[Python] Can't detect GPU with cuDNN 8.6 and CUDA 11.8

I am trying to use TensorFlow with my NVIDIA GeForce RTX 4090 GPU on Windows 11. I installed CUDA 11.8 and cuDNN 8.6 but TensorFlow fails to detect the GPU and computations are performed on the CPU.

• Latest NVIDIA drivers were installed. Verified using nvidia-smi.
• Downloaded and installed CUDA Toolkit 11.8 from the NVIDIA website.
• Downloaded and installed cuDNN 8.6 ensuring the files were copied to the CUDA directory.
• Initially installed tensorflow-gpu but faced issues, so I switched to TensorFlow since GPU support is included by default in TensorFlow 2.x.
• Verified that CUDA and cuDNN were correctly installed and functional using the CUDA samples provided by NVIDIA.
• Ensured the environment variables for CUDA and cuDNN were set correctly in the system PATH.

Python code for verification:

import tensorflow as tf
print("TensorFlow version:", tf.__version__)
print("Num GPUs Available: ", len(tf.config.experimental.list_physical_devices('GPU')))

# Simple matrix multiplication to test computation
a = tf.constant([[1.0, 2.0], [3.0, 4.0]])
b = tf.constant([[1.0, 1.0], [1.0, 1.0]])
c = tf.matmul(a, b)
print("Result of matrix multiplication:\n", c)


Output:

No GPU found.
2024-07-31 14:23:01.757763: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.
To enable the following instructions: SSE SSE2 SSE3 SSE4.1 SSE4.2 AVX AVX2 AVX512F AVX512_VNNI AVX512_BF16 FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.
Result of matrix multiplication:
tf.Tensor(
[[1. 3.]
[3. 7.]], shape=(2, 2), dtype=float32)


TensorFlow still fails to detect the GPU.

Continue reading...