Computer Concepts

The concept

Almost every computer follows the von Neumann (stored-program) model: the program and its data sit together in main memory, and the CPU fetches instructions one at a time. The machine has five functional units — input, memory, ALU, control unit, output. The CPU is the ALU + control unit + registers, on one chip.

The ALU (Arithmetic Logic Unit) does arithmetic (+, −) and logic (AND, OR, NOT, compare). The Control Unit processes no data at all: it decodes each instruction and sends control signals (memory read, memory write, which ALU operation) to every other unit, in step with the clock. Registers are tiny, fastest storage inside the CPU: PC, IR, MAR, MDR (also called MBR), the accumulator and the flag (status) register.

The CPU repeats the instruction cycle forever: fetch (MAR ← PC; MDR ← M[MAR]; IR ← MDR; PC ← PC + 1), decode (the CU reads the opcode), execute (ALU, memory or I/O action; flags updated). Then it checks for interrupts and starts again. Speed is set by the clock: a 2 GHz clock gives 2 × 10⁹ cycles per second, so one cycle lasts 0.5 ns.

The concept

An instruction is a bit pattern divided into fields: an opcode (which operation) and one or more operand/address fields (where the data is). With n opcode bits the CPU can have at most 2ⁿ different operations. By the number of addresses, instructions are 3-address (ADD R1, R2, R3), 2-address (ADD R1, R2), 1-address (ADD X, the other operand is the accumulator) and 0-address (ADD on a stack, used for postfix evaluation).

The addressing mode tells the CPU how to find the operand; the memory address it finally uses is the effective address (EA). Immediate: the operand is in the instruction (no memory access). Direct: EA = address field. Indirect: EA = M[address field], so one extra memory access. Register: operand is in a register. Register indirect: EA = contents of a register. Indexed: EA = address field + index register (good for arrays). Relative: EA = PC + offset (good for branches and relocatable code). Implied: the operand is fixed by the opcode itself (e.g. CMA, complement the accumulator).

The concept

Input devices turn real-world data into signals the computer understands: keyboard, mouse, scanner, microphone, webcam, joystick, light pen, barcode reader, biometric scanner. Three special readers are favourite exam questions: MICR (Magnetic Ink Character Recognition) reads the magnetic-ink numbers at the bottom of bank cheques; OMR (Optical Mark Reader) detects filled bubbles on answer sheets; OCR (Optical Character Recognition) turns printed text in an image into editable text. A touch screen is both an input and an output device.

Output devices: monitors (CRT, LCD, LED), speakers, projectors, plotters (large engineering drawings) and printers. Impact printers strike a ribbon: dot-matrix (speed in characters per second, cps), line printer (lines per minute, lpm), daisy wheel. Non-impact printers do not hammer the paper: inkjet, laser (pages per minute, ppm), thermal (bill printers). Print quality is measured in dpi (dots per inch).

Backup devices keep a second copy of data: magnetic tape (cheap, huge, sequential — used for archives), external hard disks, optical discs (CD ≈ 700 MB, DVD ≈ 4.7 GB single layer, Blu-ray ≈ 25 GB single layer), flash storage (pen drive, SSD, memory card) and cloud/network storage. Backup types: full (everything), incremental (changes since the last backup of any type — quickest to take, slowest to restore) and differential (changes since the last full backup — restore needs only the full plus the latest differential).

The concept

Memory is arranged as a hierarchy: registers → cache (L1, L2, L3) → main memory (RAM) → secondary storage (SSD, HDD) → backup storage (tape, optical). Moving down, capacity rises and cost per bit falls, but access time grows. The CPU directly accesses only registers, cache and main memory (primary memory); secondary storage is reached through I/O. Cache works because of locality of reference: temporal (a word used now will be used again soon) and spatial (its neighbours will be used soon).

