Tensorflow repeated success messages and NUMA node read warning

Question

I've just installed cuda 11.2 via the runfile, and tensorflow via pip install tensorflow on Ubuntu 20.04 with Python 3.8. I get a bizzare readout when creating a tensor and memory usage on my RTX 3090 goes to 95%

Python 3.8.5 (default, May 27 2021, 13:30:53) 
[GCC 9.3.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import tensorflow as tf
2021-06-25 10:42:08.881025: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcudart.so.11.0
>>> a = tf.zeros(1)
2021-06-25 10:42:16.739723: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcuda.so.1
2021-06-25 10:42:16.775681: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:937] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2021-06-25 10:42:16.776749: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1733] Found device 0 with properties: 
pciBusID: 0000:09:00.0 name: GeForce RTX 3090 computeCapability: 8.6
coreClock: 1.695GHz coreCount: 82 deviceMemorySize: 23.69GiB deviceMemoryBandwidth: 871.81GiB/s
2021-06-25 10:42:16.776781: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcudart.so.11.0
2021-06-25 10:42:16.779208: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcublas.so.11
2021-06-25 10:42:16.779252: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcublasLt.so.11
2021-06-25 10:42:16.780078: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcufft.so.10
2021-06-25 10:42:16.780261: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcurand.so.10
2021-06-25 10:42:16.780973: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcusolver.so.11
2021-06-25 10:42:16.781346: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcusparse.so.11
2021-06-25 10:42:16.781423: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcudnn.so.8
2021-06-25 10:42:16.781476: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:937] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2021-06-25 10:42:16.782026: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:937] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2021-06-25 10:42:16.782642: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1871] Adding visible gpu devices: 0
2021-06-25 10:42:16.782902: I tensorflow/core/platform/cpu_feature_guard.cc:142] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations:  AVX2 FMA
To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.
2021-06-25 10:42:16.783252: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:937] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2021-06-25 10:42:16.783876: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1733] Found device 0 with properties: 
pciBusID: 0000:09:00.0 name: GeForce RTX 3090 computeCapability: 8.6
coreClock: 1.695GHz coreCount: 82 deviceMemorySize: 23.69GiB deviceMemoryBandwidth: 871.81GiB/s
2021-06-25 10:42:16.783920: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:937] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2021-06-25 10:42:16.784470: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:937] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2021-06-25 10:42:16.785250: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1871] Adding visible gpu devices: 0
2021-06-25 10:42:16.785276: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcudart.so.11.0
2021-06-25 10:42:17.059795: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1258] Device interconnect StreamExecutor with strength 1 edge matrix:
2021-06-25 10:42:17.059823: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1264]      0 
2021-06-25 10:42:17.059828: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1277] 0:   N 
2021-06-25 10:42:17.059954: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:937] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2021-06-25 10:42:17.060459: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:937] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2021-06-25 10:42:17.061025: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:937] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2021-06-25 10:42:17.061578: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1418] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 21542 MB memory) -> physical GPU (device: 0, name: GeForce RTX 3090, pci bus id: 0000:09:00.0, compute capability: 8.6)

Any ideas?

Answer 1

I was able to replicate your issue

(base) XXXXXX@XXXXX-Xlaptop:~$ python
Python 3.7.7 (default, Mar 26 2020, 15:48:22) 
[GCC 7.3.0] :: Anaconda, Inc. on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import tensorflow
2021-07-08 16:22:12.609456: I tensorflow/stream_executor/cuda/cudart_stub.cc:29] Ignore above cudart dlerror if you do not have a GPU set up on your machine.
>>> import tensorflow as tf
>>> a = tf.zeros(1)
2021-07-08 16:23:07.820538: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcuda.so.1
2021-07-08 16:23:07.896686: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:156] kernel driver does not appear to be running on this host (mothukuru-glaptop): /proc/driver/nvidia/version does not exist
2021-07-08 16:23:07.897348: I tensorflow/core/platform/cpu_feature_guard.cc:142] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations:  AVX2 FMA
To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.

They are just the information message as they are prefixed with I, if it is the error message they would be prefixed with E or W for warnings.

To suppress the logging you can try as shown below

>>> import tensorflow.compat.v1 as tf
>>> tf.logging.set_verbosity(tf.logging.ERROR)
>>> a = tf.zeros(1)