GPU acceleration

[mrba59]

Hello i still don't know if the models available use GPU acceleration or full cpu. My problem is that when i am running inference on a remote server, the cpu load explode and make all process killed. Even for my collegues ...

I tried this code

t1 = ants.image_read("1_OFSEP_defaced.nii.gz")
flair = ants.image_read("1_flair_to_t1_registered.nii.gz")
wmh = antspynet.sysu_media_wmh_segmentation(flair, t1, verbose=True)

and i get this message from tensorflow

2025-02-04 15:17:57.006333: W tensorflow/core/common_runtime/gpu/gpu_device.cc:1960] Cannot dlopen some GPU libraries. Please make sure the missing libraries mentioned above are installed 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...

I have installed tf == 2.13.1 with cuda 11.8. Is it normal that the model is not using GPU ?

Thanks

[ntustison]

Do you explicitly set the number of threads?

import tensorflow as tf

# Set the number of threads for intra-op parallelism (within an operation)
tf.config.threading.set_intra_op_parallelism_threads(4)

# Set the number of threads for inter-op parallelism (between operations)
tf.config.threading.set_inter_op_parallelism_threads(2)

[akhil4674]

Try Rebuilding your environment ,

Install nvidia drivers (not cuda) >> Build python env >> Install tensorflow[and-cuda]