RAM is volatile. SRAM uses flip-flops — fast, no refresh, costly — and is used for cache. DRAM stores each bit as charge on a capacitor — dense and cheap but must be refreshed — and is used for main memory. ROM is non-volatile and holds firmware such as the BIOS/boot program: mask ROM (written at the factory), PROM (programmed once by the user), EPROM (erased with ultraviolet light), EEPROM (erased electrically, byte by byte) and flash (erased electrically in blocks — pen drives, SSDs).

A memory chip is described as words × bits per word. A chip with n address lines has 2ⁿ locations; with m data lines, each location holds m bits. So capacity = 2ⁿ × m bits. CET uses binary units: 1 KB = 2¹⁰ bytes, 1 MB = 2²⁰, 1 GB = 2³⁰, 1 TB = 2⁴⁰; a nibble is 4 bits and a byte is 8 bits.

The concept

In a positional number system with base (radix) r, each digit is multiplied by a power of r that depends on its position. Binary (base 2) uses 0–1, octal (base 8) uses 0–7, decimal (base 10) uses 0–9, and hexadecimal (base 16) uses 0–9 and A–F (A = 10 … F = 15). To convert any base to decimal, multiply each digit by its place value and add: 1011.1₂ = 8 + 0 + 2 + 1 + 0.5 = 11.5.

To convert a decimal integer to base r, divide repeatedly by r and read the remainders from bottom to top. For a decimal fraction, multiply repeatedly by r and read the integer parts from top to bottom. Some fractions never end: 0.1₁₀ = 0.000110011…₂, which is why a computer cannot store 0.1 exactly.

Binary ↔ octal and binary ↔ hex need no division at all, because 8 = 2³ and 16 = 2⁴. Group bits in threes (octal) or fours (hex), starting from the binary point — leftwards for the integer part and rightwards for the fraction — padding with zeros. For octal ↔ hex, go through binary.

The concept

Computers store only bits, so every character gets an agreed number (its code). ASCII is a 7-bit code with 128 characters: 0–31 are control characters, 32 is space, 48–57 are the digits ‘0’–‘9’, 65–90 are A–Z and 97–122 are a–z. Extended ASCII uses 8 bits (256 characters). IBM mainframes used EBCDIC, an 8-bit code. Unicode covers all scripts and is stored as UTF-8 (1 to 4 bytes per character: ASCII characters take 1 byte, Devanagari letters take 3), UTF-16 (2 or 4 bytes) or UTF-32 (always 4 bytes).

Numbers can also be stored digit by digit. BCD (Binary Coded Decimal, 8421 code) writes each decimal digit in 4 bits: 395 = 0011 1001 0101. The patterns 1010 to 1111 are invalid in BCD. Excess-3 adds 3 to each digit before coding (5 → 1000). Gray code changes only one bit between consecutive numbers.

A parity bit is an extra bit chosen so that the total number of 1s is even (even parity) or odd (odd parity). The receiver counts the 1s; a wrong count means an error. Parity detects any odd number of flipped bits (for example, one), but misses two flipped bits, and cannot tell which bit is wrong, so it cannot correct errors (a Hamming code can).

The concept

Binary addition uses 0 + 0 = 0, 0 + 1 = 1, 1 + 1 = 10 (write 0, carry 1) and 1 + 1 + 1 = 11. Signed numbers can be stored in three ways in n bits. Sign-magnitude: the MSB is the sign and the rest is the size. One’s complement: a negative number is the positive number with every bit inverted. Two’s complement: invert every bit and add 1 (equivalently 2ⁿ − x). Sign-magnitude and one’s complement both have two zeros (+0 and −0); two’s complement has only one zero, which is why every modern CPU uses it.

To find A − B in two’s complement, add A to the two’s complement of B and ignore any carry out of the MSB. If the MSB of the answer is 1, the answer is negative; take its two’s complement to read the size. Overflow happens only when two numbers of the same sign are added and the result has the opposite sign — equivalently, when the carry into the MSB differs from the carry out of it. Adding a positive and a negative number can never overflow.

