 @ 25ad6884
/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License. */
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License. */
#pragma once
......@@ -35,18 +35,18 @@ extern void *cublas_dso_handle;
* note: default dynamic linked libs
*/
#ifdef PADDLE_USE_DSO
#define DECLARE_DYNAMIC_LOAD_CUBLAS_WRAP(__name) \
struct DynLoad__##__name { \
template <typename... Args> \
inline cublasStatus_t operator()(Args... args) { \
typedef cublasStatus_t (*cublasFunc)(Args...); \
std::call_once(cublas_dso_flag, \
paddle::platform::dynload::GetCublasDsoHandle, \
&cublas_dso_handle); \
void *p_##__name = dlsym(cublas_dso_handle, #__name); \
return reinterpret_cast<cublasFunc>(p_##__name)(args...); \
} \
}; \
#define DECLARE_DYNAMIC_LOAD_CUBLAS_WRAP(__name) \
struct DynLoad__##__name { \
template <typename... Args> \
inline cublasStatus_t operator()(Args... args) { \
typedef cublasStatus_t (*cublasFunc)(Args...); \
std::call_once(cublas_dso_flag, []() { \
cublas_dso_handle = paddle::platform::dynload::GetCublasDsoHandle(); \
}); \
void *p_##__name = dlsym(cublas_dso_handle, #__name); \
return reinterpret_cast<cublasFunc>(p_##__name)(args...); \
} \
}; \
extern DynLoad__##__name __name
#else
#define DECLARE_DYNAMIC_LOAD_CUBLAS_WRAP(__name) \
......
paddle/fluid/platform/dynload/cudnn.cc
浏览文件 @ 25ad6884
......@@ -44,7 +44,8 @@ CUDNN_DNN_ROUTINE_EACH_R7(DEFINE_WRAP);
#ifdef PADDLE_USE_DSO
bool HasCUDNN() {
std::call_once(cudnn_dso_flag, GetCUDNNDsoHandle, &cudnn_dso_handle);
std::call_once(cudnn_dso_flag,
[]() { cudnn_dso_handle = GetCUDNNDsoHandle(); });
return cudnn_dso_handle != nullptr;
}
......
paddle/fluid/platform/dynload/cudnn.h
浏览文件 @ 25ad6884
......@@ -30,19 +30,19 @@ extern bool HasCUDNN();
#ifdef PADDLE_USE_DSO
extern void EnforceCUDNNLoaded(const char* fn_name);
#define DECLARE_DYNAMIC_LOAD_CUDNN_WRAP(__name) \
struct DynLoad__##__name { \
template <typename... Args> \
auto operator()(Args... args) -> decltype(__name(args...)) { \
using cudnn_func = decltype(__name(args...)) (*)(Args...); \
std::call_once(cudnn_dso_flag, \
paddle::platform::dynload::GetCUDNNDsoHandle, \
&cudnn_dso_handle); \
EnforceCUDNNLoaded(#__name); \
void* p_##__name = dlsym(cudnn_dso_handle, #__name); \
return reinterpret_cast<cudnn_func>(p_##__name)(args...); \
} \
}; \
#define DECLARE_DYNAMIC_LOAD_CUDNN_WRAP(__name) \
struct DynLoad__##__name { \
template <typename... Args> \
auto operator()(Args... args) -> decltype(__name(args...)) { \
using cudnn_func = decltype(__name(args...)) (*)(Args...); \
std::call_once(cudnn_dso_flag, []() { \
cudnn_dso_handle = paddle::platform::dynload::GetCUDNNDsoHandle(); \
}); \
EnforceCUDNNLoaded(#__name); \
void* p_##__name = dlsym(cudnn_dso_handle, #__name); \
return reinterpret_cast<cudnn_func>(p_##__name)(args...); \
} \
}; \
extern struct DynLoad__##__name __name
#else
......
