在没有 AVX 的情况下在 CPU 上运行 TensorFlow 2.0

Question

我想安装和使用 TensorFlow 2.0。我有一台装有 Windows 10 的 PC、一个 Geforce GTX 1080 Ti GPU 和一个不支持 AVX的旧Intel Xeon X5660 CPU。

现在，我的问题是每当我尝试在这台机器上运行任何 TensorFlow 代码时都会出现 DLL 导入错误。我知道这个存储库为传统 CPU 提供解决方案，但不幸的是我在那里找不到任何 TensorFlow 2.0 包。

任何帮助将不胜感激。谢谢你。

Answer 1

存储库中有一个全新的轮文件：

https://github.com/fo40225/tensorflow-windows-wheel

以下文件运行良好：

https://github.com/fo40225/tensorflow-windows-wheel/blob/master/2.0.0/py37/GPU/cuda101cudnn76sse2/tensorflow_gpu-2.0.0-cp37-cp37m-win_amd64.whl

正如 Readme.md 中所述：

“第一次执行TensorFlow需要时间编译。”

看看这个测试：

>>>import tensorflow as tf
tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library cudart64_101.dll

>>>print(tf.__version__)
2.0.0

>>>from tensorflow.python.client import device_lib
>>>print(device_lib.list_local_devices())

tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library nvcuda.dll
tensorflow/core/common_runtime/gpu/gpu_device.cc:1618] Found device 0 with properties: 
name: GeForce GTX 1050 major: 6 minor: 1 memoryClockRate(GHz): 1.531
GPU libraries are statically linked, skip dlopen check.
Adding visible gpu devices: 0
Device interconnect StreamExecutor with strength 1 edge matrix:
     0
0:   N
Created TensorFlow device (/device:GPU:0 with 1340 MB memory) -> physical GPU (device: 0, name: GeForce GTX 1050, pci bus id: 0000:01:00.0, compute capability: 6.1)

[name: "/device:CPU:0"
device_type: "CPU"
memory_limit: 268435456
locality {
}
incarnation: 4456898788177247918
, name: "/device:GPU:0"
device_type: "GPU"
memory_limit: 1406107238
locality {
  bus_id: 1
  links {
  }
}
incarnation: 3224787151756357043
physical_device_desc: "device: 0, name: GeForce GTX 1050, pci bus id: 0000:01:00.0, compute capability: 6.1"
]