To widen a two’s-complement number to more bits, copy the sign bit into the new places (sign extension): −3 is 1101 in 4 bits and 11111101 in 8 bits.

The concept

Floating point stores a real number as ± M × B^E: a sign, a mantissa M and an exponent E (base B = 2 in computers). A number is normalized when the mantissa has no leading zeros, so no bit is wasted. Two conventions appear in books: older textbooks use 0.1xxx × 2^E (first bit after the point is 1); IEEE 754 uses 1.xxx × 2^E. Read the question to see which one is meant.

IEEE 754 single precision (32 bits, C’s float) has 1 sign bit, 8 exponent bits and 23 fraction bits. The exponent is stored with a bias of 127 (stored E = actual exponent + 127), and the leading 1 is not stored (hidden bit), so value = (−1)^S × 1.F × 2^{E − 127}. Double precision (64 bits) uses 1 + 11 + 52 bits with bias 1023.

Special patterns: E = 0 and F = 0 is ±0; E = 255 and F = 0 is ±∞; E = 255 and F ≠ 0 is NaN (not a number, e.g. 0/0); E = 0 and F ≠ 0 is a denormal (very tiny) number. So normal exponents run from −126 to +127. A number such as 0.1 is a repeating fraction in binary and is stored only approximately; numbers made of halves, quarters and eighths (0.5, 0.75, 0.625) are exact.

The concept

Boolean algebra works on the values 0 and 1 with three basic operations: AND (·), OR (+) and NOT (′ or a bar). Other gates are combinations: NAND = NOT of AND, NOR = NOT of OR, XOR (1 when the inputs differ), XNOR (1 when the inputs are equal). NAND and NOR are universal gates: any circuit can be built using only one of them. A truth table for n inputs has 2ⁿ rows, and there are 2²ⁿ different Boolean functions of n variables.

Key laws: A + 0 = A, A·1 = A, A + 1 = 1, A·0 = 0, A + A = A, A + A′ = 1, A·A′ = 0, (A′)′ = A; distributive A + BC = (A + B)(A + C); absorption A + AB = A, A(A + B) = A, and A + A′B = A + B. De Morgan’s laws: (A + B)′ = A′·B′ and (A·B)′ = A′ + B′ — “break the bar, change the sign”.

In a Venn diagram, AND is the overlap (intersection), OR is the union, NOT is everything outside the circle, and XOR is the two side parts without the overlap. A Karnaugh map places minterms in Gray-code order (00, 01, 11, 10) so neighbouring cells differ in one variable. Group adjacent 1s in blocks of 1, 2, 4 or 8 (groups may wrap around the edges); in each group, the variables that stay constant form one product term.

The concept

At block level a computer is CPU + memory + I/O modules joined by a system bus with three parts. The address bus carries the location the CPU wants (one-way: CPU → memory/I/O); with n address lines the CPU can address 2ⁿ locations. The data bus carries the data itself (two-way); its width is the number of bits moved in one transfer. The control bus carries signals such as memory read/write, I/O read/write, interrupt request, bus request/grant and the clock. Devices connect through interfaces/ports, addressed either as part of memory (memory-mapped I/O) or with separate I/O instructions (isolated I/O).

The processor and I/O can talk in three ways. Programmed I/O: the CPU keeps testing the device’s status bit (busy-waiting), wasting CPU time. Interrupt-driven I/O: the device sends an interrupt when it is ready; the CPU does other work meanwhile but still moves every word itself. DMA (Direct Memory Access): the CPU gives the DMA controller the start address, word count and direction; the controller takes the bus (bus request / bus grant, HOLD/HLDA) and moves the whole block between device and memory, by cycle stealing or in burst mode, then interrupts the CPU once at the end. DMA suits fast bulk devices such as disks.

