cudart64_100.dll；cublas64_100.dll等没有找到的解决办法

将路径下的

C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.0\bin

cudart64_10.dll；cublas64_10.dll等重命名为_100.dll

 

2022-04-29 09:39:51.131894: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library cudart64_100.dll
2022-04-29 09:39:51.134541: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library cublas64_100.dll
2022-04-29 09:39:51.136832: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library cufft64_100.dll
2022-04-29 09:39:51.138116: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library curand64_100.dll
2022-04-29 09:39:51.141285: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library cusolver64_100.dll
2022-04-29 09:39:51.143360: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library cusparse64_100.dll

执行测试代码：

from tensorflow.python.client import device_lib
import tensorflow as tf
print(device_lib.list_local_devices())
print(tf.test.is_gpu_available())

结果为True