paddle/fluid/platform/dynload/cupti.h
浏览文件 @ 25ad6884
......@@ -11,14 +11,15 @@ distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License. */
#pragma once
#ifdef PADDLE_WITH_CUPTI
#include <cuda.h>
#include <cupti.h>
#include <dlfcn.h>
#include <mutex>
#include <mutex> // NOLINT
#include "paddle/fluid/platform/dynload/dynamic_loader.h"
namespace paddle {
......@@ -36,18 +37,18 @@ extern void *cupti_dso_handle;
* note: default dynamic linked libs
*/
#ifdef PADDLE_USE_DSO
#define DECLARE_DYNAMIC_LOAD_CUPTI_WRAP(__name) \
struct DynLoad__##__name { \
template <typename... Args> \
inline CUptiResult CUPTIAPI operator()(Args... args) { \
typedef CUptiResult CUPTIAPI (*cuptiFunc)(Args...); \
std::call_once(cupti_dso_flag, \
paddle::platform::dynload::GetCUPTIDsoHandle, \
&cupti_dso_handle); \
void *p_##__name = dlsym(cupti_dso_handle, #__name); \
return reinterpret_cast<cuptiFunc>(p_##__name)(args...); \
} \
}; \
#define DECLARE_DYNAMIC_LOAD_CUPTI_WRAP(__name) \
struct DynLoad__##__name { \
template <typename... Args> \
inline CUptiResult CUPTIAPI operator()(Args... args) { \
typedef CUptiResult CUPTIAPI (*cuptiFunc)(Args...); \
std::call_once(cupti_dso_flag, []() { \
cupti_dso_handle = paddle::platform::dynload::GetCUPTIDsoHandle(); \
}); \
void *p_##__name = dlsym(cupti_dso_handle, #__name); \
return reinterpret_cast<cuptiFunc>(p_##__name)(args...); \
} \
}; \
extern DynLoad__##__name __name
#else
#define DECLARE_DYNAMIC_LOAD_CUPTI_WRAP(__name) \
......
paddle/fluid/platform/dynload/curand.h
浏览文件 @ 25ad6884
......@@ -11,12 +11,13 @@ distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License. */
#pragma once
#include <curand.h>
#include <dlfcn.h>
#include <mutex>
#include <mutex> // NOLINT
#include "paddle/fluid/platform/dynload/dynamic_loader.h"
namespace paddle {
......@@ -25,18 +26,18 @@ namespace dynload {
extern std::once_flag curand_dso_flag;
extern void *curand_dso_handle;
#ifdef PADDLE_USE_DSO
#define DECLARE_DYNAMIC_LOAD_CURAND_WRAP(__name) \
struct DynLoad__##__name { \
template <typename... Args> \
curandStatus_t operator()(Args... args) { \
typedef curandStatus_t (*curandFunc)(Args...); \
std::call_once(curand_dso_flag, \
paddle::platform::dynload::GetCurandDsoHandle, \
&curand_dso_handle); \
void *p_##__name = dlsym(curand_dso_handle, #__name); \
return reinterpret_cast<curandFunc>(p_##__name)(args...); \
} \
}; \
#define DECLARE_DYNAMIC_LOAD_CURAND_WRAP(__name) \
struct DynLoad__##__name { \
template <typename... Args> \
curandStatus_t operator()(Args... args) { \
typedef curandStatus_t (*curandFunc)(Args...); \
std::call_once(curand_dso_flag, []() { \
curand_dso_handle = paddle::platform::dynload::GetCurandDsoHandle(); \
}); \
void *p_##__name = dlsym(curand_dso_handle, #__name); \
return reinterpret_cast<curandFunc>(p_##__name)(args...); \
} \
}; \
extern DynLoad__##__name __name
#else
#define DECLARE_DYNAMIC_LOAD_CURAND_WRAP(__name) \
......
