错误1

RuntimeError: module compiled against API version 0xf but this version of numpy is 0xd

pip install numpy -U

Fix：把numpy安装到最新版》1.24.1

错误2

Collecting tensorflow-gpu
  Using cached https://mirrors.aliyun.com/pypi/packages/8a/45/fa31ced1db38f9424f262dfbf35747fe5378b5c808cecb373c8cb8e515d3/tensorflow-gpu-2.12.0.tar.gz (2.6 kB)
  Preparing metadata (setup.py) ... error
  error: subprocess-exited-with-error

  × python setup.py egg_info did not run successfully.
  │ exit code: 1
  ╰─> [39 lines of output]
      Traceback (most recent call last):
        File "C:\Users\73915\.conda\envs\py38\lib\site-packages\setuptools\_vendor\packaging\requirements.py", line 35, in __init__
          parsed = parse_requirement(requirement_string)
        File "C:\Users\73915\.conda\envs\py38\lib\site-packages\setuptools\_vendor\packaging\_parser.py", line 64, in parse_requirement
          return _parse_requirement(Tokenizer(source, rules=DEFAULT_RULES))
        File "C:\Users\73915\.conda\envs\py38\lib\site-packages\setuptools\_vendor\packaging\_parser.py", line 82, in _parse_requirement
          url, specifier, marker = _parse_requirement_details(tokenizer)
        File "C:\Users\73915\.conda\envs\py38\lib\site-packages\setuptools\_vendor\packaging\_parser.py", line 126, in _parse_requirement_details
          marker = _parse_requirement_marker(
        File "C:\Users\73915\.conda\envs\py38\lib\site-packages\setuptools\_vendor\packaging\_parser.py", line 147, in _parse_requirement_marker
          tokenizer.raise_syntax_error(
        File "C:\Users\73915\.conda\envs\py38\lib\site-packages\setuptools\_vendor\packaging\_tokenizer.py", line 163, in raise_syntax_error
          raise ParserSyntaxError(
      setuptools.extern.packaging._tokenizer.ParserSyntaxError: Expected end or semicolon (after name and no valid version specifier)
          python_version>"3.7"
                        ^

      The above exception was the direct cause of the following exception:

Fix：pip install tf-nightly-gpu， 安装nightly版。我有一台其它机器，好些只能安装pip install tensorflow==2.9.3 。tensorflow 现在兼容性，安装文档做得实在太差了。

错误3

Could not load dynamic library 'cusolver64_10.dll'; dlerror: cusolver64_10.dll not found

我安装的是cuda 11.7，所以找不到10，

Fix：C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.7\bin 中，把cusolver64_11.dll改名成cusolver64_10.dll

基本上把以上错误搞好，就可以了。

$ python ~/.script/gpu/tf_check.py
2023-05-23 11:23:00.436724: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library cudart64_110.dll
tf version= 2.5.0-dev20210204
2023-05-23 11:23:03.316810: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library nvcuda.dll
2023-05-23 11:23:03.440978: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1770] Found device 0 with properties:
pciBusID: 0000:73:00.0 name: NVIDIA GeForce RTX 3090 computeCapability: 8.6
coreClock: 1.695GHz coreCount: 82 deviceMemorySize: 24.00GiB deviceMemoryBandwidth: 871.81GiB/s
2023-05-23 11:23:03.441306: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1770] Found device 1 with properties:
pciBusID: 0000:d5:00.0 name: NVIDIA GeForce RTX 3090 computeCapability: 8.6
coreClock: 1.695GHz coreCount: 82 deviceMemorySize: 24.00GiB deviceMemoryBandwidth: 871.81GiB/s
2023-05-23 11:23:03.441444: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library cudart64_110.dll
2023-05-23 11:23:03.459304: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library cublas64_11.dll
2023-05-23 11:23:03.459516: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library cublasLt64_11.dll
2023-05-23 11:23:03.470889: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library cufft64_10.dll
2023-05-23 11:23:03.474038: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library curand64_10.dll
2023-05-23 11:23:03.985571: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library cusolver64_10.dll
2023-05-23 11:23:03.992641: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library cusparse64_11.dll
2023-05-23 11:23:03.994094: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library cudnn64_8.dll
2023-05-23 11:23:03.994290: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1908] Adding visible gpu devices: 0, 1