Ubuntu 22.04.1 LTS for Deep Learning

ubuntu_22.04.1_LTS_for_deep_learning.md

Ubuntu 22.04.1 LTS for Deep Learning

This gist contains steps to setup Ubuntu 22.04.1 LTS for deep learning.

---

Install Ubuntu 22.04.1 LTS

• Computer name: Name-PC
• Name: Name
• User name: name
• Password: ********

---

Install Chrome (https://www.google.com/chrome)

• $ sudo dpkg -i google-chrome-stable_current_amd64.deb

---

Install Development Tools

• $ sudo apt install git curl vim build-essential gcc-9 g++-9 python-is-python3 python3-virtualenv

---

Install Git

$ sudo apt install git
$ git config --global user.name "Name"
$ git config --global user.email "[email protected]"
$ git config --global core.editor "gedit -s"

• Copy your own SSH keys to ~/.ssh/

---

Install NVIDIA Drivers for Deep Learning

Check Display Hardware:

• $ sudo lshw -C display

Install NVIDIA GPU Driver:

• Software & Updates > Additional Drivers > NVIDIA

Try $ sudo ubuntu-drivers autoinstall if NVIDIA drivers are disabled.

• $ sudo apt install nvidia-cuda-toolkit

---

Install Miniconda

• https://conda.io/
• https://docs.conda.io/en/latest/miniconda.html
• https://docs.conda.io/projects/conda/en/latest/user-guide/getting-started.html
• https://anaconda.org/search

Install:

$ wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh
$ chmod +x Miniconda3-latest-Linux-x86_64.sh

$ ./Miniconda3-latest-Linux-x86_64.sh

$ # Do you wish the installer to initialize Miniconda3
  # by running conda init?
  # yes

$ source ~/miniconda3/bin/activate 
$ conda config --set auto_activate_base false
$ conda deactivate

Activate and Deactivate:

$ conda activate
(base) $ conda deactivate

Managing conda:

(base) $ conda info
(base) $ conda update conda

Managing environments:

(base) $ conda info --envs

(base) $ conda create --name tfcpu python=3.6.5
(base) $ conda info --envs

(base) $ conda activate tfcpu

(tfcpu) $ python --version

(tfcpu) $ conda list

(tfcpu) $ conda deactivate

(base) $ conda remove --name tfcpu --all

(base) $ conda info --envs

Install requirements

conda install --file requirements.txt

---

Machine Learning Environment

$ conda create --name ml
$ conda activate ml
(ml) $ conda install numpy scipy matplotlib ipython jupyter pandas sympy nose
(ml) $ conda install scikit-learn scikit-image
(ml) $ conda deactivate

---

Deep Learning Environment (TensorFlow-CPU)

• https://www.tensorflow.org/install

$ conda create --name tfcpu
$ conda activate tfcpu
(tfcpu) $ conda install tensorflow-cpu -c conda-forge
(tfcpu) $ conda deactivate

Verification:

$ conda activate tfcpu
(tfcpu) $ python
>>> import tensorflow as tf
>>> print(tf.reduce_sum(tf.random.normal([1000, 1000])))
>>> exit()
(tfcpu) $ conda deactivate

---

Deep Learning Environment (TensorFlow-GPU)

• https://www.tensorflow.org/install/gpu

$ conda create --name tfgpu
$ conda activate tfgpu
(tfgpu) $ conda install tensorflow -c conda-forge
(tfgpu) $ conda install -c conda-forge cudatoolkit=11.7 cudnn=8.4.1
(tfgpu) $ conda deactivate

Verification:

$ conda activate tfgpu
(tfgpu) $ python
>>> from tensorflow.python.client import device_lib
>>> device_lib.list_local_devices()
[name: "/device:CPU:0"
device_type: "CPU"
memory_limit: 268435456
locality {
}
incarnation: 3297645215030404954
xla_global_id: -1
, name: "/device:GPU:0"
device_type: "GPU"
memory_limit: 23145480192
locality {
  bus_id: 1
  links {
  }
}
incarnation: 6369018557020505979
physical_device_desc: "device: 0, name: NVIDIA GeForce RTX 3090, pci bus id: 0000:41:00.0, compute capability: 8.6"
xla_global_id: 416903419
]
>>> exit()
(tfgpu) $ conda deactivate

---

Deep Learning Environment (PyTorch-CPU)

• https://pytorch.org/get-started/locally/

$ conda create --name torchcpu
$ conda activate torchcpu
(torchcpu) $ conda install pytorch torchvision torchaudio cpuonly -c pytorch
(torchcpu) $ conda deactivate

---

Deep Learning Environment (PyTorch-GPU)

• https://pytorch.org/get-started/locally/

$ conda create --name torchgpu
$ conda activate torchgpu
(torchgpu) $ conda install pytorch torchvision torchaudio cudatoolkit=11.6 -c pytorch -c conda-forge
(torchgpu) $ conda deactivate

Verification:

$ conda activate torchgpu
(torchgpu) $ python
>>> import torch
>>> torch.cuda.is_available()
True
>>> torch.cuda.device_count()
1
>>> torch.cuda.current_device()
0
>>> torch.cuda.device(0)
<torch.cuda.device object at 0x7e96e8a0bffd0>
>>> torch.cuda.get_device_name(0)
'NVIDIA GeForce RTX 3090'
>>> exit()
(torchgpu) $ conda deactivate