hi，请问是遇到新问题了吗 看昨天的import paddle的DLL load failed的问题解决了。执行下 nvidia-smi 看看

是的，新问题。昨天的问题解决了。上个问题我已经关闭，非常感谢帮助。

看不太懂这个输出的含义

C:\Program Files\NVIDIA Corporation\NVSMI>nvidia-smi
Tue Feb 25 12:00:43 2020
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 392.58 Driver Version: 392.58 |
|-------------------------------+----------------------+----------------------+
| GPU Name TCC/WDDM | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
|===============================+======================+======================|
| 0 NVS 5400M WDDM | 00000000:01:00.0 N/A | N/A |
| N/A 44C P0 N/A / N/A | 54MiB / 1024MiB | N/A Default |
+-------------------------------+----------------------+----------------------+

+-----------------------------------------------------------------------------+
| Processes: GPU Memory |
| GPU PID Type Process name Usage |
|=============================================================================|
| 0 Not Supported |
+-----------------------------------------------------------------------------+

我的机器上安装过CUDA 10.2, 后来感觉dll import有问题时，就补充安装了10.0，2个版本都增补了对应cudnn. 在昨天重装python3.7,解决了 dll 问题后。我调整了path的设置，只保留了 10.0 CUDA的路径。
显卡驱动曾经重新安装过，没看出什么报错；
唯一可能的问题是，我的机器是T530，有2个显卡，Intel HD graphics 4000, Nvidia NVS 5400M.
paddle 是怎么决定用哪个显卡的？

可能是因为有多个卡的原因，设置一下环境变量set FLAGS_selected_gpus=0,1 或者 set CUDA_VISIBLE_DEVICES=0,1试试。

跑的时候设置一下 set GLOG_v=5，再看一下有没更完整的报错信息内容，目前这个"but no GPU found in runtime."有点简略

增加了SET环境变量，就好了。应该是2个图像卡的问题。
非常感谢。

高兴太早了，还是有问题。

paddle.fluid.install_check.run_check()
Running Verify Paddle Program ...
2020-02-25 16:47:51,041-WARNING: You are using GPU version Paddle, But Your CUDA Device is not set properly
Original Error is

C++ Call Stacks (More useful to developers):

Windows not support stack backtrace yet.

Error Message Summary:

Error: cudaGetDeviceCount failed in paddle::platform::GetCUDADeviceCountImpl, error code : 35, Please see detail in https://docs.nvidia.com/cuda/cuda-runtime-api/group__CUDART__TYPES.html#group__CUDART__TYPES_1g3f51e3575c2178246db0a94a430e0038: CUDA driver version is insufficient for CUDA runtime version at (D:\1.7.0\paddle\paddle\fluid\platform\gpu_info.cc:72)

0

1.你这个显卡的驱动没有安装正确，建议通过 GeForce Experience 来自动下载驱动安装。手动安装驱动可能版本出错http://cn.download.nvidia.com/GFE/GFEClient/3.20.2.34/GeForce_Experience_v3.20.2.34.exe 2. 显卡硬件没有装好。大概率是第一个的问题。另外set CUDA_VISIBLE_DEVICES=0，Windows下仅支持单卡

问个事情：

1. 如果我的机器上有多个版本的 CUDA，是否可以和平共存？只需要在PATH中，调整顺序来切换希望激活的版本？
2. 与CUDA10、CUDA9，以及与CPU配合的paddle版本可以共存吗？系统是怎么决定哪个paddle版本会运营呢？
3. 如果系统有多个图像卡，如何指定哪个

安装了GeForce Experience后，自动安装的版本是版本392.58，检查CUDA10.0和驱动版本感觉不对应，好像只能支持CUDA9，所以试图回退安装9.0。
CUDA 9.0的前置安装时，安装CUDA9.0+ PATCH1/2/3/4, 自己将驱动的版本又下降到385.54.

我如何判断哪个网卡是0，哪个是1？

C:\Users\SidneyZ>python
Python 3.7.6 (tags/v3.7.6:43364a7ae0, Dec 19 2019, 00:42:30) [MSC v.1916 64 bit (AMD64)] on win32
Type "help", "copyright", "credits" or "license" for more information.