paddle/fluid/platform/dynload/dynamic_loader.cc
浏览文件 @ 25ad6884
......@@ -11,12 +11,14 @@ distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License. */
#include "paddle/fluid/platform/dynload/dynamic_loader.h"
#include <dlfcn.h>
#include <memory>
#include <mutex>
#include <mutex> // NOLINT
#include <string>
#include "gflags/gflags.h"
#include "glog/logging.h"
#include "paddle/fluid/platform/dynload/cupti_lib_path.h"
......@@ -65,22 +67,21 @@ static inline std::string join(const std::string& part1,
return ret;
}
static inline void GetDsoHandleFromDefaultPath(std::string& dso_path,
void** dso_handle,
int dynload_flags) {
static inline void* GetDsoHandleFromDefaultPath(const std::string& dso_path,
int dynload_flags) {
VLOG(3) << "Try to find library: " << dso_path
<< " from default system path.";
// default search from LD_LIBRARY_PATH/DYLD_LIBRARY_PATH
*dso_handle = dlopen(dso_path.c_str(), dynload_flags);
void* dso_handle = dlopen(dso_path.c_str(), dynload_flags);
// DYLD_LIBRARY_PATH is disabled after Mac OS 10.11 to
// bring System Integrity Projection (SIP), if dso_handle
// is null, search from default package path in Mac OS.
#if defined(__APPLE__) || defined(__OSX__)
if (nullptr == *dso_handle) {
dso_path = join("/usr/local/cuda/lib/", dso_path);
*dso_handle = dlopen(dso_path.c_str(), dynload_flags);
if (nullptr == *dso_handle) {
if (nullptr == dso_handle) {
dso_handle =
dlopen(join("/usr/local/cuda/lib/", dso_path).c_str(), dynload_flags);
if (nullptr == dso_handle) {
if (dso_path == "libcudnn.dylib") {
LOG(WARNING) << "Note: [Recommend] copy cudnn into /usr/local/cuda/ \n "
"For instance, sudo tar -xzf "
......@@ -91,28 +92,29 @@ static inline void GetDsoHandleFromDefaultPath(std::string& dso_path,
}
}
#endif
return dso_handle;
}
static inline void GetDsoHandleFromSearchPath(const std::string& search_root,
const std::string& dso_name,
void** dso_handle,
bool throw_on_error = true) {
static inline void* GetDsoHandleFromSearchPath(const std::string& search_root,
const std::string& dso_name,
bool throw_on_error = true) {
int dynload_flags = RTLD_LAZY | RTLD_LOCAL;
*dso_handle = nullptr;
void* dso_handle = nullptr;
std::string dlPath = dso_name;
if (search_root.empty()) {
GetDsoHandleFromDefaultPath(dlPath, dso_handle, dynload_flags);
dso_handle = GetDsoHandleFromDefaultPath(dlPath, dynload_flags);
} else {
// search xxx.so from custom path
dlPath = join(search_root, dso_name);
*dso_handle = dlopen(dlPath.c_str(), dynload_flags);
dso_handle = dlopen(dlPath.c_str(), dynload_flags);
// if not found, search from default path
if (nullptr == *dso_handle) {
if (nullptr == dso_handle) {
LOG(WARNING) << "Failed to find dynamic library: " << dlPath << " ("
<< dlerror() << ")";
dlPath = dso_name;
GetDsoHandleFromDefaultPath(dlPath, dso_handle, dynload_flags);
dso_handle = GetDsoHandleFromDefaultPath(dlPath, dynload_flags);
}
}
auto error_msg =
......@@ -124,70 +126,71 @@ static inline void GetDsoHandleFromSearchPath(const std::string& search_root,
"using the DYLD_LIBRARY_PATH is impossible unless System "
"Integrity Protection (SIP) is disabled.";
if (throw_on_error) {
PADDLE_ENFORCE(nullptr != *dso_handle, error_msg, dlPath, dlerror());
} else if (nullptr == *dso_handle) {
