failed to run cuBLAS routine: CUBLAS_STATUS_NOT_SUPPORTED

[fratamico]

When running the style.py, I get the above error message. Full stack trace below.

I'm running the following:
python style.py --style mypic.jpg --checkpoint-dir my_checkpoint_dir --content-weight 1.5e1 --checkpoint-iterations 1000 --batch-size 10

Versions:

• CUDA Version: 11.2
• Python 3.7.9
• Tensorflow 2.1.0
• GeForce RTX 3090

I've tried allowing GPU memory growth setting the GPU memory fraction as specified here, but still get he same error. Has anyone else solved this?

If using Keras pass *_constraint arguments to layers.
UID: 50
2021-01-06 18:29:08.600763: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2021-01-06 18:43:17.614048: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10
2021-01-06 18:44:26.855463: E tensorflow/stream_executor/cuda/cuda_blas.cc:428] failed to run cuBLAS routine: CUBLAS_STATUS_NOT_SUPPORTED
2021-01-06 18:44:26.855496: E tensorflow/stream_executor/cuda/cuda_blas.cc:2301] Internal: failed BLAS call, see log for details
Traceback (most recent call last):
  File "/home/lauren/anaconda3/envs/tf-gpu/lib/python3.7/site-packages/tensorflow_core/python/client/session.py", line 1367, in _do_call
    return fn(*args)
  File "/home/lauren/anaconda3/envs/tf-gpu/lib/python3.7/site-packages/tensorflow_core/python/client/session.py", line 1352, in _run_fn
    target_list, run_metadata)
  File "/home/lauren/anaconda3/envs/tf-gpu/lib/python3.7/site-packages/tensorflow_core/python/client/session.py", line 1445, in _call_tf_sessionrun
    run_metadata)
tensorflow.python.framework.errors_impl.InternalError: Blas xGEMMBatched launch failed : a.shape=[10,65536,64], b.shape=[10,65536,64], m=64, n=64, k=65536, batch_size=10
	 [[{{node ArithmeticOptimizer/FoldTransposeIntoMatMul_MatMul}}]]

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "style.py", line 167, in <module>
    main()
  File "style.py", line 147, in main
    for preds, losses, i, epoch in optimize(*args, **kwargs):
  File "src/optimize.py", line 118, in optimize
    train_step.run(feed_dict=feed_dict)
  File "/home/lauren/anaconda3/envs/tf-gpu/lib/python3.7/site-packages/tensorflow_core/python/framework/ops.py", line 2391, in run
    _run_using_default_session(self, feed_dict, self.graph, session)
  File "/home/lauren/anaconda3/envs/tf-gpu/lib/python3.7/site-packages/tensorflow_core/python/framework/ops.py", line 5347, in _run_using_default_session
    session.run(operation, feed_dict)
  File "/home/lauren/anaconda3/envs/tf-gpu/lib/python3.7/site-packages/tensorflow_core/python/client/session.py", line 960, in run
    run_metadata_ptr)
  File "/home/lauren/anaconda3/envs/tf-gpu/lib/python3.7/site-packages/tensorflow_core/python/client/session.py", line 1183, in _run
    feed_dict_tensor, options, run_metadata)
  File "/home/lauren/anaconda3/envs/tf-gpu/lib/python3.7/site-packages/tensorflow_core/python/client/session.py", line 1361, in _do_run
    run_metadata)
  File "/home/lauren/anaconda3/envs/tf-gpu/lib/python3.7/site-packages/tensorflow_core/python/client/session.py", line 1386, in _do_call
    raise type(e)(node_def, op, message)
tensorflow.python.framework.errors_impl.InternalError: Blas xGEMMBatched launch failed : a.shape=[10,65536,64], b.shape=[10,65536,64], m=64, n=64, k=65536, batch_size=10
	 [[{{node ArithmeticOptimizer/FoldTransposeIntoMatMul_MatMul}}]]```

[syu-tan]

i solved its issue on same envs. maybe problem is 3090 .
please try pip uninstall tensorflow and pip install tensorflow==2.4.1
it doesn't recognize the name of the GPU, but it works fine.