I have Jetpack 4.6.1 in which cuda-10.2 is present.
Do I still need to download and install cudnn and cudatoolkit for my environment or it is already present in cuda-10.2(which comes with jetpack) ?
Hi @abdullah13122000, if you have already flashed the JetPack 4.6.1 SD card image, you should already have CUDA Toolkit and cuDNN installed. You can check that you have CUDA Toolkit under /usr/local/cuda and the cuDNN headers here:
ls /usr/include/aarch64-linux-gnu/cudnn*
/usr/include/aarch64-linux-gnu/cudnn_adv_infer_v8.h /usr/include/aarch64-linux-gnu/cudnn_cnn_infer_v8.h /usr/include/aarch64-linux-gnu/cudnn_ops_train_v8.h
/usr/include/aarch64-linux-gnu/cudnn_adv_train_v8.h /usr/include/aarch64-linux-gnu/cudnn_cnn_train_v8.h /usr/include/aarch64-linux-gnu/cudnn_v8.h
/usr/include/aarch64-linux-gnu/cudnn_backend_v8.h /usr/include/aarch64-linux-gnu/cudnn_ops_infer_v8.h /usr/include/aarch64-linux-gnu/cudnn_version_v8.h
I have checked that cuda-10.2 is added to path by executing nvcc --version, Do I still need to add cuDNN and Cudatoolkit to my environment.
I have this question because I am having the following error in which it is mentioned that probably cuDNN failed to initialize:
Frame #: 1
Traceback (most recent call last):
File “/home/tt/archiconda3/envs/nf9/lib/python3.6/site-packages/tensorflow/python/eager/function.py”, line 1717, in _call_impl
cancellation_manager)
File “/home/tt/archiconda3/envs/nf9/lib/python3.6/site-packages/tensorflow/python/eager/function.py”, line 1794, in _call_with_structured_signature
self._structured_signature_check_missing_args(args, kwargs)
File “/home/tt/archiconda3/envs/nf9/lib/python3.6/site-packages/tensorflow/python/eager/function.py”, line 1813, in _structured_signature_check_missing_args
raise TypeError(f"{self._structured_signature_summary()} missing "
TypeError: signature_wrapper(*, input_1) missing required arguments: input_1.During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File “object_tracker.py”, line 304, in
app.run(main)
File “/home/tt/archiconda3/envs/nf9/lib/python3.6/site-packages/absl/app.py”, line 308, in run
_run_main(main, args)
File “/home/tt/archiconda3/envs/nf9/lib/python3.6/site-packages/absl/app.py”, line 254, in _run_main
sys.exit(main(argv))
File “object_tracker.py”, line 128, in main
pred_bbox = infer(batch_data)
File “/home/tt/archiconda3/envs/nf9/lib/python3.6/site-packages/tensorflow/python/eager/function.py”, line 1707, in call
return self._call_impl(args, kwargs)
File “/home/tt/archiconda3/envs/nf9/lib/python3.6/site-packages/tensorflow/python/eager/function.py”, line 1721, in _call_impl
cancellation_manager)
File “/home/tt/archiconda3/envs/nf9/lib/python3.6/site-packages/tensorflow/python/eager/function.py”, line 1774, in _call_with_flat_signature
return self._call_flat(args, self.captured_inputs, cancellation_manager)
File “/home/tt/archiconda3/envs/nf9/lib/python3.6/site-packages/tensorflow/python/saved_model/load.py”, line 123, in _call_flat
cancellation_manager)
File “/home/tt/archiconda3/envs/nf9/lib/python3.6/site-packages/tensorflow/python/eager/function.py”, line 1960, in _call_flat
ctx, args, cancellation_manager=cancellation_manager))
File “/home/tt/archiconda3/envs/nf9/lib/python3.6/site-packages/tensorflow/python/eager/function.py”, line 603, in call
ctx=ctx)
File “/home/tt/archiconda3/envs/nf9/lib/python3.6/site-packages/tensorflow/python/eager/execute.py”, line 59, in quick_execute
inputs, attrs, num_outputs)
tensorflow.python.framework.errors_impl.UnknownError: Failed to get convolution algorithm. This is probably because cuDNN failed to initialize, so try looking to see if a warning log message was printed above.
[[node model/conv2d/Conv2D
(defined at object_tracker.py:73)
]] [Op:__inference_signature_wrapper_3992]Errors may have originated from an input operation.
Input Source operations connected to node model/conv2d/Conv2D:
In[0] model/zero_padding2d/Pad:
In[1] model/conv2d/Conv2D/ReadVariableOp:Operation defined at: (most recent call last)
File “object_tracker.py”, line 304, in
app.run(main)File “/home/tt/archiconda3/envs/nf9/lib/python3.6/site-packages/absl/app.py”, line 308, in run
_run_main(main, args)File “/home/tt/archiconda3/envs/nf9/lib/python3.6/site-packages/absl/app.py”, line 254, in _run_main
sys.exit(main(argv))File “object_tracker.py”, line 73, in main
saved_model_loaded = tf.saved_model.load(FLAGS.weights, tags=[tag_constants.SERVING])Function call stack:
signature_wrapper → _wrapped_model
@abdullah13122000 you shouldn’t need to add additional things to your path. I’m not familiar with this code, but I think the errors you are encountering are primarily related to this initial exception that occurs, as opposed to the other later exception that mentions cuDNN and occurs as a result of this initial exception. So I would investigate that initial exception first and see if that resolves the issue.
Also, when I start TensorFlow and create a session, I get messages like this when it loads the CUDA/cuDNN libraries (this is for TF1):
2023-04-25 16:36:34.771833: I tensorflow/stream_executor/platform/default/dso_loader.cc:50] Successfully opened dynamic library libcudart.so.11.0
WARNING:tensorflow:Deprecation warnings have been disabled. Set TF_ENABLE_DEPRECATION_WARNINGS=1 to re-enable them.
TensorFlow version: 1.15.5
2023-04-25 16:36:36.695896: I tensorflow/stream_executor/platform/default/dso_loader.cc:50] Successfully opened dynamic library libcuda.so.1
2023-04-25 16:36:36.703631: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1049] ARM64 does not support NUMA - returning NUMA node zero
2023-04-25 16:36:36.703830: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1674] Found device 0 with properties:
name: Orin major: 8 minor: 7 memoryClockRate(GHz): 0.624
pciBusID: 0000:00:00.0
2023-04-25 16:36:36.703899: I tensorflow/stream_executor/platform/default/dso_loader.cc:50] Successfully opened dynamic library libcudart.so.11.0
2023-04-25 16:36:36.711617: I tensorflow/stream_executor/platform/default/dso_loader.cc:50] Successfully opened dynamic library libcublas.so.11
2023-04-25 16:36:36.760557: I tensorflow/stream_executor/platform/default/dso_loader.cc:50] Successfully opened dynamic library libcufft.so.10
2023-04-25 16:36:36.762263: I tensorflow/stream_executor/platform/default/dso_loader.cc:50] Successfully opened dynamic library libcurand.so.10
2023-04-25 16:36:36.763676: I tensorflow/stream_executor/platform/default/dso_loader.cc:50] Successfully opened dynamic library libcusolver.so.11
2023-04-25 16:36:36.765762: I tensorflow/stream_executor/platform/default/dso_loader.cc:50] Successfully opened dynamic library libcusparse.so.11
2023-04-25 16:36:36.766692: I tensorflow/stream_executor/platform/default/dso_loader.cc:50] Successfully opened dynamic library libcudnn.so.8
If you continue having problems with your TensorFlow installation, you might want to try the l4t-tensorflow container which comes with it pre-installed.
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.