1. Can’t find libdevice directory ${CUDA_DIR}/nvvm/libdevice

解决方法：

找到你的coda目录。

 copy coda目录下的nvvm目录到你的运行目录

2. Couldn't get ptxas/nvlink version string: INTERNAL: Couldn't invoke ptxas --version

2023-07-31 23:27:33.819439: I tensorflow/tsl/platform/default/subprocess.cc:304] Start cannot spawn child process: No such file or directory
2023-07-31 23:27:33.884567: I tensorflow/tsl/platform/default/subprocess.cc:304] Start cannot spawn child process: No such file or directory
2023-07-31 23:27:33.884648: W tensorflow/compiler/xla/stream_executor/gpu/asm_compiler.cc:109] Couldn't get ptxas version : FAILED_PRECONDITION: Couldn't get ptxas/nvlink version string: INTERNAL: Couldn't invoke ptxas --version
2023-07-31 23:27:33.948057: I tensorflow/tsl/platform/default/subprocess.cc:304] Start cannot spawn child process: No such file or directory
2023-07-31 23:27:33.948203: F tensorflow/compiler/xla/service/gpu/nvptx_compiler.cc:504] ptxas returned an error during compilation of ptx to sass: 'INTERNAL: Failed to launch ptxas'  If the error message indicates that a file could not be written, please verify that sufficient filesystem space is provided.
/tmp/tmptz3pbgu_: line 3: 15024 Aborted                 python /mnt/c/apps/PycharmProjects/ai_service/nlp/translation/LSTM_encoder_decoder.py

解决方法：

conda install -c nvidia cuda-nvcc