装的包：
tensorflow2.8
python 3.9
cuda11.4 同版cuDNN
配置：
显卡：英伟达 MX450
报错：
2022-05-23 15:54:28.129536: I tensorflow/stream_executor/cuda/cuda_dnn.cc:368] Loaded cuDNN version 8204
2022-05-23 15:54:28.509751: E tensorflow/stream_executor/cuda/cuda_blas.cc:232] failed to create cublas handle: CUBLAS_STATUS_NOT_INITIALIZED
2022-05-23 15:54:28.509835: E tensorflow/stream_executor/cuda/cuda_blas.cc:234] Failure to initialize cublas may be due to OOM (cublas needs some free memory when you initialize it, and your deep-learning framework may have preallocated more than its fair share), or may be because this binary was not built with support for the GPU in your machine.
2022-05-23 15:54:28.510240: W tensorflow/core/framework/op_kernel.cc:1745] OP_REQUIRES failed at conv_ops.cc:1120 : NOT_FOUND: No algorithm worked! Error messages:
Profiling failure on CUDNN engine 1#TC: UNKNOWN: CUDNN_STATUS_EXECUTION_FAILED
in tensorflow/stream_executor/cuda/cuda_dnn.cc(4141): 'cudnnConvolutionForward( cudnn.handle(), alpha, input_nd_.handle(), input_data.opaque(), filter_.handle(), filter_data.opaque(), conv_.handle(), ToConvForwardAlgo(algo), scratch_memory.opaque(), scratch_memory.size(), beta, output_nd_.handle(), output_data.opaque())'