PADDLE_ENFORCE(nullptr != dso_handle, error_msg, dlPath, dlerror());
} else if (nullptr == dso_handle) {
LOG(WARNING) << string::Sprintf(error_msg, dlPath, dlerror());
}
return dso_handle;
}
void GetCublasDsoHandle(void** dso_handle) {
void* GetCublasDsoHandle() {
#if defined(__APPLE__) || defined(__OSX__)
GetDsoHandleFromSearchPath(FLAGS_cuda_dir, "libcublas.dylib", dso_handle);
return GetDsoHandleFromSearchPath(FLAGS_cuda_dir, "libcublas.dylib");
#else
GetDsoHandleFromSearchPath(FLAGS_cuda_dir, "libcublas.so", dso_handle);
return GetDsoHandleFromSearchPath(FLAGS_cuda_dir, "libcublas.so");
#endif
}
void GetCUDNNDsoHandle(void** dso_handle) {
void* GetCUDNNDsoHandle() {
#if defined(__APPLE__) || defined(__OSX__)
GetDsoHandleFromSearchPath(FLAGS_cudnn_dir, "libcudnn.dylib", dso_handle,
false);
return GetDsoHandleFromSearchPath(FLAGS_cudnn_dir, "libcudnn.dylib", false);
#else
GetDsoHandleFromSearchPath(FLAGS_cudnn_dir, "libcudnn.so", dso_handle, false);
return GetDsoHandleFromSearchPath(FLAGS_cudnn_dir, "libcudnn.so", false);
#endif
}
void GetCUPTIDsoHandle(void** dso_handle) {
void* GetCUPTIDsoHandle() {
std::string cupti_path = cupti_lib_path;
if (!FLAGS_cupti_dir.empty()) {
cupti_path = FLAGS_cupti_dir;
}
#if defined(__APPLE__) || defined(__OSX__)
GetDsoHandleFromSearchPath(cupti_path, "libcupti.dylib", dso_handle, false);
return GetDsoHandleFromSearchPath(cupti_path, "libcupti.dylib", false);
#else
GetDsoHandleFromSearchPath(cupti_path, "libcupti.so", dso_handle, false);
return GetDsoHandleFromSearchPath(cupti_path, "libcupti.so", false);
#endif
}
void GetCurandDsoHandle(void** dso_handle) {
void* GetCurandDsoHandle() {
#if defined(__APPLE__) || defined(__OSX__)
GetDsoHandleFromSearchPath(FLAGS_cuda_dir, "libcurand.dylib", dso_handle);
return GetDsoHandleFromSearchPath(FLAGS_cuda_dir, "libcurand.dylib");
#else
GetDsoHandleFromSearchPath(FLAGS_cuda_dir, "libcurand.so", dso_handle);
return GetDsoHandleFromSearchPath(FLAGS_cuda_dir, "libcurand.so");
#endif
}
void GetWarpCTCDsoHandle(void** dso_handle) {
void* GetWarpCTCDsoHandle() {
#if defined(__APPLE__) || defined(__OSX__)
GetDsoHandleFromSearchPath(FLAGS_warpctc_dir, "libwarpctc.dylib", dso_handle);
return GetDsoHandleFromSearchPath(FLAGS_warpctc_dir, "libwarpctc.dylib");
#else
GetDsoHandleFromSearchPath(FLAGS_warpctc_dir, "libwarpctc.so", dso_handle);
return GetDsoHandleFromSearchPath(FLAGS_warpctc_dir, "libwarpctc.so");
#endif
}
void GetLapackDsoHandle(void** dso_handle) {
void* GetLapackDsoHandle() {
#if defined(__APPLE__) || defined(__OSX__)
GetDsoHandleFromSearchPath(FLAGS_lapack_dir, "liblapacke.dylib", dso_handle);
return GetDsoHandleFromSearchPath(FLAGS_lapack_dir, "liblapacke.dylib");
#else
GetDsoHandleFromSearchPath(FLAGS_lapack_dir, "liblapacke.so", dso_handle);
return GetDsoHandleFromSearchPath(FLAGS_lapack_dir, "liblapacke.so");
#endif
}
void GetNCCLDsoHandle(void** dso_handle) {
void* GetNCCLDsoHandle() {
#if defined(__APPLE__) || defined(__OSX__)
GetDsoHandleFromSearchPath(FLAGS_nccl_dir, "libnccl.dylib", dso_handle);
return GetDsoHandleFromSearchPath(FLAGS_nccl_dir, "libnccl.dylib");
#else
GetDsoHandleFromSearchPath(FLAGS_nccl_dir, "libnccl.so", dso_handle);
return GetDsoHandleFromSearchPath(FLAGS_nccl_dir, "libnccl.so");
#endif
}
......
