记一次windows上安装gpu遇到的问题--动态链接库

目的

• 本机电脑基础配置MX150显卡
• 安装gpu目的是，希望用gpu来加速模型训练

安装步骤

暴力安装tensorflow_gpu

[~]# conda install tensorflow_gpu
这个过程相对时间比较久

先查看本机显卡驱动信息

[~]# nvidia-smi
# 以下信息可知有一张显卡，cuda版本为11，驱动版本450
Sun Jan 16 01:36:35 2022
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 451.67       Driver Version: 451.67       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 MX150      WDDM  | 00000000:01:00.0 Off |                  N/A |
| N/A   48C    P8    N/A /  N/A |     64MiB /  2048MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+

+-----------------------------------------------------------------------------+
| Processes:                                                                  |
|  GPU   GI   CI        PID   Type   Process name                  GPU Memory |
|        ID   ID                                                   Usage      |
|=============================================================================|
|  No running processes found                                                 |
+-----------------------------------------------------------------------------+

查看gpu是否真的可用

[~]# python
>>> import tensorflow as tf
>>> tf.test.is_gpu_available()
2022-01-16 01:42:25.448896: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1720] Found device 0 with properties:
pciBusID: 0000:01:00.0 name: GeForce MX150 computeCapability: 6.1
coreClock: 1.5315GHz coreCount: 3 deviceMemorySize: 2.00GiB deviceMemoryBandwidth: 44.76GiB/s
2022-01-16 01:42:25.459216: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library cudart64_110.dll
2022-01-16 01:42:25.464589: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library cublas64_11.dll
2022-01-16 01:42:25.469195: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library cublasLt64_11.dll
2022-01-16 01:42:25.473979: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library cufft64_10.dll
2022-01-16 01:42:25.497739: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library curand64_10.dll
2022-01-16 01:42:25.529822: W tensorflow/stream_executor/platform/default/dso_loader.cc:60] Could not load dynamic library 'cusolver64_10.dll'; dlerror: cusolver64_10.dll not found
2022-01-16 01:42:25.536670: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library cusparse64_11.dll
2022-01-16 01:42:25.541340: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library cudnn64_8.dll
2022-01-16 01:42:25.546266: W tensorflow/core/common_runtime/gpu/gpu_device.cc:1757] Cannot dlopen some GPU libraries. Please make sure the missing libraries mentioned above are instal
led properly if you would like to use GPU. Follow the guide at https://www.tensorflow.org/install/gpu for how to download and setup the required libraries for your platform.
Skipping registering GPU devices...
2022-01-16 01:42:25.561557: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1261] Device interconnect StreamExecutor with strength 1 edge matrix:
2022-01-16 01:42:25.566103: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1267]      0
2022-01-16 01:42:25.569507: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1280] 0:   N
2022-01-16 01:42:25.573164: I tensorflow/compiler/jit/xla_gpu_device.cc:99] Not creating XLA devices, tf_xla_enable_xla_devices not set
False

• 关键信息是cusolver64_10.dll 无法找到
• 可能是一个bug，因为在安装目录存在cusolver64_11.dll的文件，而动态链接确是10，鉴于这个原因，将安装目录bin下cusolver64_11.dll改为cusolver64_10.dll

再次查看gpu是否真的可用

[~]# python
>>> import tensorflow as tf
>>> tf.test.is_gpu_available()
>2022-01-16 02:02:15.288957: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1720] Found device 0 with properties:
pciBusID: 0000:01:00.0 name: GeForce MX150 computeCapability: 6.1
coreClock: 1.5315GHz coreCount: 3 deviceMemorySize: 2.00GiB deviceMemoryBandwidth: 44.76GiB/s
2022-01-16 02:02:15.300269: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library cudart64_110.dll
2022-01-16 02:02:15.304805: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library cublas64_11.dll
2022-01-16 02:02:15.310304: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library cublasLt64_11.dll
2022-01-16 02:02:15.318343: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library cufft64_10.dll
2022-01-16 02:02:15.323535: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library curand64_10.dll
2022-01-16 02:02:15.328497: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library cusolver64_10.dll
2022-01-16 02:02:15.333582: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library cusparse64_11.dll
2022-01-16 02:02:15.340556: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library cudnn64_8.dll
2022-01-16 02:02:15.346280: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1862] Adding visible gpu devices: 0
2022-01-16 02:02:15.349912: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1261] Device interconnect StreamExecutor with strength 1 edge matrix:
2022-01-16 02:02:15.355085: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1267]      0
2022-01-16 02:02:15.358194: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1280] 0:   N
2022-01-16 02:02:15.463779: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1406] Created TensorFlow device (/device:GPU:0 with 1342 MB memory) -> physical GPU (device: 0, name: GeF
orce MX150, pci bus id: 0000:01:00.0, compute capability: 6.1)
2022-01-16 02:02:15.473924: I tensorflow/compiler/jit/xla_gpu_device.cc:99] Not creating XLA devices, tf_xla_enable_xla_devices not set
True

>>> from tensorflow.python.client import device_lib
>>> print(device_lib.list_local_devices())
2022-01-16 02:01:12.457057: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1720] Found device 0 with properties:
pciBusID: 0000:01:00.0 name: GeForce MX150 computeCapability: 6.1
coreClock: 1.5315GHz coreCount: 3 deviceMemorySize: 2.00GiB deviceMemoryBandwidth: 44.76GiB/s
2022-01-16 02:01:12.468355: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library cudart64_110.dll
2022-01-16 02:01:12.472850: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library cublas64_11.dll
2022-01-16 02:01:12.478089: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library cublasLt64_11.dll
2022-01-16 02:01:12.484316: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library cufft64_10.dll
2022-01-16 02:01:12.490559: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library curand64_10.dll
2022-01-16 02:01:12.812244: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library cusolver64_10.dll
2022-01-16 02:01:12.818152: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library cusparse64_11.dll
2022-01-16 02:01:12.823162: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library cudnn64_8.dll
2022-01-16 02:01:12.868417: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1862] Adding visible gpu devices: 0
2022-01-16 02:01:19.558513: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1261] Device interconnect StreamExecutor with strength 1 edge matrix:
2022-01-16 02:01:19.565232: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1267]      0
2022-01-16 02:01:19.569404: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1280] 0:   N
2022-01-16 02:01:19.665554: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1406] Created TensorFlow device (/device:GPU:0 with 1342 MB memory) -> physical GPU (device: 0, name: GeF
orce MX150, pci bus id: 0000:01:00.0, compute capability: 6.1)
2022-01-16 02:01:19.776437: I tensorflow/compiler/jit/xla_gpu_device.cc:99] Not creating XLA devices, tf_xla_enable_xla_devices not set
[name: "/device:CPU:0"
device_type: "CPU"
memory_limit: 268435456
locality {
}
incarnation: 17822183251633891573
, name: "/device:GPU:0"
device_type: "GPU"
memory_limit: 1408043827
locality {
  bus_id: 1
  links {
  }
}
incarnation: 8018427731587780304
physical_device_desc: "device: 0, name: GeForce MX150, pci bus id: 0000:01:00.0, compute capability: 6.1"
]

• gpu已经可用