import paddle
import paddly.fluid
Traceback (most recent call last):
File "", line 1, in
ModuleNotFoundError: No module named 'paddly'
import paddle.fluid
paddle.fluid.install_check.run_check()
Running Verify Paddle Program ...
W0225 20:29:53.582741 16840 device_context.cc:237] Please NOTE: device: 0, CUDA Capability: 21, Driver API Version: 9.0, Runtime API Version: 9.0
W0225 20:29:54.531765 16840 device_context.cc:245] device: 0, cuDNN Version: 7.6.
D:\Program Files\Python\Python37\lib\site-packages\paddle\fluid\executor.py:782: UserWarning: The following exception is not an EOF exception.
"The following exception is not an EOF exception.")
Traceback (most recent call last):
File "", line 1, in
File "D:\Program Files\Python\Python37\lib\site-packages\paddle\fluid\install_check.py", line 124, in run_check
test_simple_exe()
File "D:\Program Files\Python\Python37\lib\site-packages\paddle\fluid\install_check.py", line 120, in test_simple_exe
exe0.run(startup_prog)
File "D:\Program Files\Python\Python37\lib\site-packages\paddle\fluid\executor.py", line 783, in run
six.reraise(*sys.exc_info())
File "D:\Program Files\Python\Python37\lib\site-packages\six.py", line 703, in reraise
raise value
File "D:\Program Files\Python\Python37\lib\site-packages\paddle\fluid\executor.py", line 778, in run
use_program_cache=use_program_cache)
File "D:\Program Files\Python\Python37\lib\site-packages\paddle\fluid\executor.py", line 831, in _run_impl
use_program_cache=use_program_cache)
File "D:\Program Files\Python\Python37\lib\site-packages\paddle\fluid\executor.py", line 905, in _run_program
fetch_var_name)
paddle.fluid.core_avx.EnforceNotMet:

C++ Call Stacks (More useful to developers):

Windows not support stack backtrace yet.

Python Call Stacks (More useful to users):

File "D:\Program Files\Python\Python37\lib\site-packages\paddle\fluid\framework.py", line 2594, in _prepend_op
attrs=kwargs.get("attrs", None))
File "D:\Program Files\Python\Python37\lib\site-packages\paddle\fluid\initializer.py", line 191, in call
stop_gradient=True)
File "D:\Program Files\Python\Python37\lib\site-packages\paddle\fluid\framework.py", line 2476, in create_parameter
initializer(param, self)
File "D:\Program Files\Python\Python37\lib\site-packages\paddle\fluid\layer_helper_base.py", line 353, in create_parameter
**attr._to_kwargs(with_initializer=True))
File "D:\Program Files\Python\Python37\lib\site-packages\paddle\fluid\dygraph\layers.py", line 113, in create_parameter
default_initializer)
File "D:\Program Files\Python\Python37\lib\site-packages\paddle\fluid\dygraph\nn.py", line 921, in init
shape=[output_dim], attr=bias_attr, dtype=dtype, is_bias=True)
File "D:\Program Files\Python\Python37\lib\site-packages\paddle\fluid\install_check.py", line 38, in init
param_attr=ParamAttr(initializer=Constant(value=0.1)))
File "D:\Program Files\Python\Python37\lib\site-packages\paddle\fluid\install_check.py", line 112, in test_simple_exe
simple_layer0 = SimpleLayer(input_size=2)
File "D:\Program Files\Python\Python37\lib\site-packages\paddle\fluid\install_check.py", line 124, in run_check
test_simple_exe()
File "", line 1, in

Error Message Summary:

Error: Failed to create Cudnn handle in DeviceContext
[Hint: CUDNN_STATUS_ARCH_MISMATCH] at (D:\1.7.0\paddle\paddle\fluid\platform\device_context.cc:283)
[operator < fill_constant > error]

1。 SET CUDA_VISIBLE_DEVICES=1 时，

import paddle.fluid
W0225 20:40:02.858937 15284 init.cc:127] Compiled with WITH_GPU, but no GPU found in runtime.
paddle.fluid.install_check.run_check()
Running Verify Paddle Program ...
2020-02-25 20:40:32,204-WARNING: You are using GPU version Paddle, But Your CUDA Device is not set properly

Error Message Summary:

Error: cudaGetDeviceCount failed in paddle::platform::GetCUDADeviceCountImpl, error code : 38, Please see detail in https://docs.nvidia.com/cuda/cuda-runtime-api/group__CUDART__TYPES.html#group__CUDART__TYPES_1g3f51e3575c2178246db0a94a430e0038: no CUDA-capable device is detected at (D:\1.7.0\paddle\paddle\fluid\platform\gpu_info.cc:72)

文档中的38号错误，没有说明

2. SET CUDA_VISIBLE_DEVICES=0时，

import paddle.fluid 正常
paddle.fluid.install_check.run_check()

Error Message Summary:

Error: Failed to create Cudnn handle in DeviceContext
[Hint: CUDNN_STATUS_ARCH_MISMATCH] at (D:\1.7.0\paddle\paddle\fluid\platform\device_context.cc:283)
[operator < fill_constant > error]

