Open Wasim04 opened 5 months ago
@Wasim04, I am trying to execute the mentioned code with the GPU and CPU environments and the time taken to execute the code is more than expected. Could you please allow some duration to analyse the issue and provide the update on the same. Thank you!
@tilakrayal Thanks for your response. Please take your time.
@tilakrayal Hi, just wondering if you had a chance to look into this issue?
Same issue here but not with tensorflow - the error seem to come from jax==0.4.30
Issue type
Bug
Have you reproduced the bug with TensorFlow Nightly?
Yes
Source
source
TensorFlow version
2.15.0.post1
Custom code
Yes
OS platform and distribution
Rocky Linux 8.9
Mobile device
No response
Python version
3.10.12
Bazel version
No response
GCC/compiler version
No response
CUDA/cuDNN version
8.9.2.26
GPU model and memory
NVIDIA A100-SXM4-80GB
Current behavior?
I am trying to run a simple denoising autoencoder. my training data and label data are 900 samples of healpy maps with nside 64 resolution, loaded as numpy array. After normalising the maps, I used tf.data.Dataset.from_tensor_slices, to create dataset. when I used random noise to create these maps and ran on jupyter notebook, although took ages to initiate training after doing model.fit(), but it did run and produced some result. knowing that the model works, I tried to run on GPU with real data. this is where the issue started. it shows the following error: Check failed: ret == 0 (11 vs. 0)Thread tf_data_private_threadpool creation via pthread_create() failed. Aborted (core dumped), and the process stops.
Standalone code to reproduce the issue
Relevant log output