paddle/fluid/platform/dynload/dynamic_loader.h
浏览文件 @ 25ad6884
......@@ -18,55 +18,13 @@ namespace paddle {
namespace platform {
namespace dynload {
/**
* @brief load the DSO of CUBLAS
*
* @param **dso_handle dso handler
*
*/
void GetCublasDsoHandle(void** dso_handle);
/**
* @brief load the DSO of CUDNN
*
* @param **dso_handle dso handler
*
*/
void GetCUDNNDsoHandle(void** dso_handle);
void GetCUPTIDsoHandle(void** dso_handle);
/**
* @brief load the DSO of CURAND
*
* @param **dso_handle dso handler
*
*/
void GetCurandDsoHandle(void** dso_handle);
/**
* @brief load the DSO of warp-ctc
*
* @param **dso_handle dso handler
*
*/
void GetWarpCTCDsoHandle(void** dso_handle);
/**
* @brief load the DSO of lapack
*
* @param **dso_handle dso handler
*
*/
void GetLapackDsoHandle(void** dso_handle);
/**
* @brief load the DSO of NVIDIA nccl
*
* @param **dso_handle dso handler
*
*/
void GetNCCLDsoHandle(void** dso_handle);
void* GetCublasDsoHandle();
void* GetCUDNNDsoHandle();
void* GetCUPTIDsoHandle();
void* GetCurandDsoHandle();
void* GetWarpCTCDsoHandle();
void* GetLapackDsoHandle();
void* GetNCCLDsoHandle();
} // namespace dynload
} // namespace platform
......
paddle/fluid/platform/dynload/nccl.cc
浏览文件 @ 25ad6884
......@@ -25,11 +25,6 @@ void *nccl_dso_handle;
NCCL_RAND_ROUTINE_EACH(DEFINE_WRAP);
void LoadNCCLDSO() {
platform::call_once(nccl_dso_flag,
[] { GetNCCLDsoHandle(&nccl_dso_handle); });
}
} // namespace dynload
} // namespace platform
} // namespace paddle
paddle/fluid/platform/dynload/nccl.h
浏览文件 @ 25ad6884
......@@ -11,12 +11,13 @@ distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License. */
#pragma once
#include <dlfcn.h>
#include <nccl.h>
#include <mutex>
#include <mutex> // NOLINT
#include "paddle/fluid/platform/call_once.h"
#include "paddle/fluid/platform/dynload/dynamic_loader.h"
......@@ -28,18 +29,19 @@ extern std::once_flag nccl_dso_flag;
extern void* nccl_dso_handle;
#ifdef PADDLE_USE_DSO
extern void LoadNCCLDSO();
#define DECLARE_DYNAMIC_LOAD_NCCL_WRAP(__name) \
struct DynLoad__##__name { \
template <typename... Args> \
auto operator()(Args... args) -> decltype(__name(args...)) { \
using nccl_func = decltype(__name(args...)) (*)(Args...); \
paddle::platform::dynload::LoadNCCLDSO(); \
void* p_##__name = dlsym(nccl_dso_handle, #__name); \
return reinterpret_cast<nccl_func>(p_##__name)(args...); \
} \
}; \
#define DECLARE_DYNAMIC_LOAD_NCCL_WRAP(__name) \
struct DynLoad__##__name { \
template <typename... Args> \
auto operator()(Args... args) -> decltype(__name(args...)) { \
using nccl_func = decltype(__name(args...)) (*)(Args...); \
std::call_once(nccl_dso_flag, []() { \
nccl_dso_handle = paddle::platform::dynload::GetNCCLDsoHandle(); \
}); \
void* p_##__name = dlsym(nccl_dso_handle, #__name); \
return reinterpret_cast<nccl_func>(p_##__name)(args...); \
} \
}; \
extern DynLoad__##__name __name
#else
#define DECLARE_DYNAMIC_LOAD_NCCL_WRAP(__name) \
......