1.多个CUDA可以同时安装，调整CUDA_PATH 和PATH中的环境变量即可。2. 不同版本的paddle不能共同使用，Paddle是装到python里去的，只能装一个。你可以用 pip install -U paddlepaddle-gpu==1.6.3.post97 -i https://pypi.tuna.tsinghua.edu.cn/simple ，重新安装CUDA9的paddle，会卸载之前的paddle

1. 再次运行GeForce Experience, 将驱动恢复成了本机最高版本392.58
2. 使用你提供的指令，将paddle从1.7.0，重新安装退到了1.6.3，post97 CUDA9.0
   安装时，报错
   换用了百度的镜像，现象一样；
   重新卸载 pip uninstall paddlepaddle-gpu, 再安装，现象一样。
   是我下载的文件出问题了？还是我的解压出问题了？

C:\Program Files\NVIDIA Corporation\NVSMI>python -m pip install -U paddlepaddle-gpu==1.6.3.post97 -i https://pypi.tuna.tsinghua.edu.cn/simple
Looking in indexes: https://pypi.tuna.tsinghua.edu.cn/simple
Collecting paddlepaddle-gpu==1.6.3.post97
Downloading https://pypi.tuna.tsinghua.edu.cn/packages/8a/67/7e944c6e5f94894217b8a302b075b1ff8c06f54abc55cdb3de0f2c32325a/paddlepaddle_gpu-1.6.3.post97-cp37-cp37m-win_amd64.whl (281.4 MB)
|████████████████████████████████| 281.4 MB 51 kB/s
ERROR: Exception:
Traceback (most recent call last):
File "D:\Program Files\Python\Python37\lib\site-packages\pip_internal\cli\base_command.py", line 186, in _main
status = self.run(options, args)
File "D:\Program Files\Python\Python37\lib\site-packages\pip_internal\commands\install.py", line 331, in run
resolver.resolve(requirement_set)
File "D:\Program Files\Python\Python37\lib\site-packages\pip_internal\legacy_resolve.py", line 177, in resolve
discovered_reqs.extend(self._resolve_one(requirement_set, req))
File "D:\Program Files\Python\Python37\lib\site-packages\pip_internal\legacy_resolve.py", line 333, in _resolve_one
abstract_dist = self._get_abstract_dist_for(req_to_install)
File "D:\Program Files\Python\Python37\lib\site-packages\pip_internal\legacy_resolve.py", line 282, in _get_abstract_dist_for
abstract_dist = self.preparer.prepare_linked_requirement(req)
File "D:\Program Files\Python\Python37\lib\site-packages\pip_internal\operations\prepare.py", line 482, in prepare_linked_requirement
hashes=hashes,
File "D:\Program Files\Python\Python37\lib\site-packages\pip_internal\operations\prepare.py", line 287, in unpack_url
hashes=hashes,
File "D:\Program Files\Python\Python37\lib\site-packages\pip_internal\operations\prepare.py", line 164, in unpack_http_url
unpack_file(from_path, location, content_type)
File "D:\Program Files\Python\Python37\lib\site-packages\pip_internal\utils\unpacking.py", line 252, in unpack_file
flatten=not filename.endswith('.whl')
File "D:\Program Files\Python\Python37\lib\site-packages\pip_internal\utils\unpacking.py", line 139, in unzip_file
shutil.copyfileobj(fp, destfp)
File "D:\Program Files\Python\Python37\lib\shutil.py", line 79, in copyfileobj
buf = fsrc.read(length)
File "D:\Program Files\Python\Python37\lib\zipfile.py", line 930, in read
data = self._read1(n)
File "D:\Program Files\Python\Python37\lib\zipfile.py", line 1006, in _read1
data = self._decompressor.decompress(data, n)
zlib.error: Error -3 while decompressing data: invalid block type

你这个是pip安装过程中就已经报错了吗 你先卸载掉原来的吧

pip uninstall可以成功卸载吗

我昨晚用 pip uninstall 将大部分第三方安装包卸载，试图安装paddle，没有成功；
刚刚已经将整个python卸载，并重新安装了。
错误现象还是一样的。

C:\Users\SidneyZ>python --version
Python 3.7.6

C:\Users\SidneyZ>pip list
Package Version

pip 19.2.3
setuptools 41.2.0
WARNING: You are using pip version 19.2.3, however version 20.0.2 is available.
You should consider upgrading via the 'python -m pip install --upgrade pip' command.

