解决 Could not load dynamic library cudart64_101.dll 或 tensorflow 2.3.0 CUDA 10.0 问题

环境：

• python 3.6.8
• tensorflow-gpu 2.3.0rc0
• CUDA 10.0
• cudnn 7.4

PS：CUDA10.0 安装方式:地址

 

错误：

Could not load dynamic library ‘cudart64_101.dll‘
Could not load dynamic library ‘cublas64_10.dll‘
Could not load dynamic library ‘cufft64_10.dll‘
Could not load dynamic library ‘curand64_10.dll‘
Could not load dynamic library ‘cusolver64_10.dll‘
Could not load dynamic library ‘cusparse64_10.dll‘

 

版本：

 

DLL：

 

解决：

• 复制 上述的 文件（缺什么复制什么）
• 拷贝到 C盘 Windows/System32
• 修改 其 名称

 

测试：

import tensorflow as tf
if tf.test.gpu_device_name():
    print('Default GPU Device: {}'.format(tf.test.gpu_device_name()))
else:
    print("Please install GPU version of TF")