paddle/fluid/platform/dynload/warpctc.h
浏览文件 @ 25ad6884
......@@ -15,9 +15,10 @@ limitations under the License. */
#pragma once
#include <dlfcn.h>
#include <mutex>
#include "ctc.h"
#include <mutex> // NOLINT
#include "paddle/fluid/platform/dynload/dynamic_loader.h"
#include "warpctc/include/ctc.h"
namespace paddle {
namespace platform {
......@@ -31,18 +32,18 @@ extern void* warpctc_dso_handle;
* (for each function) to dynamic load warpctc routine
* via operator overloading.
*/
#define DYNAMIC_LOAD_WARPCTC_WRAP(__name) \
struct DynLoad__##__name { \
template <typename... Args> \
auto operator()(Args... args) -> decltype(__name(args...)) { \
using warpctcFunc = decltype(__name(args...)) (*)(Args...); \
std::call_once(warpctc_dso_flag, \
paddle::platform::dynload::GetWarpCTCDsoHandle, \
&warpctc_dso_handle); \
void* p_##_name = dlsym(warpctc_dso_handle, #__name); \
return reinterpret_cast<warpctcFunc>(p_##_name)(args...); \
} \
}; \
#define DYNAMIC_LOAD_WARPCTC_WRAP(__name) \
struct DynLoad__##__name { \
template <typename... Args> \
auto operator()(Args... args) -> decltype(__name(args...)) { \
using warpctcFunc = decltype(__name(args...)) (*)(Args...); \
std::call_once(warpctc_dso_flag, []() { \
warpctc_dso_handle = paddle::platform::dynload::GetWarpCTCDsoHandle(); \
}); \
void* p_##_name = dlsym(warpctc_dso_handle, #__name); \
return reinterpret_cast<warpctcFunc>(p_##_name)(args...); \
} \
}; \
extern DynLoad__##__name __name
#define DECLARE_DYNAMIC_LOAD_WARPCTC_WRAP(__name) \
......
paddle/fluid/platform/gpu_info.h
浏览文件 @ 25ad6884
......@@ -23,9 +23,6 @@ limitations under the License. */
namespace paddle {
namespace platform {
//! Environment variable: fraction of GPU memory to use on each device.
const char kEnvFractionGpuMemoryToUse[] = "PADDLE_FRACTION_GPU_MEMORY_TO_USE";
//! Get the total number of GPU devices in system.
int GetCUDADeviceCount();
......
paddle/fluid/pybind/tensor_py.h
浏览文件 @ 25ad6884
......@@ -13,6 +13,7 @@ See the License for the specific language governing permissions and
limitations under the License. */
#pragma once
#include <Python.h>
#include <string>
#include <tuple>
#include <vector>
......
