tensorflow-gpu版本运行出错：failed to create cublas handle: CUBLAS_STATUS_INTERNAL_ERROR

2020-06-15 20:50:15.674461: E tensorflow/stream_executor/cuda/cuda_blas.cc:238] failed to create cublas handle: CUBLAS_STATUS_INTERNAL_ERROR
2020-06-15 20:50:15.675589: E tensorflow/stream_executor/cuda/cuda_blas.cc:238] failed to create cublas handle: CUBLAS_STATUS_INTERNAL_ERROR
2020-06-15 20:50:15.676041: W tensorflow/stream_executor/stream.cc:2041] attempting to perform BLAS operation using StreamExecutor without BLAS support

原因是对应cublas64_10.dll出错，重新下载替代原文件即可

链接：https://pan.baidu.com/s/13LkUq7D4v1vw2Tbx81K1PA
提取码：b2uw

注：

该错误代码问题与：TensorFlow failed to create cublas handle: CUBLAS_STATUS_ALLOC_FAILED不一样，TensorFlow failed to create cublas handle: CUBLAS_STATUS_ALLOC_FAILED是由于显卡内存不足，可在tensorflow2程序开头加上

import tensorflow as tfconfig = tf.ConfigProto()config.gpu_options.allow_growth = Truesession = tf.Session(config=config, ...)

或者固定分配使用空间：

gpu_options = tf.GPUOptions(per_process_gpu_memory_fraction=0.333)sess = tf.Session(config=tf.ConfigProto(gpu_options=gpu_options))