C:\Users\SidneyZ>python -m ensurepip --upgrade
Looking in links: c:\Users\SidneyZ\AppData\Local\Temp\tmp_y95z8c1
Requirement already up-to-date: setuptools in d:\program files\python\python37\lib\site-packages (41.2.0)
Requirement already up-to-date: pip in d:\program files\python\python37\lib\site-packages (19.2.3)

C:\Users\SidneyZ>python -c "import platform;print(platform.architecture()[0]);print(platform.machine())"
64bit
AMD64

C:\Users\SidneyZ>python -m pip install paddlepaddle-gpu==1.6.3.post97 -i https://mirror.baidu.com/pypi/simple
Looking in indexes: https://mirror.baidu.com/pypi/simple
Collecting paddlepaddle-gpu==1.6.3.post97
Using cached https://mirror.baidu.com/pypi/packages/8a/67/7e944c6e5f94894217b8a302b075b1ff8c06f54abc55cdb3de0f2c32325a/paddlepaddle_gpu-1.6.3.post97-cp37-cp37m-win_amd64.whl
ERROR: Exception:
Traceback (most recent call last):
File "D:\Program Files\Python\Python37\lib\site-packages\pip_internal\cli\base_command.py", line 188, in main
status = self.run(options, args)
File "D:\Program Files\Python\Python37\lib\site-packages\pip_internal\commands\install.py", line 345, in run
resolver.resolve(requirement_set)
File "D:\Program Files\Python\Python37\lib\site-packages\pip_internal\legacy_resolve.py", line 196, in resolve
self._resolve_one(requirement_set, req)
File "D:\Program Files\Python\Python37\lib\site-packages\pip_internal\legacy_resolve.py", line 359, in _resolve_one
abstract_dist = self._get_abstract_dist_for(req_to_install)
File "D:\Program Files\Python\Python37\lib\site-packages\pip_internal\legacy_resolve.py", line 307, in _get_abstract_dist_for
self.require_hashes
File "D:\Program Files\Python\Python37\lib\site-packages\pip_internal\operations\prepare.py", line 199, in prepare_linked_requirement
progress_bar=self.progress_bar
File "D:\Program Files\Python\Python37\lib\site-packages\pip_internal\download.py", line 1064, in unpack_url
progress_bar=progress_bar
File "D:\Program Files\Python\Python37\lib\site-packages\pip_internal\download.py", line 928, in unpack_http_url
unpack_file(from_path, location, content_type, link)
File "D:\Program Files\Python\Python37\lib\site-packages\pip_internal\utils\misc.py", line 735, in unpack_file
flatten=not filename.endswith('.whl')
File "D:\Program Files\Python\Python37\lib\site-packages\pip_internal\utils\misc.py", line 631, in unzip_file
shutil.copyfileobj(fp, destfp)
File "D:\Program Files\Python\Python37\lib\shutil.py", line 79, in copyfileobj
buf = fsrc.read(length)
File "D:\Program Files\Python\Python37\lib\zipfile.py", line 930, in read
data = self._read1(n)
File "D:\Program Files\Python\Python37\lib\zipfile.py", line 1006, in _read1
data = self._decompressor.decompress(data, n)
zlib.error: Error -3 while decompressing data: invalid block type
WARNING: You are using pip version 19.2.3, however version 20.0.2 is available.
You should consider upgrading via the 'python -m pip install --upgrade pip' command.

从上面的日志中，可以看到Python已经重新安装过了，pip list里面几乎是空的；我也手动删除过Python的目录。
pip的cache目录在python37的目录下吗？还是在windows的什么地方？
我机器的SET 是设置之前设置过CUDA_VISIABLEDEVICE的，现在也删除了，现象还是这个样子。
还有一个可能，我机器上还有之前安装vs2019时，一起安装的python, 我通过windows store给卸载掉了；不过从where python中，还能看到那个目录，目录下面已经没有东西了。这个不会再有影响了吧。

卸载了VS2019的PYTHON组件，并重启了机器，但where python 依然返回含有MicrosoftWindowsApp 程序的目录。
怎么能够把VS2019安装的Python删除干净呢？

C:\Users\SidneyZ\AppData\Local\Microsoft\WindowsApps\目录 和 系统的TEMP 目录使用管理员权限也不能删除。

C:\Users\SidneyZ\AppData\Local\Microsoft\WindowsApps>where python
C:\Users\SidneyZ\AppData\Local\Microsoft\WindowsApps\python.exe
D:\Program Files\Python\Python37\python.exe

C:\Users\SidneyZ\AppData\Local\Microsoft\WindowsApps>dir
驱动器 C 中的卷是 Win 10 Pro x64
卷的序列号是 3EB7-3F09

C:\Users\SidneyZ\AppData\Local\Microsoft\WindowsApps 的目录

2020/02/24 21:22