python/paddle/fluid/debuger.py
浏览文件 @ 25ad6884
......@@ -16,6 +16,7 @@ import sys
import re
from graphviz import GraphPreviewGenerator
import proto.framework_pb2 as framework_pb2
from google.protobuf import text_format
_vartype2str_ = [
"UNK",
......@@ -100,7 +101,7 @@ def repr_var(vardesc):
def pprint_program_codes(program_desc):
reprs = []
for block_idx in range(program_desc.num_blocks()):
for block_idx in range(program_desc.desc.num_blocks()):
block_desc = program_desc.block(block_idx)
block_repr = pprint_block_codes(block_desc)
reprs.append(block_repr)
......@@ -127,7 +128,7 @@ def pprint_block_codes(block_desc, show_backward=False):
if type(block_desc) is not framework_pb2.BlockDesc:
block_desc = framework_pb2.BlockDesc.FromString(
block_desc.serialize_to_string())
block_desc.desc.serialize_to_string())
var_reprs = []
op_reprs = []
for var in block_desc.vars:
......@@ -237,13 +238,13 @@ def draw_block_graphviz(block, highlights=None, path="./temp.dot"):
# draw parameters and args
vars = {}
for var in desc.vars:
shape = [str(i) for i in var.lod_tensor.tensor.dims]
if not shape:
shape = ['null']
# TODO(gongwb): format the var.type
# create var
if var.persistable:
varn = graph.add_param(
var.name, var.type, shape, highlight=need_highlight(var.name))
var.name,
str(var.type).replace("\n", "<br />", 1),
highlight=need_highlight(var.name))
else:
varn = graph.add_arg(var.name, highlight=need_highlight(var.name))
vars[var.name] = varn
......@@ -268,4 +269,4 @@ def draw_block_graphviz(block, highlights=None, path="./temp.dot"):
for var in op.outputs:
add_op_link_var(opn, var, True)
graph(path, show=True)
graph(path, show=False)
python/paddle/fluid/framework.py
浏览文件 @ 25ad6884
......@@ -640,6 +640,20 @@ class Operator(object):
"""
return self.desc.block_attr(name)
def all_attrs(self):
"""
Get the attribute dict
Returns(dict): The Operator's attribute dict
"""
attr_names = self.attr_names
attr_map = {}
for n in attr_names:
if n == 'sub_block':
attr_map[n] = self.block_attr(n)
else:
attr_map[n] = self.attr(n)
return attr_map
class Block(object):
def __init__(self, program, idx):
......@@ -838,7 +852,7 @@ class Block(object):
def sync_with_cpp(self):
"""
Sync with the desc on the c++ end.
Sync from the desc on the c++ end.
This method is used to synchronize the c++ desc instance generated by backward.
"""
......@@ -951,6 +965,13 @@ class Block(object):
if var.type == core.VarDesc.VarType.STEP_SCOPES:
ret_var = self.create_var(
name=var.name, persistable=var.persistable, type=var.type)
elif var.type == core.VarDesc.VarType.SELECTED_ROWS:
ret_var = self.create_var(
name=var.name,
shape=var.shape,
dtype=var.dtype,
type=var.type,
persistable=True)
else:
ret_var = self.create_var(
name=var.name,
......
python/paddle/fluid/graphviz.py
浏览文件 @ 25ad6884
......@@ -83,7 +83,7 @@ class Graph(object):
file = open(dot_path, 'w')
file.write(self.__str__())
image_path = os.path.join(
os.path.dirname(__file__), dot_path[:-3] + "pdf")
os.path.dirname(dot_path), dot_path[:-3] + "pdf")
cmd = ["dot", "-Tpdf", dot_path, "-o", image_path]
subprocess.Popen(
cmd,
......@@ -199,7 +199,7 @@ class GraphPreviewGenerator(object):
else:
self.graph.show(path)
def add_param(self, name, data_type, shape, highlight=False):
def add_param(self, name, data_type, highlight=False):
label = '\n'.join([
'<<table cellpadding="5">',
' <tr>',
......@@ -214,11 +214,6 @@ class GraphPreviewGenerator(object):
str(data_type),
' </td>'
' </tr>',
' <tr>',
' <td>',
'[%s]' % 'x'.join(shape),
' </td>'
' </tr>',
'</table>>',
])
return self.graph.node(
......
