Difference: PreparingThePC (3 vs. 4)

Revision 42023-09-21 - UliRaich

Line: 1 to 1

META TOPICPARENT	name="TinyML"

Preparing the PC for Tensorflow

Setting up a virtual Python environment

Line: 35 to 35

Changed:

<
<

TensorFlow relies on the cuda toolkit

and on the NVidia display driver. I had difficulties to install these with the .deb files but succeeded to do so using the .run file.

>
>

When looking at the Software & Updates program we see that the nvidia-driver-535 is used to control the NVidia GeForce GTX 950 module. Make sure that the nouveau driver is disabled!

Changed:

<
<

>
>

Changed:

<
<

To run the installer script I had to boot into single user mode, avoiding to start the desktop system. This can be done typing "e" at the boot screen and adding "single" to the boot command.

>
>

TensorFlow relies on the cuda toolkit

and on the NVidia display driver. Even after installing the newest version of cuda, TensorFlow did not see the GPU. An intensive search with Google finally showed that (at the moment of writing this page) TensorFlow version 2 only works with cuda 11.8 and cuDNN 8.6.

Changed:

<
<

In the "Additional Drivers" tab of the Software & Updates program you see that we are using the manually installed NVidia driver:

>
>

https://medium.com/@juancrrn/installing-cuda-and-cudnn-in-ubuntu-20-04-for-deep-learning-dad8841714d6

describes how to install cuda-11.8 and cudnn-8.6 and how to test the installation.

Changed:

<
<

>
>

The script testGPU.py below shows the number of GPUs that TensorFlow has fouind: #!/home/uli/.virtualenvs/AI/bin/python
import tensorflow as tf
print("Num GPUs Available: ", len(tf.config.list_physical_devices('GPU')))

Changed:

<
<

When trying a very simple TensorFlow program: #!/usr/bin/python3
import tensorflow as tf
import numpy as np
from tensorflow import keras
model = tf.keras.Sequential([keras.layers.Dense(units=1, input_shape=[1])])

>
>

After installation of cuda-11.8 and cuDNN-8.6 this is what I get, when I run the script:

Changed:

<
<

I get:

>
>

Changed:

<
<

>
>

To get rid of these warnings I first had to install tensorrt from the NVidia WEB site

To irradicate the NUMA mode warning I followed:

https://askubuntu.com/questions/1379119/how-to-set-the-numa-node-for-an-nvidia-gpu-persistently

When running the testGPU program now, things seem to be ok and I should be able to use my NVidia GPU.

Uli Raich - 2022-01-31

Comments

Changed:

<
<

<--/commentPlugin-->

>
>

<--/commentPlugin-->

META FILEATTACHMENT	attachment="cuda_download.png" attr="" comment="" date="1643626582" name="cuda_download.png" path="cuda_download.png" size="52637" user="UliRaich" version="1"
META FILEATTACHMENT	attachment="nvidia.png" attr="" comment="" date="1643626582" name="nvidia.png" path="nvidia.png" size="13000" user="UliRaich" version="1"

Line: 64 to 72

META FILEATTACHMENT	attachment="libcudart.png" attr="" comment="" date="1643626913" name="libcudart.png" path="libcudart.png" size="37871" user="UliRaich" version="1"
META FILEATTACHMENT	attachment="gpu_good.png" attr="" comment="" date="1643626582" name="gpu_good.png" path="gpu_good.png" size="46971" user="UliRaich" version="1"
META FILEATTACHMENT	attachment="graphicsCard.png" attr="" comment="" date="1643626913" name="graphicsCard.png" path="graphicsCard.png" size="39171" user="UliRaich" version="1"

Added:

>
>

META FILEATTACHMENT	attachment="displayDriver.png" attr="" comment="" date="1695319151" name="displayDriver.png" path="displayDriver.png" size="104774" user="UliRaich" version="1"
META FILEATTACHMENT	attachment="GPU_warnings.png" attr="" comment="" date="1695320777" name="GPU_warnings.png" path="GPU_warnings.png" size="118104" user="UliRaich" version="1"
META FILEATTACHMENT	attachment="GPU_ok.png" attr="" comment="" date="1695323567" name="GPU_ok.png" path="GPU_ok.png" size="41303" user="UliRaich" version="1"

View topic | History: r5 < r4 < r3 < r2 | More topic actions...