你好，请教个关于并行计算的问题！ about parl HOT 19 CLOSED

你好，感谢对PARL的关注。
1.关于并行这块。PARL目前已经有并行通讯的接口，通过这个接口来实现并行算法是支持跨系统以及跨平台的。我们自己也有使用GPU训练（centos系统） + CPU仿真（Ubuntu系统）的情况。
2.目前PARL还不支持Windows系统，主要在于Windows的命令管理方式和Linux存在较大差距，这个会在未来的版本中解决(#178)。

from parl.

谢谢，这是秒回复啊，我刚才描述的架构模式下有一个地方没有提到，就是在CPU集群中没有parl在运行，cpu集群只是提供仿真环境的运行，不过强化学习算法可以通过grpc启动仿真环境并与仿真环境交互，不知道这样会不会影响PARL的并行运行？

from parl.

噢，这个是不影响的，有问题可以随时抛出来：）

from parl.

你好，PARL可以在多GPU卡上实现神经网络模型并行计算么？

from parl.

可以的！
我们注意到离线训练(offline training)的时候，经常有训练上亿级别的数据需求。所以提供了example来举例如何使用。实际上通过一行代码就可以完成转换了～
具体参考这个PR（#193），马上就会合并到主干。

from parl.

你好，我看到有parl.compile来完成并行计算，可以简单介绍下这个调用在多机多卡上并行计算的原理么？谢谢

from parl.

Paddle的实现是data-parallel，就是把一个大batch的数据分发的不同的卡上分别做计算，各自计算梯度并通过all reduce的操作合并梯度，把梯度刷到每块卡上，更新其模型参数。
请关注下PARL哈，后续还会有更多新example，包括multi-agent上的工作。

from parl.

喔，那我理解这应该是数据并行，而不是模型并行，就是说把一个大模型拆成几块放到不同机器的GPU上跑，不知我得理解对不对？
嗯，我对PARL这块很关注，我也了解其他比如rllib、rlpyt等框架，对多智能体的支持还是很需要的。

from parl.

喔，那我理解这应该是数据并行，而不是模型并行，就是说把一个大模型拆成几块放到不同机器的GPU上跑，不知我得理解对不对？

是的，数据并行。

感谢关注！

from parl.

这个数据并行机制底层是依托paddle实现的么？可以用pytorch来实现parl数据并行么？

from parl.

这种对计算性能要求高的并行计算是依托Paddle底层的C++、Cuda代码实现的哈，所以torch用不了这部分的工作。

from parl.

我看了下要合并的并行计算文件，这种并行要是在多机上实现，在什么地方设置多机多卡地址呢？

from parl.

目前我们只提供单机多卡的RL计算example，多机多卡这块还没有提供example的打算。
这块paddle目前已经支持的了，具体可以参考paddle的文档。

from parl.

请问有没有结合PaddlePaddle EDL 和PARL的案例？

from parl.

目前没有哈，有使用问题可以抛出来我们再解决：）

from parl.

我看了下代码，目前parl的并行计算所利用的资源还是CPU吧？没有涉及GPU吧

from parl.

是的，目前针对CPU

from parl.