python/paddle/fluid/layers/io.py
浏览文件 @ 25ad6884
......@@ -255,7 +255,32 @@ def _copy_reader_var_(block, var):
new_var.desc.set_shapes(var.desc.shapes())
new_var.desc.set_dtypes(var.desc.dtypes())
new_var.persistable = True
return monkey_patch_reader_methods(new_var)
return new_var
def _copy_reader_create_op_(block, op):
input_param_names = op.input_names
new_input_map = {}
for param_name in input_param_names:
new_input_map[param_name] = []
arg_names = op.input(param_name)
for arg_name in arg_names:
new_input_map[param_name].append(block.var(arg_name))
output_param_names = op.output_names
new_output_map = {}
for param_name in output_param_names:
new_output_map[param_name] = []
arg_names = op.output(param_name)
for arg_name in arg_names:
new_output_map[param_name].append(block.var(arg_name))
new_op = block.append_op(
type=op.type,
inputs=new_input_map,
outputs=new_output_map,
attrs=op.all_attrs())
return new_op
def open_recordio_file(filename, shapes, lod_levels, dtypes):
......@@ -283,8 +308,9 @@ def open_recordio_file(filename, shapes, lod_levels, dtypes):
startup_var.desc.set_dtypes(dtypes)
startup_var.persistable = True
return _copy_reader_var_(default_main_program().current_block(),
startup_var)
main_prog_var = _copy_reader_var_(default_main_program().current_block(),
startup_var)
return monkey_patch_reader_methods(main_prog_var)
def open_files(filenames, thread_num, shapes, lod_levels, dtypes):
......@@ -313,22 +339,25 @@ def open_files(filenames, thread_num, shapes, lod_levels, dtypes):
startup_var.desc.set_dtypes(dtypes)
startup_var.persistable = True
return _copy_reader_var_(default_main_program().current_block(),
startup_var)
main_prog_var = _copy_reader_var_(default_main_program().current_block(),
startup_var)
return monkey_patch_reader_methods(main_prog_var)
def __create_decorated_reader__(op_type, reader, attrs):
var_name = unique_name(op_type)
startup_blk = default_startup_program().current_block()
startup_var = startup_blk.create_var(name=var_name)
startup_blk.append_op(
startop_op = startup_blk.append_op(
type=op_type,
inputs={'UnderlyingReader': reader},
outputs={'Out': [startup_var]},
attrs=attrs)
startup_var.persistable = True
return _copy_reader_var_(default_main_program().current_block(),
startup_var)
main_prog_block = default_main_program().current_block()
main_prog_var = _copy_reader_var_(main_prog_block, startup_var)
_copy_reader_create_op_(main_prog_block, startop_op)
return monkey_patch_reader_methods(main_prog_var)
def create_shuffle_reader(reader, buffer_size):
......
python/paddle/fluid/tests/unittests/test_debugger.py
浏览文件 @ 25ad6884
......@@ -51,7 +51,9 @@ class TestDebugger(unittest.TestCase):
outputs={"Out": mul_out},
attrs={"x_num_col_dims": 1})
print(debuger.pprint_program_codes(p.desc))
print(debuger.pprint_program_codes(p))
debuger.draw_block_graphviz(p.block(0), path="./test.dot")
if __name__ == '__main__':
......
python/paddle/fluid/tests/unittests/test_recordio_reader.py
浏览文件 @ 25ad6884
......@@ -15,8 +15,8 @@
import unittest
import paddle.fluid as fluid
import paddle
import paddle.dataset.mnist as mnist
import paddle.v2 as paddle
import paddle.v2.dataset.mnist as mnist
class TestRecordIO(unittest.TestCase):
......
python/setup.py.in
浏览文件 @ 25ad6884
......@@ -107,6 +107,7 @@ package_dir={
# So that package points to other directory.
'paddle.fluid.proto.profiler': '${PADDLE_BINARY_DIR}/paddle/fluid/platform',
'paddle.fluid.proto': '${PADDLE_BINARY_DIR}/paddle/fluid/framework',
'paddle.fluid': '${PADDLE_BINARY_DIR}/python/paddle/fluid',
}
if '${WITH_FLUID_ONLY}'== 'OFF':
package_dir['py_paddle']='${PADDLE_BINARY_DIR}/python/py_paddle'
......
Markdown is supported
0% 或 .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑！
想要评论请 注册 或