适配NVIDIA GeForce GTX 480的TensorFlow+Keras版本选择咨询

Microsoft Windows [Version 10.0.19045.2965]
(c) Microsoft Corporation. All rights reserved.


C:\Users\pc>nvcc -V
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2017 NVIDIA Corporation
Built on Fri_Sep__1_21:08:32_Central_Daylight_Time_2017
Cuda compilation tools, release 9.0, V9.0.176


C:\Users\pc>nvidia-smi
Fri Nov 10 14:02:45 2023
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 391.35                 Driver Version: 391.35                    |
|-------------------------------+----------------------+----------------------+
| 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 GTX 480    WDDM  | 00000000:03:00.0 N/A |                  N/A |
| 44%   63C   P12    N/A /  N/A |    360MiB /  1536MiB |     N/A      Default |
+-------------------------------+----------------------+----------------------+

+-----------------------------------------------------------------------------+
| Processes:                                                       GPU Memory |
|  GPU       PID   Type   Process name                             Usage      |
|=============================================================================|
|    0                    Not Supported                                       |
+-----------------------------------------------------------------------------+

C:\Users\pc>

C:\Users\pc>pip install tensorflow==2.0.0
ERROR: Could not find a version that satisfies the requirement tensorflow==2.0.0 
(from versions: 2.12.0rc0, 2.12.0rc1, 2.12.0, 2.12.1, 2.13.0rc0, 2.13.0rc1, 2.13.0rc2, 
2.13.0, 2.13.1, 2.14.0rc0, 2.14.0rc1, 2.14.0, 2.15.0rc0, 2.15.0rc1)
ERROR: No matching distribution found for tensorflow==2.0.0

• 核心限制：你的GeForce GTX 480属于NVIDIA Fermi架构，TensorFlow从2.10版本起不再支持该架构；同时当前Python3.11不兼容支持CUDA9.0的早期TensorFlow版本，需要先调整Python版本。
• 版本组合：
  1. 降级Python到3.7版本（这是同时兼容TensorFlow 2.9.x和CUDA9.0的最高Python版本）
  2. 安装TensorFlow 2.9.3（最后一个支持Fermi架构的TensorFlow版本，完美匹配CUDA9.0）
  3. Keras无需单独安装，TensorFlow 2.x版本已内置对应版本的Keras，会随TensorFlow自动部署
• 安装命令（调整Python版本后执行）：

  pip install tensorflow==2.9.3
  

• 补充：如果不愿降级Python，可选择安装CPU版本TensorFlow，但GPU无法参与训练，速度会大幅下降。

内容的提问来源于stack exchange，提问作者user366312