When an interrupt arrives, the CPU finishes the current instruction, saves the PC and flags on the stack, finds the address of the Interrupt Service Routine (ISR) — often from an interrupt vector table — runs it, and on the return instruction restores the saved state. Hardware interrupts come from devices and are maskable (can be disabled) or non-maskable (NMI, e.g. power failure). Software interrupts / traps are caused by instructions (system calls) or by errors (divide by zero, invalid opcode, page fault — called exceptions). With several devices, priority is set by polling order in software or by daisy chaining in hardware, where the device nearest the CPU has the highest priority.

The concept

Languages are grouped in generations. Machine language (1GL) is pure binary — the only thing a CPU executes; fast but machine-dependent and hard to write. Assembly language (2GL) replaces bit patterns with mnemonics (MOV, ADD, JMP) and labels; each statement usually becomes exactly one machine instruction, so it is still machine-dependent (not portable) but gives full control of the hardware — used in device drivers and embedded code. High-level languages (3GL: C, C++, Java, Python) are close to English and maths, portable, and one statement becomes many machine instructions. 4GLs such as SQL describe what you want, not how.

An assembler converts assembly into machine code (a two-pass assembler builds a symbol table of labels in pass 1). A compiler translates the whole high-level program into object code and lists all errors after scanning it; the result runs fast and needs no compiler at run time (C, C++). An interpreter translates and executes one statement at a time, stops at the first error, produces no object file and runs slower (Python, BASIC; Java bytecode is run by the JVM). A cross-compiler produces code for a different machine.

Building a C program goes: preprocessor (handles #include and #define, removes comments) → compiler (C to assembly) → assembler (to an object file, .o/.obj) → linker (joins object files with library code such as printf, resolves external references, produces the executable) → loader (copies the executable into memory, relocates it and starts it). Syntax errors are caught by the compiler; a function declared but never defined gives a linker error; dividing by zero is a run-time error.

The concept

C’s basic types (assume a typical system): char 1 byte, short 2, int 4, long 4 or 8, float 4, double 8. A signed n-bit integer ranges from −2ⁿ⁻¹ to 2ⁿ⁻¹ − 1 (a 4-byte int is about ±2.1 × 10⁹); unsigned ranges from 0 to 2ⁿ − 1. sizeof is an operator that gives a size in bytes. Integer division drops the fraction (7 / 2 is 3), and % works only on integers, with the result taking the sign of the left operand (-7 % 2 is −1).

Operators from highest to lowest precedence: postfix () [] ++ -- → unary ! ~ ++ -- - (type) sizeof → * / % → + - → shifts << >> → relational < <= > >= → == != → & → ^ → | → && → || → ?: → assignment → comma. Operators of equal precedence group left to right, except unary operators, ?: and assignment, which group right to left. && and || short-circuit: the right side is not evaluated if the left side already decides the answer.

In mixed expressions C converts the smaller type to the larger (implicit conversion: char/short → int → long → float → double); assigning to a smaller type truncates. A cast forces a conversion: (float)7 / 2 is 3.5, but (float)(7 / 2) is 3.0 because the division already happened in int. Common printf formats: %d/%i int, %u unsigned, %ld long, %f float/double (6 decimals by default), %e exponent form, %c char, %s string, %x hex, %o octal, %p pointer, %% a percent sign; %5d sets width 5, %-5d left-justifies, %.2f prints 2 decimals.

The concept

if / else chooses between paths; any non-zero value counts as true. An else always pairs with the nearest unmatched if (the dangling-else rule), whatever the indentation. if (x = 0) is an assignment, not a comparison: it sets x to 0 and the condition is false. switch works only on integer/char values; control jumps to the matching case label and falls through every following case until a break or the end. default runs when no case matches, wherever it is written.

C has three loops. for (init; condition; update) and while (condition) test before each pass, so they may run zero times. do { … } while (condition); tests after each pass, so it always runs at least once. In a for loop, continue jumps to the update part (e.g. i++) and then the test; break leaves only the innermost loop or switch.

For output prediction, never guess: make a small table with one column per variable and one row per pass, and update it line by line. Watch post-increment/decrement inside conditions: in while (n-- > 2) the old value of n is compared and then n is decreased — even on the final, false test. A stray semicolon after for (…); or while (…); gives the loop an empty body.

