5.1 Computer Systems Organization

The main components of a computer system include the processor, the main memory, the disk, and I/O devices.
The components interact to execute computer programs efficiently.
Applications are written in high-level languages that use a program to translate the language into the machine-level programs that computers understand.
This computer system organization is mostly the same in your tablet or smartphone as it is in the huge systems running services like Facebook or Google. They just differ as it comes to how powerful each piece is, how many CPUs are there, the size of the memory and disk, and so on.

5.2 Computer Levels of Abstraction

The levels from problem definition to assembly language are related to the computer science field. People studying computer science explore how to solve a problem using an algorithm, how to translate this algorithm into a programming language, then how to translate this programming language into a language that computers can understand.
Computer processors operate at various levels of abstraction going from the digital logic level up to the microarchitecture level and the machine language level. The highest level of abstraction for application programs is that obtained by writing programs in a high-level language. Using a compiler makes it possible to generate a representation of such programs at a lower level of abstraction known as assembly language. That representation uses instructions that are part of the instruction set architectures (ISA) specific to the processor family in use.
The operating system (OS) is the only piece of software that can directly access the hardware of a computer. All other programs must interface with the OS to have it achieve a specific task in a secure fashion by scheduling the corresponding process. The OS stores data and programs on disk in an organized way, using a file system, and allocates memory to programs.
Upcoming computer designs call for new computing abstractions that will deviate from the traditional binary logic. For example, quantum computing uses qubits and neuromorphic computing uses hardware neurons.

The most frequently used data items are stored inside a computer.
An integer is a number that does not have a floating point; 7 is an integer but 3.14 is not. Integers are divided into two categories: signed and unsigned. Signed integers can be positive, negative, or 0. Unsigned integers are 0 or positive; therefore, they do not need a sign because they will never be negative. This is why we call them unsigned.
Real numbers, known as floating point numbers, are the numbers that represent fractions. Integers cannot represent numbers like 3.14 or –1.25. This is the role of floating point numbers.
Characters and symbols on your keyboard are represented inside the computer as 1s and 0s. Every character has its own code.
The range of numbers that n-bit binary number can present if interpreted as a signed or an unsigned integer is helpful to you when you write a program because if you know the values that a variable in your program may take, you can make a precise decision about the type of that variable when you declare it.

An x86-64 Intel processor uses a complex instruction set computer architecture (CISC). This type of architecture combines many simple instructions into a simple complex one. Other processor architectures use a reduced instruction set architecture (RISC) based on simpler instructions. RISC V is a relatively new RISC ISA that is getting popular than CISC. All new high-tech companies use RISC instruction sets in their hardware. For example, ARM assembly, which is a RISC assembly, is used in most portable devices.
Assembly language makes use of processor instructions that are part of the instruction set architecture (ISA) of the processor being used. These instructions can be converted to binary code, referred to as machine language code, using an assembler.
Assembly language provides specific instruction to perform common operations such as addition and multiplication of signed and unsigned integers.

We must use different technologies to build a near ideal memory because each technology has some good characteristics and a few shortcomings. Therefore, we use multiple technologies to get the best of all. Various types of memory, such as DRAM and SRAM, are available and differ in access speed and associated costs.
Those technologies are organized as a hierarchy. Technologies that are fast but expensive are at the top of the hierarchy, but their storage capacity is not big. As we go down the hierarchy, the technology is slower, but cheaper; hence, we use a lot of storage from it.
That is, characteristics of each technology determine its place in the hierarchy.
By paying attention to locality in accessing the data, the programmer can get the best performance from the hierarchy.

Processors evolved from a single cycle implementation to multicore.
The norm now is to have a multicore processor working in tandem with one or more accelerator chips.
The norm in software development is to write parallel code for such a heterogeneous system. Sequential programming will soon be dated.