CPU Design: ALU, CU, Registers, and Processing

CPU Design

• The CPU comprises the Arithmetic Logic Unit (ALU), Control Unit (CU), and Registers.

Control Unit (CU)

• The CU is responsible for controlling the execution of each instruction.

Arithmetic Logic Unit (ALU)

• The ALU is designed to add, and subtraction needs a negative number to be added.
• The ALU handles logical comparisons like True/False and performs general calculations.

Register

• The Register stores the current instructions and data being processed.

Serial Processing

• Serial processing involves processing one instruction at a time, utilizes one core, and breaks the problem into sequentially executed instructions.

Parallel Processing

• Parallel processing entails running program instructions simultaneously on multiple microprocessors to reduce processing time.
• Instructions are executed on different processors, such as when a user runs multiple programs concurrently.

Hyperthreading

• A single processor chip duplicates its registers to preload the next instructions into a second set of registers.
• This enables two sets of instructions to run concurrently, referred to as a thread.
• Context switching between processes is hastened by removing data saving and loading times due to the appearance as two logical CPUs within one physical CPU with dual register sets.

Multiprocessing

• Involves having multiple CPUs, referred to as cores, on a single CPU chip, which duplicates the whole core.
• More cores allow for more parallel processes.
• Effectiveness requires support from both the motherboard and the operating system.

Processor Cache

• Processor speed is faster than RAM speed.
• Cache memory is made from Static RAM (SRAM) chips with faster access speeds than Dynamic RAM (DRAM).
• Cache memory temporarily retains pre-fetched instructions and data from RAM, anticipating their next use by the CPU.
• Level 1 (L1) cache is a small memory inside the CPU that operates at the same speed as the CPU's internal speed.
• Level 2 (L2) cache is larger than L1, located slightly further away, and operates at half the CPU's internal speed between the CPU and RAM.
• Level 3 (L3) cache is found on motherboards, is larger, slower than L1/L2 and closer to the CPU than RAM.
• The CPU searches for information in the order of L1, then L2, then L3, and slowest from RAM.
• L3 cache's purpose is dependent on the CPU's design.
• Cache memory stores data for quicker future access, resulting in a cache hit when data is found. A cache miss transpires when data isn't in the cache.

RAM (Random Access Memory)

• RAM exists as chips on modules plugged into dedicated slots.
• RAM stores currently used programs and data linked to it.
• DDR (Double Data Rate) is a commonly used type of RAM.
• DRAM is dynamic RAM and is the chip used for memory.
• SDRAM synchronizes RAM chips with the CPU, eliminating wait times during refreshing, which allows for faster CPU instruction feeding.
• DDR transfers data twice as fast per clock pulse compared to single data rate transfers..
• Faster RAM speed leads to quicker instruction delivery to the cache.

Latency

• Latency measures the time it takes for a component to respond; RAM has varied latency, which can cause wasted clock ticks.
• Fewer clock ticks indicate a faster processor.

Types of Memory

• L0: registers hold words retrieved from cache.

• L1: on-chip L1 cache holds cache lines retrieved from L2 cache.

• L2: off-chip L2 cache holds cache lines retrieved from memory.

• L3: main memory (DRAM) holds disk blocks retrieved from local disks.

• L4: local secondary storage (local disks) hold files retrieved from remote network servers.

• L5: remote secondary storage (distributed file systems, Web servers)

• Storage devices are larger, slower, and cheaper as you go down the list. The reverse is true they are smaller, faster and more costly.

Motherboard

• Connects different circuits.
• Provides communication between components.
• Provides power.

System Clock

• Overclocking increases component speeds beyond their specified limits.
• Per component overclocking increases the CPU speed relative to the system clock, by changing the clock multiplication factor.
• The whole system overclocking increases the system clock speed and is applied to all components that use and multiply the system clock.

Clocking Multiplication

• Clocking multiplication involves increasing the system clock to raise a component's frequency through overclocking and increasing the clock multiplier.

Cooling

• Methods include Fans, Liquid Cooling, and Heat Sinks.

Speed vs Throughput

• Speed/bandwidth (theoretical speed) is measured in Megabits per second (Mbps), which indicates the bus's maximum speed.
• Due to latency, throughput/actual data speed, is often reduced.

Front Side Bus (FSB)

• FSB is a parallel bus that connects components to the motherboard.
• FSB connections replaced the point-to-point serial connections in other connections.
• Typically connects the CPU and RAM.
• A data bus transfers actual instructions/data between the CPU and RAM.
• An address bus transfers physical addresses of data between the CPU and RAM.
• Control buses manage commands between the CPU and RAM, essentially sending/reading signals.

External Buses

• PCI Express slots connect graphic cards.
• The number of slots available depends on the motherboard.
• These slots have independent clocks, preventing dependence on the front-side bus for timing.
• SATA is a computer bus interface used to connect mass storage devices like hard drives to a computer's motherboard.
• SATA I speed is 1.5GHz with a throughput of 150MB/s.
• SATA II speed is 3GHz with a throughput of 300MB/s.
• SATA III speed is 3GHz with a throughput of 600MB/s.
• USB connects peripherals to the computer for communication and power.
  • USB 1.0 (Low-Speed): 1.5 Mbps
  • USB 1.1 (Full-Speed): 12 Mbps
  • USB 2.0 (Full-Speed): 480 Mbps
  • USB 3.0 (SuperSpeed): 5 Gbps
  • USB 3.1 (SuperSpeed): 10 Gbps
• NVMe connects SSDs to the motherboard and enables higher speed read/write operations via SATA communication bus.
• The M.2 format is for installing high-speed, low-power devices in small spaces.

Secondary Storage

• Cloud storage is remote storage on an external server for access from any device.
• Size determines the amount of remote storage, where more storage results in higher costs.
• Speed measures the time for accessing a service provider's server over the internet.
• Cost is determined by the amount of cloud storage used, billed annually or monthly.

Other Types of Caching

• Disk Cache is hardware that speeds up disk read/write times and can be kept in general RAM.
• Browser Cache stores website copies accessed on the local hard drive, including downloaded web files.
• Web Cache stores recently accessed web pages on a network server with the most recently downloaded pages.
• The first network request to a webpage accesses the internet and then stores a copy in the proxy server's cache.
• Subsequent access from a user accesses the web page from the browser if conditions are met.

System Software

Start Up and Boot Software

• During computer startup, hardware devices are initialized.
• Passed to RAM after initial checks, which leads to a boot sequence.
• The bootloader performs the boot sequence and loads the operating system.

BIOS (Basic Input/Output System)

• It is firmware on the ROM chip that uses nonvolatile memory, and loads the OS.
• The Power-On-Self-Test (POST) checks the hardware.
• An inventory is obtained from the Complementary Metal-Oxide Semiconductor (CMOS).
• Hardware is tested for proper function, the Master Boot Record (MBR) is located on the hard drive and then loads and runs the OS kernel.

Kernel

• Core of the OS in RAM, the kernel handles input/output requests from software, translating them into instructions for the CPU.
• It also manages memory and peripherals like keyboards and monitors.

CMOS

• Volatile memory that stores hardware and user settings, including hardware installed and the current date and time settings.
• The BIOS accesses it during the boot process to obtain data.

UEFI (Unified Extended Firmware Interface)

• UEFI is used by modern computers for efficient system startups.
• New hardware with UEFI is needed , as it is not interchangable.
• It can boot from drives larger than 2.2TB
• It can run in 32-bit or 64-bit mode.
• It has more address space, resulting in faster boot process and user-friendly features like graphics and cursor support.

Interrupts

• Signals sent to the CPU from hardware or software indicate a need for attention.
• Interrupts can be generated by either hardware or software.

Software Interrupts

• Generated by programs during events like division by zero or I/O requests.
• This can occur when a process has used its allocated CPU cycles.

Hardware Interrupts

• Signals from a device (e.g., keyboard key press, printer out of paper).
• When the CPU receives an interrupt, it pauses the current task to execute the interrupt handler program based on priority.

IRQ (Interrupt Request)

• Each device possesses a unique IRQ for identification by the CPU during interrupts, found in the Device Manager on Windows.

I/O Range

• Act as a buffer, or a temporary storage area in RAM, to accelerate slow I/O devices.
• A complete file may load into a buffer in memory, giving the CPU faster data access.

Processing Techniques

• Designed to optimize computer system efficiency through multitasking and multithreading.

Multitasking

• Allows the OS to run many programs concurrently. The system suspends or interrupts a task when it can't use the processor, freeing the processor for use by another.
• Multiple RAM-stored programs are processed with short bursts of CPU time.
• Os uses process scheduling for fair CPU time distribution.

Multithreading

• It allows a program to run several different threads (parts of a program that run independently) simultaneously such as navigating one website while another downloads in a web browser.
• Hyperthreading from Intel allows processors to run two instruction threads simultaneously through two register sets on the CPU.
• Multithreading is when more than one task in the same program can be executed at the same time.

Virtual Memory

• Enlarges the addressable memory by using swapping pages between RAM and hard disk.
• A slower device is used when the faster RAM runs out of space.
• Is used by most apps use a GUI.

Translator

• A programming tool that converts code into another form, such as translating source code into machine code.
• These are compilers, assemblers or interpreters.
• Most programming languages are text-based.

High-Level Languages

• Abstract the computer's workings, making them user-friendly.
• Python, Delphi, and Java are examples.

Low-Level Languages

• Closely mirror machine language.
• Machine language: binary code of 1s and 0s.
• Assembly code is one machine language instruction but with readable notation.
• Assembly is more readable than machine code.

IDEs (Integrated Development Environments)

• Offer editing features like code completion and error checking during programmer designs.

Interpreters

• Read, convert to machine code, and execute each program line, discarding the machine code afterward, and re-interpreting with each execution.

Compilers

• Transform compiled languages into executable forms before execution.

Machine Code Generation

• Compilers create a machine-specific object code file from source code that can only execute on the compiled machine,. Delphi compiles into .exe files which can only be run on Windows..

Intermediate Representations

• Languages are compiled into bytecode (an intermediate representation) so that bytecode does not need to re-read the source file to execute.
• Bytecode must be interpreted for execution.
• Java uses .class bytecode files that interpreted in the Java Virtual Machine.

Machine Cycle

• Consists of four stages:
  1. FETCH: Retrieve instruction from RAM.
  2. DECODE: Understand the instruction using the CPU's instruction set.
  3. EXECUTE: Perform the instruction, using data if necessary.
  4. STORE: Save result in main memory or an output device.
• The CPU only stores current instructions and the CU sends read/write signals to memory to fetch data.
• The instructions are broken down and stored temporarily in registers.
• The accumulator is a register in the ALU that stores results for RAM storage. The data get loaded from ram into the accumulator if data is loaded.