(Lecture 2: 8/26) Desktop Computer Hardware, Processor Architectures, and Infrastructure Notes

Course Announcements & External Resources

• Canvas Media & Podcast Links

  • Announcements posted on Canvas include direct links to external video and audio resources, such as YouTube and Apple Podcasts.

  • Required listening includes a business and economics podcast episode focusing on the viability of deploying data centers into outer space.

  • This concept originated from formal conversations with the "Moonshot" group at Google.

• Definition of a Moonshot Project

  • A Moonshot refers to an exceptionally ambitious, highly difficult project designed to address a complex problem (frequently termed a "wicked problem").

  • Key characteristics:

  • Probability of success is extremely small.

  • If solved successfully, the outcome is completely transformational to business and society.

  • Exam Note: The formal definition and characteristics of a "moonshot" are explicit examination topics.

• Freakonomics & Economic Frameworks

  • Created by economists Steve Levitt and Stephen Dubner (from the University of Chicago), introduced in academic coursework by Dr. Dow.

  • Freakonomics examines core economic and business principles through accessible, non-traditional real-world scenarios.

• Microsoft Teams Integration

  • Canvas contains an active access request link for the dedicated Microsoft Teams channel.

  • Accessing this channel provides full access to archived class meeting recordings, useful for reviewing lecture materials prior to examinations and quizzes.

Desktop Computer Architecture & Hardware Overview

• Desktop Computer Enclosure Terminologies

  • Common terms for the physical enclosure housing compute hardware include the case, tower, hardware unit, or colloquially (though technically imprecise) the CPU.

  • The enclosure houses the primary processing power, internal memory, power conversion systems, and system storage.

• Peripheral Devices Classification

  • Any external hardware device connected to the central computer enclosure is classified as a peripheral.

  • Input Peripherals: Devices designed to feed data, commands, or signals into the computer system.

  • Examples: Keyboard, Mouse, Microphone.

  • Touchscreen monitors function as input peripherals when interacting via direct physical contact.

  • Output Peripherals: Devices designed to output, display, or physically manifest data processed by the computer.

  • Examples: Standard Monitors, Displays, Projectors, Audio Speakers, Hardware Printers, Indicator Lights.

  • CNC Machine (Computer Numerically Controlled): A computer-controlled machining tool utilized in industrial manufacturing to sculpt or manufacture physical parts, serving as a specialized output peripheral.

  • Hybrid Peripherals: Touch-interface monitors act as both input devices (touch gestures) and output devices (visual display).

Electrical Power Standards & Physical Connectivity

• Universal Serial Bus (USB) Interface

  • USB represents a standardized physical interface created to replace dozens of legacy proprietary connectors with a single universal protocol.

  • Standard USB-A and USB-B ports operate at a regulated output voltage of $5\,\text{V}$ DC (Direct Current).

• Alternating Current (AC) vs. Direct Current (DC) Standards

  • Alternating Current (AC): High-voltage line current supplied directly from municipal electric utilities through standard wall outlets.

  • North American Standard: $110\,\text{V}$ to $120\,\text{V}$ AC at $60\,\text{Hz}$.

  • European / UK Standard: $220\,\text{V}$ to $230\,\text{V}$ AC (or universal operational range of $110\,\text{V}$ to $240\,\text{V}$ AC).

  • Direct Current (DC): Stable, continuous low-voltage direct power required by internal electronic microprocessors, microchips, and integrated circuits.

  • Modern desktop internal computer components primarily run on $12\,\text{V}$ DC.

  • External charging blocks ("wall warts") convert high-voltage AC wall power down to low-voltage DC power (e.g., transforming $110\text{--}240\,\text{V}$ AC input into $5\,\text{V}$ DC output at $3\,\text{A}$).

• Case Toolless Design

  • A toolless case features chassis latches, side panel thumbscrews, and modular slots that allow complete internal hardware assembly and disassembling without requiring screwdrivers or manual tools.

Non-Volatile Storage Technologies & Media Systems

• Solid-State Drives (SSD)

  • SSDs represent modern non-volatile, persistent storage technology that retains data permanently regardless of power state.

  • Common consumer desktop standard: $250\,\text{GB}$ SATA SSD.

  • SATA (Serial Advanced Technology Attachment): A internal bus interface protocol connecting storage drives to the computer motherboard using ribbon data cables.

  • Features: SSDs use solid-state flash memory without moving parts, offering substantially faster read/write speeds, reduced power draw, and physical durability (resistant to impact, drops, and temperature fluctuations).

• Data Volatility vs. Non-Volatile Persistence

  • Volatile Data: Data stored temporarily that is instantly lost or erased the moment power is disconnected (e.g., Random Access Memory).

  • Non-Volatile Data: Data stored permanently that persists indefinitely after full power down (e.g., SSD, HDD).

• Data Destruction & Forensic Persistence

  • Persistent digital data on solid-state media is extremely resilient to physical destruction.

  • Government agencies (FBI, NSA) and specialized forensic recovery contractors possess technology capable of extracting usable data from physically burned SSD drives.

  • Extreme thermal destruction methods—such as burning with high-temperature thermite (a pyrotechnic composition used in heavy cast iron industrial welding)—are frequently recoverable via specialized laboratory forensic analysis.

• Hard Disk Drives (HDD)

  • Legacy storage technology utilizing spinning magnetic platters read by mechanical physical heads.

  • Drawbacks relative to SSDs: Slower performance, higher mechanical failure rates, larger physical footprint, and higher overall energy consumption.

  • Purchasing Advice: Machines utilizing HDDs as primary system drives should be avoided in modern computing environments ($2026$ onward).

• Optical Media Technologies (CD-ROM / DVD-ROM)

  • Compact Disc (CD): Legacy digital optical format.

  • Standard Storage Capacity: $700\,\text{MB}$ (equivalent to roughly $6\,\text{hours}$ of compressed audio).

  • Digital Data Storage Unit Conversion Metrics:

  • $1000\,\text{KB} = 1\,\text{MB}$

  • $1000\,\text{MB} = 1\,\text{GB}$

  • Formula comparison: A single $250\,\text{GB}$ SSD ($250000\,\text{MB}$) stores an equivalent capacity of approximately $320$ to $357$ individual optical CDs:       $\frac{250000\,\text{MB}}{700\,\text{MB}} \approx 357.14$

  • Physical vulnerabilities of optical media: Scratching, thermal warping/melting from direct solar exposure, and microwave radiation destruction.

• Economic Trade-offs: Cloud Streaming vs. Physical Media

  • Advantages of Streaming (Spotify, Apple Music): Immediate, frictionless physical convenience; instant global access to vast media libraries (e.g., Dolly Parton catalog) without physical storage footprint or thermal damage risk.

  • Disadvantages of Streaming:

  • Lack of Ownership: Consumers lease streaming access rights rather than owning perpetual digital assets. Platforms can revoke access arbitrarily (e.g., Sony PlayStation removing over $500$ purchased digital movie licenses without financial compensation).

  • Perpetual Cost Structure: Ongoing recurring monthly subscription fees (e.g., $4\,\text{USD}$ to $5\,\text{USD}$ per month indefinitely) instead of a single physical product purchase.

Internal System Components & Cooling Dynamics

• Power Supply Unit (PSU)

  • An essential internal desktop hardware block designed to step down line voltage AC ($110\,\text{V}\text{--}120\,\text{V}$ AC) and rectify it into clean, multi-rail DC output ($12\,\text{V}$ DC and $5\,\text{V}$ DC) to power internal active electronics.

• Thermal Management Systems

  • High-density silicon chips generate substantial heat during electrical switching.

  • Heat Sink: A heavy metal mass (typically aluminum or copper) designed with micro-fins to maximize surface area, conducting heat away from processing chips via thermal conduction.

  • Cooling Fan: Mounted adjacent or directly on top of the heat sink to force ambient air across fins, transferring thermal energy out of the enclosure via forced convection.

• Central Processing Unit (CPU)

  • The primary logic and calculation processor ("brain") of the computer, situated beneath the main cooling fan and heat sink on the motherboard.

  • Operates at a fundamental level in binary ($1$s and $0$s, representing ON and OFF states), driven by billions of microscopic switching transistors on a silicon die.

• Graphics Processing Unit (GPU) & Video Cards

  • Dedicated processing expansion card traditionally engineered for rendering graphical outputs, 3D textures, and computer video displays (e.g., legacy $2\,\text{GB}$ models such as the GeForce GT series).

  • Modern GPU Applications: Modern artificial intelligence workloads, deep learning algorithms, and massive parallel neural network processing.

  • Industrial GPU Market Standards:

  • High-performance enterprise units (e.g., NVIDIA A100) are explicitly optimized for large language model (LLM) processing.

  • Pricing: Secondary/used market prices for individual NVIDIA A100 GPUs range from $3,900\,\text{USD}$ to $18,750\,\text{USD}$, depending heavily on cumulative operational hours and electrical degradation.

• Data Center Scale & Artificial Intelligence Infrastructure Economics

  • Institutional Computing Nodes: University computing infrastructure often operates dynamic clusters (e.g., $12$ to $15$ active nodes of NVIDIA A100 GPUs), representing equipment investments approaching $400,000\,\text{USD}$ for processor blocks alone.

  • Hyper-Scale Data Centers:

  • Example: xAI Colossus Data Center located in Memphis.

  • Facility Construction Cost: Approximately $37\,\text{billion}\,\text{USD}$.

  • Global Market Projections:

  • Approximately $158$ hyper-scale data centers are scheduled for construction across the United States.

  • Total economic commitment calculation for large-scale data center infrastructure:       $158 \times 37\,\text{billion}\,\text{USD} = 5.846\,\text{trillion}\,\text{USD}$

  • Including smaller regional edge nodes, over $2,000$ AI data center facilities are planned nationwide.

System Memory (RAM) & Processor Architecture Standards

• Random Access Memory (RAM)

  • Represents high-speed, volatile active working memory.

  • Functional Role: Temporarily holds active operating system files, software programs, and multi-tasking workspace data, allowing smooth execution across applications.

  • Hardware Specifications:

  • Legacy benchmark shown: $4\,\text{GB}$ DDR3/DDR4 desktop module.

  • Older laptop implementations (e.g., legacy MacBook Air models older than $18$ months) frequently shipped with $8\,\text{GB}$ RAM.

  • Modern baseline recommendation for general computing ($2026$ baseline): Minimum of $16\,\text{GB}$ RAM.

  • High-end industrial computing baseline (AutoCAD processing, Adobe Premiere 4K video rendering): $64\,\text{GB}$ RAM or higher (module costs exceeding $1,000\,\text{USD}$).

• Motherboard Architecture

  • The primary internal Printed Circuit Board (PCB) that physically anchors and electrically interconnects the CPU, RAM, storage drives, GPU, PSU connectors, and peripheral expansion buses.

  • Soldered RAM: In modern ultrabooks and thin-and-light laptops, RAM chips are soldered directly onto the motherboard PCB, making aftermarket memory upgrades impossible.

• Processor Architectures: x86 vs. ARM vs. RISC-V

  • x86 Architecture (Intel, AMD):

  • The legacy instruction set architecture dominant in personal desktop PCs and laptops for over 20 years.

  • Highly powerful for legacy desktop computing, but generally higher power consumption.

  • ARM Architecture (Apple Silicon M1/M2/M3, Qualcomm Snapdragon, Surface Pro X):

  • Uses Unified Memory Architecture (UMA) for high efficiency.

  • Dynamically disables inactive processor cores (similar to multi-cylinder engines shutting down unused cylinders), enabling multi-day battery operational runtime on modern portable hardware.

  • Business Licensing Model: Device manufacturers must pay licensing royalty fees to the ARM Foundation for every manufactured processor chip.

  • RISC-V (RISC 5):

  • An open-source, license-free instruction-set processor architecture engineered to compete directly against x86 and ARM without royalty overhead.

Single-Board Computers & Internet of Things (IoT) Applications

• Raspberry Pi 0 W (Zero W) Architectural Specifications

  • A fully functional, complete single-board computer measuring only a few inches.

  • Operational Specifications:

  • CPU Clock Speed: $900\,\text{MHz}$.

  • Onboard Memory: Up to $4\,\text{GB}$ RAM.

  • Integrated Features: Onboard Wi-Fi adapter, USB ports, GPU display interface, and GPIO (General Purpose Input/Output) breakout header pins to read discrete and analog sensor signals.

  • Pricing History: Commercial launch price was $7.99\,\text{USD}$ per unit (modern production models average around $15.00\,\text{USD}$).

• Microcontrollers & Low-Cost Embedded Compute Capabilities

  • Microcontrollers can be procured in bulk for $1.00\,\text{USD}$ to $2.00\,\text{USD}$ per chip.

  • Sub-$15$ dollar full computing devices allow dense industrial and commercial deployments in physical spaces where traditional $1,200\,\text{USD}$ to $1,500\,\text{USD}$ computers are cost-prohibitive.

• Commercial & Business Analytical Scenarios

  • Automated Class Digital Signage: Replacing physical wall clocks (e.g., Primex Wireless wall clocks costing roughly $80.00\,\text{USD}$) with an $8.00\,\text{USD}$ Raspberry Pi paired with a cheap display to continuously present live class schedules, countdown timers, and automated QR attendance codes.

  • Stealth Foot-Traffic Market Research (Real-World Case Study):

  • A private commercial food vendor evaluated opening a retail location inside a university library.

  • The company deployed a hidden $15.00\,\text{USD}$ Raspberry Pi connected to wall outlet power behind a library trash container.

  • The board operated continuously on the wireless network, silently calculating ambient Wi-Fi network telemetry to log daily student foot traffic volume.

  • The captured foot-traffic data provided critical business intelligence worth thousands of dollars at a hardware loss risk of less than $15.00\,\text{USD}$.

  • Internet of Things (IoT): The network of physical object hardware embedded with micro-computing platforms, sensors, and network connectivity (e.g., consumer Ring doorbells, which contain approximately $12.00\,\text{USD}$ in internal core hardware technology while retailing near $100.00\,\text{USD}$).

Questions, Answers & Class Discussion

• Discussion on GPU Service Life and Operational Longevity

  • Question: How does cumulative run-time impact used GPU pricing in AI markets?

  • Answer: Silicon chips degrade under heavy continuous operational thermal loads. High cumulative operational hours increase component failure probability, causing heavy price discounting on secondary hardware listings ($3,900\,\text{USD}$ vs. $18,750\,\text{USD}$).

• Discussion on Retail Sales Tactics for Consumer Electronics

  • Question: Why do big-box retail associates over-recommend enterprise or high-end computer configurations to non-technical consumers?

  • Answer: Retail associates are heavily driven by store credit card activation quotas and sales commission structures, aligning their incentives toward selling high-margin gaming rigs to low-demand users (e.g., web-browsing customers).

• Discussion on Campus AI Research Computing Resources

  • Question: How do students access high-performance campus GPU nodes for specialized machine learning research?

  • Answer: Student compute allocation requests are handled through the institutional Office of Research Computing in coordination with Microinfo contacts via designated Microsoft Teams channels.