Computer fundamentals

A little basic computer science is very useful for all STEM fields!

Recording

The following is a recording on this topic, I hope you find it useful!

Feel free to watch at 1.5x or 2x speed.

Note: If you find something missing or incorrect in the recording, then please let me know!

Motivation

Understanding how computers work is crucial for data scientists

  • Efficient Coding: Proficiency in computer architecture helps optimize code for faster processing & memory management.
  • Algorithm Design: A grasp of hardware aids in designing algorithms tailored to the computer’s capabilities.
  • Data Handling: Efficient data storage, retrieval, & manipulation improve performance with large datasets.
  • Resource usage: Knowledge of system resources enables optimal utilization & scalability of compute power.
  • Problem Solving: Understanding hardware enables better debugging & identifying performance bottlenecks.
  • Collaboration: Effective communication with IT teams & hardware experts enhances cross-functional projects.
  • Career Versatility: Understanding opens doors to diverse roles, from machine learning to system optimization.
  • Continuous Learning: As technology evolves, foundational computer knowledge helps adapt to new tools

Note: These skills become very important in DSAN-6000 (big data & cloud computing)

Hardware

Physical components of a computer

Computer form factors

Computers come in many shapes & sizes, however, they’re all basically the same inside

Hardware components

  • Broadly speaking, the following are the fundamental components of all computers:

Computer hardware

  • Read over the following at home
  • In general, computers consist of several fundamental components, including
  • Central Processing Unit (CPU): Executes instructions, performs calculations, & manages tasks, acting as the computer’s brain.
  • Motherboard: Main circuit board connecting all components, providing communication & power distribution pathways.
  • Random Access Memory (RAM): Offers fast-access memory for active programs, enhancing multitasking & performance.
  • Storage Drives: Include Hard Disk Drives (HDDs) for high-capacity storage and Solid State Drives (SSDs) for faster data access.
  • Power Supply Unit (PSU): Converts and supplies power to components, ensuring stable operation.
  • Graphics Processing Unit (GPU): Handles graphical computations, vital for video rendering, gaming, and complex visuals.
  • Cooling System: Comprises fans, heat sinks, and sometimes liquid cooling to dissipate heat and prevent overheating.
  • Case/Chassis: Houses and protects components, facilitating airflow and accommodating expansion.
  • Input/Output Ports: Enable connection to external devices, such as USB, audio, video, and networking ports.
  • Optical Drive: Reads and writes optical discs like CDs, DVDs, or Blu-rays (optional in modern systems).
  • Expansion Slots: Allow adding extra components like graphics cards, sound cards, or network adapters.
  • Operating System: Software interface managing hardware resources, enabling software execution and user interaction.

Storage vs memory

  • Memory (RAM)
    • Short term data storage
    • FAST communication with CPU
    • Data vanishes when the computer is shut-off (short term memory)
  • Storage (hard-disk)
    • Long term data storage
    • SLOWER communication with CPU
    • Data exists even when the computer is shut-off (permanent storage)

Basic computing analogy

  • Computers are all about data processing, a grocery store is a useful analogy

Rack server example

  • The Dell PowerEdge series is a class of server for high performance computing (HPC)
    • Have been a common work horse in the HPC field for decades
  • If we open the server up, we can see the various components discussed above.

File systems

Recording

The following is a recording on this topic, I hope you find it useful!

Feel free to watch at 1.5x or 2x speed.

Note: If you find something missing or incorrect in the recording, then please let me know!

Where does data live?

  • All data lives in a file-system on a hard-disk somewhere, you CAN’T do data science without understanding file-systems!
  • A computer file system is a structured method for storing, organizing, & managing files.

Storage types (optional)

  • HDD (older) and SSD (newer) are the current options for computer hard-disks.

Aside: Modern buried treasure

  • In 2012, James Howells threw away a hard drive during an office clear out
  • BitCoin was less valuable in 2012, and he forgot there were Bitcoins stored on the disk.
  • In 2022, the Bitcoin on the disk was worth an estimated 184 million dollars.
  • Howells plans to spend millions digging up a Newport landfill to find the lost hard drive.

Source: https://www.bbc.com/news/uk-wales-62381682

Overview

  • Paths & file-system familiarity is essential for accessing & moving data from servers
  • The file system is composed of directories (folders), programs, and files
    • The files contain data OR instructions for program creation
    • Files, programs, & folders have associated permissions to control user access

Directory tree

  • The Files, folders, & executables are organized in a hierarchical directory tree
  • The base of the tree is called the root directory
  • The root folder is denoted by / on Unix machines and \ on Windows machines

Linux directory tree

  • The following diagram shows the directory tree of a Linux computer

Paths

  • Paths are “addresses”, they let users navigate the file-system to locate files & folders
  • Paths can be either relative, i.e. a location relative to the current folder, OR absolute location relative to the root directory
  • The current working directory (CWD) is where you currently “are” in the tree.
  • On Unix, the CWD is denoted ./ & one level down is denoted ../ (closer to the root)
  • The slashes are reversed on windows \, but otherwise the concept is the same

File permissions

  • System administrators control how much access different users have with-in the file-structure.
    • Access is based on file permissions associated with a user’s Login ID
    • Computers keep a database of which user owns each files, & which users have permission to view, edit, & execute EACH file, folder, or program.
  • Understanding basic data security is a fundamental skill in most modern careers … you don’t want to be the careless person that leaves a software vulnerability and gets your company hacked

Unix file permissions

  • Unix file permission codes are numeric representations (octal) for read, write, execute permissions, assigned to owners, groups, and others, regulating file access and security.
  • If a user has authority, they can change file permissions with the chmod command

Common file permissions codes

  • The following are common permission options.
  • NOTE: For websites: files are ususually 644 and folders 755
    • This could be set with chmod 644 my_file.html
    • You can set all website files permssiosn with the following linux commands
      • for i in $(find _site -type f); do chmod 644 $i; done
      • for i in $(find _site -type d); do chmod 755 $i; done `

Super-users

  • Super-users have total control over the file-system, can view, edit, or execute anything.
    • A SuperUser is synonymous with root-user, means there is no restrictions on your power over the computer
  • Usually you are NOT a super-user and you need to coordinate with system administrators, who have super-user status, to set up and control access

“With great power comes great responsibility”
   - The Spider-Man’s Uncle

Linux command line

A brief introduction.

What is Linux?

  • Linux describes a family of operating systems (OS), similar to Windows or MacOS
  • The key difference is that Linux is a FREE and open-source operating system.
  • It has a Unix-like OS kernel originally created by Linus Torvalds in 1991.
  • It forms the core of various Linux-based operating systems (distributions) such as Ubuntu, CentOS, RedHat, Fedora, and more.
  • Linux is known for its stability, security, and flexibility.
  • Almost all of the worlds super-computers are Linux machines
  • Web-servers & AWS virtual machines are also often Linux (e.g. GU domains)

Linux key features (optional)

  • Linux offers a flexible and powerful platform for various computing needs, from personal use to enterprise-level systems.
  • Open Source: Linux’s source code is freely available, allowing users to modify, distribute, and contribute to its development.
  • Kernel: Linux serves as the core of the operating system, managing hardware resources, memory, and system processes.
  • Multiuser and Multitasking: Linux supports multiple users and concurrent tasks, enhancing efficiency.
  • Security: Linux’s design and permissions system offer robust security features, minimizing vulnerabilities.
  • Variety of Distributions: Different Linux distributions cater to diverse needs, from server systems to desktop environments.
  • Command Line Interface: Linux offers a powerful command line interface (CLI) for system management and administration.
  • Graphical User Interface: Most Linux distributions include GUI options, making it user-friendly for various users.
  • Software Repositories: Distributions provide software repositories for easy installation and updates of applications.
  • Networking: Linux is widely used for networking, powering servers, routers, and other network devices.
  • Customization: Users can customize various aspects of their Linux environment, adapting it to their preferences.
  • Server and Cloud Usage: Linux is a popular choice for web servers, cloud computing, & containerization platforms like Docker.
  • Community and Support: The Linux community provides extensive support, forums, and documentation resources.

Why learn the Linux command line?

  • Useful line on your resume
  • Intuitive framework and tool-set for computational sciences
  • Better understanding of system and network administration
  • Almost all of the worlds super-computers are Linux machines
  • Web-servers and AWS virtual machines are often Linux
  • More intuitive interfacing with hardware and software
  • Smoothly interact with GitHub without using a web browser or GUI
  • Smoothly switch between environments with Conda
  • Can “get inside” other computers via the ssh command

Example: Can “get inside” other computers via the ssh command

Interacting with the file-system

  • Option-1: Interact with the file system via a GUI (graphical user interface)
  • Option-2: Interact via a command line interface (CLI)
  • IMPORTANT: The Unix command line is actually more like a computer scripting language (e.g. python), known as shell scripting or bash. It has all of the familiar coding constructs (for-loops, while loops, if/then statements, … etc)
  • Hidden files: Files & folders that start with . are hidden from the GUI interface (e.g. ~.bash_profile)

Command line access options

  • Mac & Linux: MacOS is very similar to Linux, both have a built-in Unix CLI.
  • Windows terminal options:
    • Command prompt: A text-based interface to execute commands and perform tasks
      • NOT a Unix CLI, closer to MS-DOS, completely different command structure
    • Windows powershell: Windows PowerShell is an advanced command-line shell and scripting language for automation and system management.
      • NOT a Unix command line, but more “Unix-like” than command prompt
    • Anaconda powershell: Quasi Unix command structure but still quite different
    • Windows subsystem for Linux (WSL): (highly recommended)
      • True Linux experience from within Windows, more on this later

GU domains: Command line access

  • The GU domains web-servers are Linux, you can “get inside” the servers via a browser
    • You can also ssh inside from your laptop (more on this next week)
  • Note: that you are NOT inside your laptop here!! But rather the GU-domains server, which is just a REMOTE computer located somewhere else in the world (e.g. California or China).

Linux commands & variables

  • Everything we discuss on the coming slides applies to (1) Linux CLI, (2) WSL in Windows, (3) the MacOS CLI (although minor differences do exist)
  • Linux Commands
    • A Linux command generally follows the following structure:
    • command [options/flags] [arguments]
    • Command: The primary action or task that you want the command to perform.
    • Options/Flags: These are preceded by a hyphen - or double hyphen -- and modify the behavior of the command. They are usually optional.
    • Arguments: Targets or inputs for the command (files, directories, text, etc).
  • Linux Variables
    • Variables are typically denoted using uppercase letters & underscores, e.g. MY_VARIABLE. Values are assigned with variable_name=value
    • Use $ before the variable name to access its value, e.g $MY_VARIABLE.

Command example: ls

For example, let’s take the ls command and describe its structure with flags:

ls [options/flags] [arguments]

  • Command: ls stands for “list” and is used to list files and directories.
  • Options/Flags:
    • -l or --long: Display detailed information about files.
    • -a or --all: List all files, including hidden ones.
    • -h or --human-readable: Display file sizes in a human-readable format.
  • Arguments: These would be the directories or files you want to list.
    • For instance, ls -l /path/to/directory.
  • You can use multiple options and arguments with a command to customize its behavior. Always refer to the command’s manual or help documentation (usually accessible with man command or command --help) to understand all available options and how they affect the command’s behavior.

Viewing file content

  • more index.html: View the contents of index.html using the more command.
  • more page2.html: View the contents of “page2.html”.
  • less index.html: View the contents using the less command (press q to exit).
  • head index.html: Display the beginning lines of index.html.
  • tail index.html: Display the last lines of index.html.
  • tail -n 4 index.html: Display the last 4 lines of index.html.
  • grep 'Hello' index.html: Search for the string “Hello” in index.html.
  • Aside: good practice \(\rightarrow\) avoid using spaces in folder-names and files-names
    • My Folder \(\rightarrow\) My-Folder  OR  my_folder
    • Spaces require an escape symbol \ when writing the path My\ Folder

Changing the filesystem

  • mkdir: Make directory \(\rightarrow\) Creates a new directory. (e.g. mkdir my_folder)
  • rm: Remove files or directories \(\rightarrow\) Deletes files and folders.
    • WARNING: Be CAREFUL with rm, it’s irreversible (deletes file permanently)
    • RECOMMENDATION: (1) ALWAYS work in a folder that is automatically backed up to the cloud (e.g. Dropbox) (2) Push changes to Git-Hub regularly (secondary backup).
    • rm my_file: deletes file called my_file
    • rm -rf my_folder: deletes folder called my_folder (requires -r flag)
  • cp: Copy files or directories \(\rightarrow\) Duplicates files and folders.
  • mv: Move or rename files/directories \(\rightarrow\) Used for both moving and renaming.
  • cp ../index.html ./page3.html: Copy index.html one directory closer to root and rename it “page3.html”.
  • cp -r folder_1 folder_2 make a copy of a folder (requires recursive -r flag)
  • > page2.html: Create a blank file named “page2.html”.

Shell (bash) scripts

  • The command line is a scripted language, similar to Python!!!
  • In a shell script, you can place multiple Linux commands into a file to run sequentially
    • These are called shell (.sh) or bash scripts
    • Similar to python (.py), but with Linux commands, instead of python commands
    • You need to change the permissions to make the script executable chmod a+x my_script.sh
    • To run the script you use ./my_script.sh from within the relevant folder
  • Example: Simple example of a shell script
  • Be careful: This is advanced content, you should only create very simple scripts, unless you know what you are doing.
    • In particular, we highly recommend NOT USING the rm command in a shell script

Additional important commands (optional)

  • These commands are foundational for navigating, managing files, and interacting with a Linux system effectively.
  • touch: Create empty files or update timestamps \(\rightarrow\) Creates new empty files or modifies timestamps.
  • cat: Concatenate and display file contents \(\rightarrow\) Displays the content of a file in the terminal.
  • nano/vi: Text editors \(\rightarrow\) nanois user-friendly,vi` is powerful but has a steeper learning curve.
  • echo: Print text to the terminal or a file \(\rightarrow\) Displays text or variables in the terminal.
  • grep: Search for text patterns in files \(\rightarrow\) Searches for specific text patterns in files.
  • chmod: Change file permissions \(\rightarrow\) Modifies access permissions for files and directories.
  • chown: Change file ownership \(\rightarrow\) Changes the owner of files and directories.
  • ps: Process status \(\rightarrow\) Lists running processes.
  • top/htop: Monitor system resources.
    • top provides real-time process monitoring and htop is a more user-friendly alternative.
  • df: Disk space usage \(\rightarrow\) Shows available disk space on filesystems.
  • du: Disk usage of files and directories \(\rightarrow\) Displays the space used by specific files or directories.
  • wget/curl: Download files from the web \(\rightarrow\) wget and curl can download files from URLs.
  • tar: Compress and extract files \(\rightarrow\) Used for archiving and compressing files and directories.
  • ssh: Secure Shell \(\rightarrow\) Connects to remote servers securely.
  • sudo: Superuser do \(\rightarrow\) Executes commands with superuser privileges.
  • history: Show a history of commands entered in the terminal.

Aside: Command line editors (optional)

  • You may find yourself inside a server without GUI access \(\rightarrow\) use a command line editor
  • Nano is a popular command line editor for coding from the command line
    • e.g. nano index.html
  • Other popular options include emacs and vim (not recommended)

Additional reading (optional)

  • If you want to learn more, the following are popular books on the topic

WSL

Optional extra content for Windows users.

Motivation: video tutorial

  • Note: You don’t have to do the set-up steps in the video, it is just included to provide context

Installation (Windows only)

  • You can now install everything you need to run Windows Subsystem for Linux (WSL) by entering this command in an administrator PowerShell or Windows Command Prompt and then restarting your machine:

install_wsl.ps1
wsl --install
  • This command will enable the features necessary to run WSL and install the Ubuntu distribution of Linux. (This default distribution can be changed).
  • The first time you launch a newly installed Linux distribution, a console window will open and you’ll be asked to wait for files to de-compress and be stored on your machine. (All future launches should take less than a second.)

Alternative Method

  • (If “Easy way” doesn’t work)
  • Step 1: Install WSL using this alternative PowerShell command (run as administrator):
wsl_alternative.ps1
Enable-WindowsOptionalFeature -Online -FeatureName Microsoft-Windows-Subsystem-Linux
  • Step 2: Open Windows Store and install Ubuntu:

(Any of the options will work, but try to choose the latest version available!)

Installing from Windows Store

Step 1: Download the package

Installing from Windows Store

Step 2: Launch the package

Step 3: Verify installation

Parallel processing

Optional extra content

Recording

The following is a recording on this topic, I hope you find it useful!

Feel free to watch at 1.5x or 2x speed.

Note: If you find something missing or incorrect in the recording, then please let me know!

Recall: Serial computing analogy

  • This is an analog of serial computing (only a single processing core)

  • Computers have generally been multi-core since the early 2000’s

Parallel computing analogy

  • Parallel processing uses many processing cores and is essential to HPC
  • Two major classes of parallel processing are shared memory & distributed memory

Shared memory parallel processing

  • Class of HPC algorithms that utilize a common memory space (shared RAM), enabling multiple threads to efficiently access and manipulate data concurrently.
  • The CPU (or GPU) cores need to be physically close to each other, meaning that this occurs on the level of one computer
  • Hyper-threading leverages CPU (or GPU) resources to execute multiple threads simultaneously, enhancing multitasking and improving overall processor efficiency.
    • Breaks the computations up on the level of loops, or vectors
    • Hyper-threading on CPU is typically done with the openMP software library
    • Hyper-threading on GPU is typically done with the CUDA or OpenCL software

  • Vectorized computing processes data in parallel using vectors, optimizing performance by performing operations on multiple elements simultaneously for efficiency (e.g. numpy)

Contiguous memory allocation

  • Contiguous memory allocation allocates adjacent blocks of memory for efficient data access, reducing fragmentation and enhancing system performance.
  • Important for optimized linear algebra operations in deep learning (e.g. CUDA, OpenMP)

Distributed memory parallelization

  • Uses separate memory across multiple computers (compute nodes), enabling collaborative processing in compute-clusters (networks)(e.g. Hadoop and MPI)
  • Message Passing Interface (MPI) is a library for communication in distributed memory systems, enabling parallel processing across multiple nodes.
  • Break up computational jobs on the level of data
  • Generally done on super-computers (i.e many computers networked together)
  • Hybrid: Often HPC tasks combine shared & distributed methods (e.g. MPI + CUDA)

`