目前在python3.6和python3.7环境下安装parl总是报错误，有关于psutil，请问如何解决？
Building wheels for collected packages: psutil
Building wheel for psutil (setup.py) ... error
ERROR: Command errored out with exit status 1:
command: /home/sunfangyi/virtualenv/PARL3.7/bin/python3.7 -u -c 'import sys, setuptools, tokenize; sys.argv[0] = '"'"'/tmp/pip-install-2vpoancx/psutil/setup.py'"'"'; file='"'"'/tmp/pip-install-2vpoancx/psutil/setup.py'"'"';f=getattr(tokenize, '"'"'open'"'"', open)(file);code=f.read().replace('"'"'\r\n'"'"', '"'"'\n'"'"');f.close();exec(compile(code, file, '"'"'exec'"'"'))' bdist_wheel -d /tmp/pip-wheel-0xgcqsrp
cwd: /tmp/pip-install-2vpoancx/psutil/
Complete output (42 lines):
running bdist_wheel
running build
running build_py
creating build
creating build/lib.linux-x86_64-3.7
creating build/lib.linux-x86_64-3.7/psutil
copying psutil/_pssunos.py -> build/lib.linux-x86_64-3.7/psutil
copying psutil/_psposix.py -> build/lib.linux-x86_64-3.7/psutil
copying psutil/_psaix.py -> build/lib.linux-x86_64-3.7/psutil
copying psutil/_psbsd.py -> build/lib.linux-x86_64-3.7/psutil
copying psutil/_common.py -> build/lib.linux-x86_64-3.7/psutil
copying psutil/init.py -> build/lib.linux-x86_64-3.7/psutil
copying psutil/_psosx.py -> build/lib.linux-x86_64-3.7/psutil
copying psutil/_compat.py -> build/lib.linux-x86_64-3.7/psutil
copying psutil/_pswindows.py -> build/lib.linux-x86_64-3.7/psutil
copying psutil/_pslinux.py -> build/lib.linux-x86_64-3.7/psutil
creating build/lib.linux-x86_64-3.7/psutil/tests
copying psutil/tests/test_process.py -> build/lib.linux-x86_64-3.7/psutil/tests
copying psutil/tests/test_aix.py -> build/lib.linux-x86_64-3.7/psutil/tests
copying psutil/tests/test_linux.py -> build/lib.linux-x86_64-3.7/psutil/tests
copying psutil/tests/runner.py -> build/lib.linux-x86_64-3.7/psutil/tests
copying psutil/tests/test_osx.py -> build/lib.linux-x86_64-3.7/psutil/tests
copying psutil/tests/test_contracts.py -> build/lib.linux-x86_64-3.7/psutil/tests
copying psutil/tests/test_bsd.py -> build/lib.linux-x86_64-3.7/psutil/tests
copying psutil/tests/test_system.py -> build/lib.linux-x86_64-3.7/psutil/tests
copying psutil/tests/test_posix.py -> build/lib.linux-x86_64-3.7/psutil/tests
copying psutil/tests/init.py -> build/lib.linux-x86_64-3.7/psutil/tests
copying psutil/tests/main.py -> build/lib.linux-x86_64-3.7/psutil/tests
copying psutil/tests/test_unicode.py -> build/lib.linux-x86_64-3.7/psutil/tests
copying psutil/tests/test_connections.py -> build/lib.linux-x86_64-3.7/psutil/tests
copying psutil/tests/test_misc.py -> build/lib.linux-x86_64-3.7/psutil/tests
copying psutil/tests/test_memory_leaks.py -> build/lib.linux-x86_64-3.7/psutil/tests
copying psutil/tests/test_windows.py -> build/lib.linux-x86_64-3.7/psutil/tests
copying psutil/tests/test_sunos.py -> build/lib.linux-x86_64-3.7/psutil/tests
running build_ext
building 'psutil._psutil_linux' extension
creating build/temp.linux-x86_64-3.7
creating build/temp.linux-x86_64-3.7/psutil
x86_64-linux-gnu-gcc -pthread -Wno-unused-result -Wsign-compare -DNDEBUG -g -fwrapv -O2 -Wall -g -fstack-protector-strong -Wformat -Werror=format-security -g -fstack-protector-strong -Wformat -Werror=format-security -Wdate-time -D_FORTIFY_SOURCE=2 -fPIC -DPSUTIL_POSIX=1 -DPSUTIL_VERSION=567 -DPSUTIL_LINUX=1 -I/usr/include/python3.7m -I/home/sunfangyi/virtualenv/PARL3.7/include/python3.7m -c psutil/_psutil_common.c -o build/temp.linux-x86_64-3.7/psutil/_psutil_common.o
psutil/_psutil_common.c:9:20: fatal error: Python.h: No such file or directory
compilation terminated.
error: command 'x86_64-linux-gnu-gcc' failed with exit status 1

