Question

我正在运行NVIDIA GeFOrce MX150。我已经安装了CUDA工具包v10，cuDNN和tensorflow-gpu。

nvcc --version
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2019 NVIDIA Corporation
Built on Wed_Oct_23_19:32:27_Pacific_Daylight_Time_2019
Cuda compilation tools, release 10.2, V10.2.89

>>> import tensorflow as tf
2019-11-23 20:57:00.621530: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library cudart64_100.dll
>>> tf.test.is_built_with_cuda()
True
>>> tf.test.is_gpu_available(cuda_only=False, min_cuda_compute_capability=None)
2019-11-23 20:57:35.881189: I tensorflow/core/platform/cpu_feature_guard.cc:142] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX AVX2
2019-11-23 20:57:35.886829: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library nvcuda.dll
2019-11-23 20:57:35.904122: E tensorflow/stream_executor/cuda/cuda_driver.cc:318] failed call to cuInit: CUDA_ERROR_UNKNOWN: unknown error
2019-11-23 20:57:35.913593: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:169] retrieving CUDA diagnostic information for host: [REDACTED]
2019-11-23 20:57:35.916327: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:176] hostname: [REDACTED]
False

有人可以帮助您解决问题吗？

Answer 1

您似乎已安装CUDA 10.2。也许值得尝试推荐的版本10.0？检查tensorflow gpu software requirements了解更多信息。 windows setup section.中也提到了这是指向CUDA 10.0 download

的链接

tf.test.is_gpu_available（cuda_only = False，min_cuda_compute_capability = None）返回False

1 个答案: