L1 Terminology PDF

1 L1 Terminology 1. Introduction 2. Computer structure 2.1. Processor 2.2. Graphics Processing Unit 2.3. Memory 3. Conclusion Introduction 2 What is Terminology? Terminology is the science that studies terms used in a specific field. It focuses o...

1 L1 Terminology 1. Introduction 2. Computer structure 2.1. Processor 2.2. Graphics Processing Unit 2.3. Memory 3. Conclusion Introduction 2 What is Terminology? Terminology is the science that studies terms used in a specific field. It focuses on the vocabulary used in scientific or technical contexts. The word "terminology" first appeared in the 19th century, as noted by the French linguist Alain Rey. Introduction Definition of Computer Science 3 Computer science is the science of automatic information processing using a machine capable of manipulating data in digital or binary form. This is basically the concept of the computer. The term "Informatique" was created in 1962 by Philippe Dreyfus, combining the words "information" and "automatic." Introduction 4 Computer A computer is an electronic device capable of processing information by executing logical and arithmetic operations based on programs that read a sequence of instructions. Introduction 5 Generations of computers 1. The First Generation (1945 - 1954): This corresponds to the first implementations of the Von Neumann architecture based on vacuum tubes. Electronic machines composed of vacuum tube circuits; Programming in binary language; Significant role; Difficulty of programming; Lack of flexibility; Processing time (10^-2 seconds). Vacuum lamps  Introduction 6 The First Generation (1945 - 1954): ENIAC (1946): the world's first fully electronic computer ENIAC (Electrical Numerical Integrator and Calculator) Introduction 7 Generations of computers 2. The Second Generation (1955 - 1963): Discovery of transistors, which replaced vacuum tube circuits; Programming in binary language; Development of advanced programming languages (such as Fortran in 1954); Space-saving; Introduction of printers and memory; Processing time (10^-3 seconds). Transistors Introduction 8 Generations of computers 3. The Third Generation (1964 - 1971): Corresponds to computers using integrated circuits; Integrated circuits, an entire electronic circuit on a small piece of silicon; Computers become affordable for small companies for the first time; Beginning of the software industry; Processing time (10^-6 seconds). Circuits intégrés  Introduction 9 Generations of computers The Fourth Generation (1972 - Present): Corresponds to the use of microprocessors, which allow for the emergence of microcomputers; Large-scale integration of integrated circuits (VLSI, LSI); Reproduction of a true micro-machine on a single chip: the microprocessor; Reduction in the space occupied by computers; Development of personal computers; Networking; Processing time (10^-9 seconds). Circuit VLSI  10 Basic computer structure The basic structure of a computer consists of several key components, each serving a specific function to make the system work: Input unit, Central unit, Output unit, Storage unit CPU Input unit Central unit Output unit The Processor (CPU) Graphics Processing Unit (GPU) Random Access Memory (RAM) Storage unit Hard Drive (HDD) or Solid-State Drive (SSD)  11 Central Unit The Central Unit contains the main components that enable the computer to function. These components are connected to the motherboard, which links all the key parts of the system. Among these components, we find: The Processor (CPU) Graphics Processing Unit (GPU) Random Access Memory (RAM) Hard Drive (HDD) or Solid-State Drive (SSD)  Central Unit 12 The central unit is connected to external peripherals to enable interaction with the user and the outside world: Input Devices: Such as the mouse, keyboard, or scanner, which allow the user to give instructions to the computer. Output Devices: Such as the monitor, printer, or speakers, which display or return the results of the computer’s actions. CPU 13 Processor : CPU (Central Processing Unit) A CPU is the primary hardware component in a computer responsible for executing instructions from programs and performing essential tasks like calculations, logic operations, and data management. It acts as the "brain" of the computer, interpreting and processing data to ensure the system runs properly. Its power is expressed in Hz.  CPU 14 Processor : CPU (Central Processing Unit) CPU Calculation Execution UAL stands for Arithmetic Logic Unit. It is a crucial component of the processor that performs all the arithmetic (mathematical) and logical operations. How it Works ? The UAL receives data from the CPU’s registers or memory. Based on the instructions given by the CPU (via the control unit), the UAL performs the specified operation. After processing, it returns the result to the CPU’s registers or memory.  CPU 15 Processor : CPU (Central Processing Unit) Data management The CPU's role can be broken down into key tasks: Plan and allocate resources Provide instructions to other components Supervise operations Data transfers => Using DMA, to avoid “wasting” calculation resources. For data transfers, the CPU often uses specialized modules like DMA (Direct Memory Access) instead of the CPU’s UAL. To avoid "wasting" computational resources. The UAL is reserved for calculations, while DMA handles data transfers. The CPU's role is to trigger or authorize the transfer, allowing DMA to perform the task.  CPU 16 Processor : Clock Speed (Frequency) and its Relation to Power Clock speed is the rate at which a CPU executes instructions, measured in Hertz (Hz). 1 Hz means 1 cycle per second, so if a CPU has a clock speed of 1 GHz (gigahertz), it means it can perform 1 billion cycles per second. Higher Clock Speed generally means the CPU can execute more instructions per second, which typically leads to better performance. However, clock speed alone is not the sole determinant of CPU power or overall system performance. Other factors like the number of cores, architecture, and cache size also play significant roles. CPU 17 Number of cores, architecture, and cache size.  CPU 18 Main Types of DMA Controllers Integrated DMA in Chipsets: Found in most modern motherboards (e.g., Intel, AMD chipsets). Example: Transfers between RAM and storage without using CPU cycles. PCIe Card DMA Controllers: Built into high-speed PCIe devices (like graphics cards and SSDs). Examples: Graphics cards transfer video data; SSDs perform quick storage reads/writes. CPU 19 DMA operating modes Single-Transfer Mode This mode transfers one data unit (like a byte or word) at a time. Demand Mode The DMA controller transfers data only when the connected device requests it. Scatter-Gather Mode The DMA controller collects data from multiple scattered locations in memory and transfers it to one destination, or takes a block of data and distributes it to different locations. CPU 20 Example (without DMA) CPU 21 Example (with DMA)  CPU 22 Processor Examples System on Chip (SoC) : Microcontroller : 1.Qualcomm Snapdragon 1.ARM Cortex-M Series 2.Apple A-Series Chips (A14, A15, etc.) 2.ATmega328P (Arduino) 3.Samsung Exynos 3.PIC Microcontrollers (by Microchip) 4.NVIDIA Tegra 4.ESP32 Processors for Embedded and Automotive Systems: 1.NXP i.MX Series 2.Renesas R-Car 3.Texas Instruments Sitara 4.Infineon AURIX 5.STMicroelectronics STM32 6.Qualcomm Snapdragon Automotive 7.Intel Atom (Automotive)  23 Basic computer structure The basic structure of a computer consists of several key components, each serving a specific function to make the system work: Input unit, Central unit, Output unit, Storage unit GPU Input unit Central unit Output unit The Processor (CPU) Graphics Processing Unit (GPU) Random Access Memory (RAM) Storage unit Hard Drive (HDD) or Solid-State Drive (SSD)  GPU The Graphics Card (GPU) 24 GPU (Graphics Processing Unit) a processor specialized in graphics calculations and massively parallel tasks. It is a component of the central unit in charge of managing the display on the screen, including Windows, desktop environments, and application windows. High-performance graphics cards also handle 3D rendering, such as in video games. The leading manufacturer of graphics cards is NVIDIA. Compared to CPU: CPU: General-purpose tasks, sequential processing. GPU: Massively parallel tasks, ideal for intensive computations. GPU 25 Applications of GPUs The GPU plays an essential role in many areas due to its ability to efficiently process massively parallel tasks. Video Games: Rendering realistic images, lighting effects, detailed textures. Scientific Computing: Complex simulations (climate, physics, biology). Artificial Intelligence: Accelerating model training (machine learning, deep learning).  GPU 26 GPU Architecture GPU is designed for Parallel Processing, thousands of lightweight cores optimized to execute many small tasks simultaneously (massive parallelism). Comparison to CPU CPU Cores: Fewer (4–16 cores typically). More powerful for sequential tasks. GPU Cores: Less powerful individually. Compensate with sheer numbers (thousands of cores).  GPU 27 Example Processing CPU/GPU Processing of one pixel =2 clock cycle Image with 36 Pixel CPU: (1 cores) Time execution = 36 pixel x 2 clock cycle Parallel processing Total execution time = 72 clock cycle GPU: (36 cores) Execution time(core 1) =1 pixel x 2 clock cycle Execution time(core 2) =1 pixel x 2 clock cycle Execution time (core 3) =1 pixel x 2 clock cycle..... Execution time (core 36) =1 pixel x 2 clock cycle Total execution time =2 clock cycle Sequential processing  GPU 28 GPU Architecture 1. Arithmetic Logic Unit (UAL) UAL (Arithmetic Logic Units) A core component of a GPU designed to perform arithmetic and logical operations. Responsible for executing mathematical calculations such as addition, subtraction, multiplication, and logical operations (AND, OR, NOT). ALUs contribute significantly to the parallelism of the GPU, allowing it to execute thousands of operations at once, which enhances the overall performance for graphics and scientific computations. ALU in Video Games: Object positions, Rotations, Shadows…  GPU 29 GPU Architecture 2. Shared Memory Shared Memory is a fast, temporary memory space in the GPU that allows multiple cores to communicate with each other quickly. Why It's Important? GPU GPU Speed: The closer the memory is to the Core 1 Core 2 core, the faster it can be accessed, reducing bottlenecks in data transfer. Optimization: By storing frequently used data in shared memory, the GPU can Shared Memory execute parallel tasks more efficiently, speeding up complex computations.  GPU 30 GPU Architecture 3. VRAM (Video Random Access Memory) VRAM is a specialized type of memory used by the GPU to store and manage data needed for rendering images, videos, and 3D graphics. It holds textures, frame buffers, and other graphical assets that the GPU needs to access quickly during processing.  GPU 31 Graphics Pipeline The Graphics Pipeline refers to the series of steps that a computer's GPU (Graphics Processing Unit) takes to render 3D images and display them on the screen. It breaks down the complex task of rendering graphics into a sequence of stages that are processed in order, each focusing on a specific aspect of image creation. Vertex Processing Pixel Mapping (Rasterization) Pixel Shading  GPU 32 Graphics Pipeline 1. Vertex Processing Vertex processing is the first step in the graphics rendering pipeline. It involves the transformation of 3D coordinates (vertices) of an object into 2D coordinates that can be displayed on the screen.  GPU 33 Graphics Pipeline 2. Pixel Mapping (Rasterization) Rasterization is the process of converting 3D models (or geometries) into 2D images by determining which pixels on the screen correspond to the shapes defined by vertices in the 3D space.  GPU Graphics Pipeline 34 3.Pixel Shading A Shader is a small program executed on the GPU to generate visual effects. Shaders control the appearance of objects in a game or simulation, such as color, texture, lighting, and shadows. There are different types of shaders, two common shading techniques are: Flat Shading: Simplifies lighting calculations by Flat Shading assigning a single color to each polygon, creating a faceted, angular look. Phong Shading: Provides smooth shading by interpolating vertex normals, resulting in more realistic lighting and smooth surfaces. Shaders are essential for creating visually impressive and Phong Shading dynamic scenes.  GPU CUDA (Compute Unified Device Architecture) 35 CUDA is a parallel computing platform and programming model developed by NVIDIA. It allows developers to harness the power of NVIDIA GPUs for tasks beyond traditional graphics processing, enabling them to perform complex calculations much faster. It Breaks down tasks into smaller sub-tasks that can be processed simultaneously by the thousands of cores in the GPU. Greatly accelerates performance for applications like artificial intelligence (AI), image processing, and scientific simulations. Developers can write code using languages like C, C++, and Fortran, and take advantage of NVIDIA’s libraries for deep learning and high-performance computing.  GPU GPU Overclocking 36 Overclocking is the process of increasing the operating frequency of the GPU to boost its performance. While this results in higher performance, it also generates more heat and can lead to increased power consumption. Use cases: Gamers and graphics professionals often use overclocking to get better performance in specific tasks, such as gaming or rendering. Risks: If not properly managed, overclocking can cause hardware damage due to excessive heat and power draw. Overclocking is a powerful tool but requires careful monitoring to avoid damaging the system.  37 GPU vs CPU Comparison between GPU and CPU let's compare the CPU and GPU directly in terms of : Performance Cost Development time Energy consumption. Performance __________________________ Performance ____________________________ 38 CPU: 4 to 16 cores, high frequency (3-5 GHz), ideal for complex, sequential tasks. GPU: Thousands of cores (e.g., NVIDIA RTX 4090 with 16,000+ cores), lower frequency (1-2 GHz), optimized for parallel tasks. ________________________ Development Time ________________________ CPU: Easy to program with common languages (Python, C++, etc.), quick development for sequential tasks. GPU: Requires specialized frameworks (CUDA, OpenCL) and expertise for parallel programming, leading to longer development time. ______________________________ Cost ______________________________ CPU: Typically less expensive. GPU: Generally more costly due to their specialized architecture. ______________________ Energy Consumption_________________________ CPU: Consumes less energy, better for energy-efficient systems. GPU: High power consumption, especially for intensive tasks like 3D rendering or AI model training (up to 450W), resulting in significant heat generation.  39 GPU CPU Ideal for parallel Best for sequential computing tasks or general tasks, such as graphics offering advantages rendering, in terms of cost, simulations, and development time, artificial intelligence, and energy thanks to its large efficiency. number of cores.  40 Basic computer structure The basic structure of a computer consists of several key components, each serving a specific function to make the system work: Input unit, Central unit, Output unit, Storage unit Memory Input unit Central unit Output unit The Processor (CPU) Graphics Processing Unit (GPU) Random Access Memory (RAM) Storage unit Hard Drive (HDD) or Solid-State Drive (SSD) Memory 41 Memory Memory is a crucial component in all computer systems, responsible for storing and accessing data. Two main categories of memory: Volatile Memory (e.g., RAM): Loses its content when the computer is turned off. Non-Volatile Memory (e.g., Hard Drives or SSDs): Retains data even when the computer is powered off.  Memory 42 Memory roles Memory acts as a bridge or interface between the processor and data. What is an interface? An interface is a mechanism that enables communication between two components. It can include data buses or communication protocols (e.g., PCIe). Interface CPU (high frequency) Hard disk (low frequency)  Memory 43 Memory roles CPU has low storage capacity, so we have to use memory. What does memory do? Provides data to the processor quickly: Memory ensures that the CPU has fast access to the data it needs for processing. Stores the results of calculations: After executing instructions, memory stores the results temporarily or permanently for future use.  Memory 44 Characteristics of Memory To differentiate between types of memory, we consider several key characteristics: 1.Capacity Measured in megabytes (MB), gigabytes (GB), or terabytes (TB). Determines the amount of data a memory can store. The higher the capacity, the better the memory can meet modern needs.  Memory 45 Characteristics of Memory 2. Access Speed: Measured in nanoseconds or Performance frequencies (MHz, GHz). 3. Volatility Volatile memory (e.g., RAM): loses data when powered off Non-volatile memory (e.g., SSD) retains data even without power. 4. Cost: The cost varies depending on the capacity, speed, and volatility of the memory. Higher performance typically results in higher costs. Memory 46 Types of Memory 1.Primary Memory: Includes RAM and Cache Memory. They store data that is actively being used by the processor for quick access. 2.Secondary Memory: Includes SSDs and Hard Drives (HDDs), These memories are used for permanent data storage. Even when the computer is powered off, the data remains preserved. Memory 47 Random Access Memory (RAM) RAM is a type of memory found in every computer, used to store temporary information. Its main advantage is its very fast read speed compared to the hard drive, allowing for smooth computer performance. RAM stands for Random Access Memory, meaning its purpose is not to store information permanently, but to access it quickly and temporarily.  Memory 48 RAM Features There are several generations of RAM. Each generation of DDR (e.g., DDR3, DDR4, DDR5) introduces improvements in speed, capacity, and energy efficiency: DDR3: Suited for older systems, Frequency up to 2.1 GHz. DDR4: Faster and more energy-efficient. Frequency up to 3.2 GHz. Consumes less power relative to frequency. DDR5:The latest generation with even higher speeds. Frequency over 4.8 GHz. Consumes less energy. Storage Capacity: RAM size ranges from several gigabytes to terabytes in high-end systems (e.g., servers). Memory 49 Cache Memory Cache Memory A ultra-fast memory located directly in the processor (or nearby). It acts as an intermediary between the CPU and main memory (RAM). Why is it important? Designed to be much faster than RAM. When placed between the CPU and RAM, it boosts overall performance significantly. Memory 50 HDD (Hard Disk Drive) The hard drive is a key component of a computer, responsible for storing data. It holds the Windows operating system, installed programs, and personal files. Traditional hard drives (HDD) were the most common storage solution. They offer large storage capacities (e.g., 1TB or 2TB) at a low cost. However, HDDs rely on mechanical components (rotating platters and read/write heads), making them slower and more prone to damage from physical shocks.  Memory 51 SSD (Solid State Drive) An SSD uses chips to store data, eliminating the need for mechanical parts like those found in traditional hard drives. This results in significantly faster data access speeds, quieter operation, and greater resistance to physical shocks. While SSDs come at a higher cost compared to HDDs, they provide much better performance, making them an ideal choice for those seeking faster and more durable storage solutions. Memory 52 ROM (Read-Only Memory) ROM, or Read-Only Memory, is a type of non-volatile memory, meaning it retains its content even without power, ensuring no data loss during power outages. It is read-only, meaning once data is stored, it cannot easily be modified or erased, ensuring the stability and security of the stored information. ROM is ideal for storing critical information such as firmware, BIOS (Basic Input/Output System), and boot programs.