The concept

An array stores elements of one type in consecutive memory. Indexes start at 0, so int a[5] has a[0] to a[4], and C does no bounds checking. In an expression the array name becomes a pointer to the first element, and a[i] means exactly *(a + i); the address of a[i] = base + i × (size of one element). A string is a char array ending in the null character '\0': strlen counts the characters before it, while sizeof on the array counts the whole array including the '\0'.

A pointer holds an address. &x gives the address of x; *p gives the value stored at the address in p. Adding 1 to a pointer moves it by one element (4 bytes for an int *, 1 byte for a char *). C passes every argument by value: the function receives a copy, so changing the parameter does not change the caller’s variable. To let a function change a variable, pass its address and work through * — this is how C does call by reference (e.g. swap(&a, &b), scanf("%d", &n)). Arrays are passed as a pointer to the first element, so a function can change the caller’s array.

A recursive function calls itself. It needs a base case that stops the calls, and every call must move closer to it. Each call gets its own copy of local variables on the stack; statements written after the recursive call run while the calls return, in reverse order. A static local variable is created once and keeps its value between calls.

The concept

An operating system manages the hardware and provides services to programs: process management, memory management, file management, device (I/O) management, security, and a user interface (shell or GUI). The kernel is its core, and programs ask it for services through system calls. Types: batch, multiprogramming, time-sharing (multitasking), real-time and distributed. A process is a program in execution, described by its PCB (Process Control Block). It moves through the states new → ready → running → waiting (blocked) → ready … → terminated; switching the CPU from one process to another is a context switch.

CPU scheduling picks the next ready process. FCFS: in arrival order, non-preemptive, suffers the convoy effect. SJF: shortest burst first; gives the minimum average waiting time; its preemptive form is SRTF; long jobs may starve. Priority: highest priority first; starvation is cured by aging. Round Robin: each process gets a fixed time quantum, then goes to the back of the queue; designed for time-sharing. For each process, turnaround time = completion − arrival and waiting time = turnaround − burst.

Paging divides a process into fixed-size pages and memory into equal-size frames; a page table maps pages to frames. It removes external fragmentation (some internal fragmentation remains in the last page). A logical address = page number + offset, and the number of pages = 2^{address bits} ÷ page size. Virtual memory keeps only the needed pages in RAM; a missing page causes a page fault; constant faulting is thrashing; replacement policies are FIFO (can show Belady’s anomaly), LRU and Optimal. Deadlock needs all four conditions together — mutual exclusion, hold and wait, no preemption, circular wait; the Banker’s algorithm avoids it. File systems organise files in directories (single-level, two-level, tree) and allocate disk blocks contiguously, as a linked list, or through an index block (FAT, NTFS, ext4).

The concept

The internet is a worldwide network of networks that connects computers using the TCP/IP protocols; it grew out of ARPANET (1969). The World Wide Web (WWW) is only one service running on it — a collection of interlinked pages (hypertext) invented by Tim Berners-Lee at CERN in 1989; email, file transfer (FTP) and video calls are other services. A web page is found by its URL (Uniform Resource Locator), for example https://www.mahacet.org/exams/mca.html: the protocol (https), the domain name (www.mahacet.org, whose last part .org is the top-level domain, like .com, .in, .edu, .gov), and the path and file name (/exams/mca.html). HTTP (HyperText Transfer Protocol, port 80) carries pages between a web server and your browser (Chrome, Firefox, Edge, Safari); HTTPS adds SSL/TLS encryption (port 443) and shows the padlock. Pages are written in HTML (HyperText Markup Language): a markup language of tags such as <p> and <a href="…"> that describe structure and links, not a programming language. Because HTML is an open standard understood by every browser on every operating system, the same document can be displayed anywhere — its main advantage. Email (electronic mail) delivers messages to an address of the form username@domain, e.g. sunil.patil@gmail.com: the user name (mailbox) before the @ and the domain name of the mail provider after it. SMTP (Simple Mail Transfer Protocol) sends mail; POP3 (Post Office Protocol) downloads it to one device; IMAP (Internet Message Access Protocol) keeps it on the server so phone and laptop stay in sync. CC (carbon copy) recipients are visible to everyone; BCC (blind carbon copy) recipients are hidden; an attachment is a file sent along with the message; unwanted bulk mail is spam.

