Tensorflow 2.0 CUDA 顯示有問題? - Cupoy

Tensorflow 2.0 CUDA 顯示有問題?

更新 tensorflow 到 2.0 會跳出一堆錯誤訊息。 刪掉 tensorflow 再重新用 `conda install tensorflow-gpu=2.0.0` 安裝也是會跳出一樣的錯誤訊息。 有試過刪掉 CUDA 再重新安裝 CUDA 10.0 和 CuDNN 7.6.0，但是 nvidia-smi 上面會顯示 CUDA 11.0 ``` Sun Sep 20 10:49:51 2020 +-----------------------------------------------------------------------------+ | NVIDIA-SMI 451.82 Driver Version: 451.82 CUDA Version: 11.0 | |-------------------------------+----------------------+----------------------+ | GPU Name TCC/WDDM | Bus-Id Disp.A | Volatile Uncorr. ECC | | Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. | |===============================+======================+======================| | 0 GeForce RTX 208... WDDM | 00000000:01:00.0 On | N/A | | 0% 59C P3 79W / 300W | 1468MiB / 11264MiB | 0% Default | +-------------------------------+----------------------+----------------------+ ``` 環境變數也有設，但重新跑 code 也是會跳出一樣的訊息。 以下是 log 顯示的訊息： ``` 2020-09-20 10:51:32.182818: E tensorflow/stream_executor/cuda/cuda_dnn.cc:319] Loaded runtime CuDNN library: 7.4.1 but source was compiled with: 7.6.0. CuDNN library major and minor version nee ds to match or have higher minor version in case of CuDNN 7.0 or later version. If using a binary install, upgrade your CuDNN library. If building from sources, make sure the library loaded at runtime is compatible with the version specified during compile configuration. 2020-09-20 10:51:32.196765: W tensorflow/core/framework/op_kernel.cc:1622] OP_REQUIRES failed at cudnn_rnn_ops.cc:1491 : Unknown: Fail to find the dnn implementation. 2020-09-20 10:51:32.201205: W tensorflow/core/common_runtime/base_collective_executor.cc:216] BaseCollectiveExecutor::StartAbort Unknown: Fail to find the dnn implementation. [[{{node CudnnRNN}}]] 2020-09-20 10:51:32.206991: W tensorflow/core/common_runtime/base_collective_executor.cc:216] BaseCollectiveExecutor::StartAbort Cancelled: [_Derived_]RecvAsync is cancelled. [[{{node Reshape_16/_62}}]] [[Adam/Adam/update/AssignSubVariableOp/_65]] 2020-09-20 10:51:32.213809: W tensorflow/core/common_runtime/base_collective_executor.cc:216] BaseCollectiveExecutor::StartAbort Cancelled: [_Derived_]RecvAsync is cancelled. [[{{node Reshape_16/_62}}]] ```