ERROR: Failed building wheel for psutil
Running setup.py clean for psutil
Failed to build psutil
Installing collected packages: psutil, pyzmq, click, protobuf, numpy, tensorboardX, absl-py, werkzeug, grpcio, markdown, tb-nightly, termcolor, cloudpickle, Jinja2, itsdangerous, flask, scipy, pyarrow, parl
Running setup.py install for psutil ... error
ERROR: Command errored out with exit status 1:
command: /home/sunfangyi/virtualenv/PARL3.7/bin/python3.7 -u -c 'import sys, setuptools, tokenize; sys.argv[0] = '"'"'/tmp/pip-install-2vpoancx/psutil/setup.py'"'"'; file='"'"'/tmp/pip-install-2vpoancx/psutil/setup.py'"'"';f=getattr(tokenize, '"'"'open'"'"', open)(file);code=f.read().replace('"'"'\r\n'"'"', '"'"'\n'"'"');f.close();exec(compile(code, file, '"'"'exec'"'"'))' install --record /tmp/pip-record-_4trv25m/install-record.txt --single-version-externally-managed --compile --install-headers /home/sunfangyi/virtualenv/PARL3.7/include/site/python3.7/psutil
cwd: /tmp/pip-install-2vpoancx/psutil/
Complete output (42 lines):
running install
running build
running build_py
creating build
creating build/lib.linux-x86_64-3.7
creating build/lib.linux-x86_64-3.7/psutil
copying psutil/_pssunos.py -> build/lib.linux-x86_64-3.7/psutil
copying psutil/_psposix.py -> build/lib.linux-x86_64-3.7/psutil
copying psutil/_psaix.py -> build/lib.linux-x86_64-3.7/psutil
copying psutil/_psbsd.py -> build/lib.linux-x86_64-3.7/psutil
copying psutil/_common.py -> build/lib.linux-x86_64-3.7/psutil
copying psutil/init.py -> build/lib.linux-x86_64-3.7/psutil
copying psutil/_psosx.py -> build/lib.linux-x86_64-3.7/psutil
copying psutil/_compat.py -> build/lib.linux-x86_64-3.7/psutil
copying psutil/_pswindows.py -> build/lib.linux-x86_64-3.7/psutil
copying psutil/_pslinux.py -> build/lib.linux-x86_64-3.7/psutil
creating build/lib.linux-x86_64-3.7/psutil/tests
copying psutil/tests/test_process.py -> build/lib.linux-x86_64-3.7/psutil/tests
copying psutil/tests/test_aix.py -> build/lib.linux-x86_64-3.7/psutil/tests
copying psutil/tests/test_linux.py -> build/lib.linux-x86_64-3.7/psutil/tests
copying psutil/tests/runner.py -> build/lib.linux-x86_64-3.7/psutil/tests
copying psutil/tests/test_osx.py -> build/lib.linux-x86_64-3.7/psutil/tests
copying psutil/tests/test_contracts.py -> build/lib.linux-x86_64-3.7/psutil/tests
copying psutil/tests/test_bsd.py -> build/lib.linux-x86_64-3.7/psutil/tests
copying psutil/tests/test_system.py -> build/lib.linux-x86_64-3.7/psutil/tests
copying psutil/tests/test_posix.py -> build/lib.linux-x86_64-3.7/psutil/tests
copying psutil/tests/init.py -> build/lib.linux-x86_64-3.7/psutil/tests
copying psutil/tests/main.py -> build/lib.linux-x86_64-3.7/psutil/tests
copying psutil/tests/test_unicode.py -> build/lib.linux-x86_64-3.7/psutil/tests
copying psutil/tests/test_connections.py -> build/lib.linux-x86_64-3.7/psutil/tests
copying psutil/tests/test_misc.py -> build/lib.linux-x86_64-3.7/psutil/tests
copying psutil/tests/test_memory_leaks.py -> build/lib.linux-x86_64-3.7/psutil/tests
copying psutil/tests/test_windows.py -> build/lib.linux-x86_64-3.7/psutil/tests
copying psutil/tests/test_sunos.py -> build/lib.linux-x86_64-3.7/psutil/tests
running build_ext
building 'psutil._psutil_linux' extension
creating build/temp.linux-x86_64-3.7
creating build/temp.linux-x86_64-3.7/psutil
x86_64-linux-gnu-gcc -pthread -Wno-unused-result -Wsign-compare -DNDEBUG -g -fwrapv -O2 -Wall -g -fstack-protector-strong -Wformat -Werror=format-security -g -fstack-protector-strong -Wformat -Werror=format-security -Wdate-time -D_FORTIFY_SOURCE=2 -fPIC -DPSUTIL_POSIX=1 -DPSUTIL_VERSION=567 -DPSUTIL_LINUX=1 -I/usr/include/python3.7m -I/home/sunfangyi/virtualenv/PARL3.7/include/python3.7m -c psutil/_psutil_common.c -o build/temp.linux-x86_64-3.7/psutil/_psutil_common.o
psutil/_psutil_common.c:9:20: fatal error: Python.h: No such file or directory
compilation terminated.
error: command 'x86_64-linux-gnu-gcc' failed with exit status 1
----------------------------------------
ERROR: Command errored out with exit status 1: /home/sunfangyi/virtualenv/PARL3.7/bin/python3.7 -u -c 'import sys, setuptools, tokenize; sys.argv[0] = '"'"'/tmp/pip-install-2vpoancx/psutil/setup.py'"'"'; file='"'"'/tmp/pip-install-2vpoancx/psutil/setup.py'"'"';f=getattr(tokenize, '"'"'open'"'"', open)(file);code=f.read().replace('"'"'\r\n'"'"', '"'"'\n'"'"');f.close();exec(compile(code, file, '"'"'exec'"'"'))' install --record /tmp/pip-record-_4trv25m/install-record.txt --single-version-externally-managed --compile --install-headers /home/sunfangyi/virtualenv/PARL3.7/include/site/python3.7/psutil Check the logs for full command output.

from parl.

请问是在什么系统环境下安装的？（linux/macos/windows?）

from parl.