Software is of three kinds. System software runs the machine itself: the operating system (Windows, Linux, macOS, Android), device drivers and language translators (compiler, interpreter, assembler). Application software does the user’s work: word processor (MS Word), spreadsheet (Excel), presentation (PowerPoint), browser, WhatsApp, Tally. Utility software maintains the system: antivirus, disk clean-up, file compression (WinRAR, 7-Zip), backup tools. Software burnt into ROM (the BIOS) is firmware; open-source software (Linux, Firefox) ships its source code, proprietary software (Windows, MS Office) does not. A file’s extension tells its type: .docx Word, .xlsx Excel, .pptx PowerPoint, .pdf portable document, .txt plain text, .jpg/.png/.gif images, .mp3 audio, .mp4 video, .zip compressed archive, .exe Windows program, .apk Android app, .html web page, .c C source. Networks are named by size: PAN (personal, Bluetooth), LAN (Local Area Network — one building or campus, such as a college lab), MAN (a city) and WAN (Wide Area Network — across cities and countries; the internet is the largest WAN). Every device on a network has an IP address: IPv4 is 32 bits written as four numbers from 0 to 255 separated by dots (192.168.1.1); IPv6 is 128 bits. DNS (Domain Name System) translates names such as www.google.com into IP addresses — the phone book of the internet. A modem (modulator–demodulator) converts the computer’s digital signals to and from the analog signals of a telephone or cable line; a router forwards data packets between networks and shares one connection among many devices; a switch or hub joins devices inside a LAN; Wi-Fi (IEEE 802.11) is a wireless LAN; an ISP (Internet Service Provider — Jio, Airtel, BSNL) connects you to the internet. Bandwidth is the maximum rate at which a link carries data, measured in bits per second (kbps, Mbps, Gbps).

Security: malware is any harmful program. A virus attaches itself to a file and spreads when that file is run; a worm copies itself across a network on its own; a Trojan horse hides inside software that looks useful; spyware steals information and ransomware locks your files for money. Antivirus software (Quick Heal from Pune, Norton, McAfee, Kaspersky) detects and removes malware and must be updated regularly. A firewall (hardware or software) filters the traffic between a trusted network and the internet according to rules. Phishing is a fake email, SMS or website pretending to be a bank or a known company to “fish” for passwords, OTPs and card numbers; a bank never asks for these. A strong password is long (at least 8–12 characters), mixes upper- and lower-case letters, digits and symbols, is not a name or a date of birth, and is backed by two-factor authentication (OTP). Encryption scrambles data so that only the holder of the key can read it. Types of computer by size and power, smallest first: microcomputer (one microprocessor — desktop, laptop/notebook, tablet, smartphone), minicomputer (mid-range, many users in a department), mainframe (thousands of users at once — banks, railway reservation) and supercomputer (the fastest, speed measured in FLOPS — weather forecasting, research; India’s PARAM 8000 was built by C-DAC Pune in 1991). By the signals they use, computers are analog, digital or hybrid. Generations are named by their switching technology: 1st (about 1940–56) vacuum tubes, machine language, punched cards (ENIAC, UNIVAC); 2nd (1956–63) transistors, assembly language and early high-level languages (FORTRAN, COBOL); 3rd (1964–71) integrated circuits, keyboards, monitors and operating systems (IBM 360); 4th (1971 onward) microprocessors with VLSI (Intel 4004), personal computers, GUI and networks; 5th (present and beyond) ULSI, parallel processing and artificial intelligence.