Programs, machine code and why we translate; compiler vs interpreter

0 · Recall Warm-up

Answer from memory first. Answers are in the Answer Vault at the end of this lesson.

• When you type a message on your phone, does the phone’s chip understand the English words? from everyday life
• A cook gets a recipe written in a language she cannot read. Name two ways she can still cook it. from everyday life
• A tourist carries a phrasebook prepared before the trip; another tourist hires a guide who translates on the spot. Who talks faster during the trip? from everyday life

1 · The Story

📖

Mr. Patel books Translation House

Mr. Patel runs a textile business in Ahmedabad. Next week he must give the same speech every evening, for seven evenings, to an audience that understands only English. His speech is written in Gujarati.

He walks into Translation House. The director, Mrs. Desai, offers him two services.

Service 1 — the live interpreter. Raj stands next to Mr. Patel every evening. Mr. Patel says a sentence in Gujarati, Raj says it in English, then the next sentence. If Mr. Patel changes a line on Thursday, Raj adapts instantly. But Raj must work every single evening, and every evening each sentence is translated all over again. If a sentence in the speech is garbled, nobody finds out until Raj reaches it on stage.

Service 2 — the booklet. A translation team reads the whole speech once, before the first evening, and prints a complete English booklet. Garbled sentences are found and reported while the team works, before anybody is on stage. Every evening the audience simply reads the booklet, which is fast. But if Mr. Patel changes a line, a new booklet must be printed.

Mr. Patel asks: “Which is better?” Mrs. Desai smiles: “Seven evenings of the same speech? Print the booklet. A one-off chat with a buyer who keeps changing his mind? Hire Raj.”

2 · Predict First

🤔

A program will be run 1000 times on different inputs and is never changed. Which service from the story should a computer use for it: the live interpreter or the printed booklet? Write your guess down. The answer is in the vault.

3 · Core Concept

The problem: people write programs, processors run machine code

A program is a list of instructions written by a person in a programming language such as C:

total = price * qty + 50;

This is a high-level language: close to mathematics and English, and one line can mean many small machine steps.

The processor (CPU) does not understand this line. It executes only machine code: instructions encoded as patterns of bits (0s and 1s). Each machine instruction does one tiny step, such as “load a number from memory”, “multiply two numbers” or “store a number in memory”.

Machine code belongs to one processor family

An Intel/AMD (x86) processor and a phone’s ARM processor use different machine codes. The same C line becomes different bit patterns for each. The source is portable; the machine code is not.

So we need a translator

A translator is a program that reads a program written in one language and either produces an equivalent program in another language or carries out its actions directly.

The program we start from is the source program. A translated program that comes out is the target program.

Compiler

A compiler reads the whole source program and produces an equivalent target program, usually machine code, before anything runs. It also reports mistakes it finds in the source. The time when the compiler works is compile time.

Later, the target program (an executable) runs on its inputs as often as you like, with no compiler present. The time when the program runs is run time.

Formal definition. A compiler is a program that maps a source program in language S to an equivalent target program in language T, and reports errors in the source program.

Interpreter

An interpreter produces no target program. It takes the source program and the input together and carries out the statements one at a time, translating as it goes. A mistake in line 40 is found only when execution reaches line 40, after lines 1–39 have already run.

Comparing the two

Hybrid. Java is compiled first into bytecode (code for an imaginary machine), and then a virtual machine interprets that bytecode, often compiling the busy parts to machine code while the program runs (just-in-time compilation).

Story → compiler term

4 · Visual

Compiler route, step 1 (compile time): translate the whole program once.

Compiler route, step 2 (run time): run the target program as often as needed, no compiler needed.

Interpreter route: every run translates again; no target program is ever produced.

5 · Worked Examples

Example 1 Easy

Classify each tool: (a) gcc, which turns a C file into an executable; (b) the python program running script.py; (c) a tool that turns C++ code into equivalent C code.

Solution

(a) Compiler: it produces a separate target program (machine code) before running.

(b) Interpreter (strictly a hybrid: it first converts the script to bytecode, then interprets the bytecode, and produces no separate executable you keep).

(c) A compiler too. Nothing in the definition says the target must be machine code. This kind is called a source-to-source compiler. In the story, it is like translating the Gujarati speech into a Hindi speech text: still a booklet made in advance.

Example 2 Medium

A program will run 1000 times. Compiling takes 2 s and each compiled run takes 0.01 s. Interpreting takes 0.5 s per run. Compare the total times. From how many runs onwards is compiling cheaper?

Solution

Compiler route: 2 + 1000 × 0.01 = 12 s. Interpreter route: 1000 × 0.5 = 500 s.

Break-even: compiling is cheaper when 2 + 0.01n < 0.5n, that is n > 2 / 0.49 ≈ 4.08. So from 5 runs onwards the booklet wins. This is Mrs. Desai’s advice to Mr. Patel in numbers: seven evenings of the same speech favour the booklet.

Example 3 GATE

Which statements are true? (i) An interpreter produces a separate target program. (ii) A compiler translates the whole program before execution. (iii) Interpreted programs usually run slower than compiled ones. (iv) A compiled program needs the compiler present every time it runs.

Solution

(i) False: an interpreter produces no target program; it executes directly. (ii) True. (iii) True: the interpreter repeats translation work during every run. (iv) False: once the target program exists, the compiler is not needed, just as the audience reads the booklet without the translation team in the hall.

Answer: (ii) and (iii).

6 · Shortcuts

⚡

One question decides it: does the tool produce a separate program that can run later without the tool? Yes → compiler. No → interpreter. Valid when: classifying a translator in an MCQ. Hybrids such as Java do both: compile to bytecode, then interpret.

⚡

Break-even runs: with compile cost C, compiled run time r and interpreted run time i, compiling is cheaper for n > C / (i − r) runs. Valid when: every run costs the same and i > r.

7 · GATE Traps

⚠️

• A compiler’s target need not be machine code: it can be another high-level language, bytecode or assembly-like text.
• An interpreter also detects errors, but only when execution reaches the faulty statement; earlier statements may already have run.
• Machine code is tied to a processor family. The source program is portable; the executable is not.
• Java is neither purely compiled nor purely interpreted: a compiler produces bytecode, then a virtual machine interprets (or just-in-time compiles) it.

8 · Memory Hook

🧠

Compiler = a booklet printed once. Interpreter = a live translator every evening.

9 · Connect the Dots

This lesson answered why translation is needed and the two ways to do it. Next question: when you compile a C program, the compiler is not the only tool involved. Which other tools touch the program before it runs, and in what order?

10 · Self-Test

GATE format. Try every question before opening the vault.

Q1MCQ1 markCompiler

Which statement best describes a compiler?

• (A) It executes the source program line by line together with the input.
• (B) It translates the whole source program into an equivalent target program before execution.
• (C) It converts machine code back into the original source program.
• (D) It only checks the spelling of the source program.

Q2MSQ1 markInterpreter

Which of the following are true of an interpreter?

• (A) It does not produce a separate target program.
• (B) A mistake in a statement is found only when execution reaches that statement.
• (C) Programs generally run faster under an interpreter than as compiled code.
• (D) The source program is needed every time the program runs.

Q3NAT2 marksBreak-even

Compiling a program takes 3 s; each compiled run takes 0.02 s. Interpreting takes 0.32 s per run. What is the smallest number of runs for which compiling (including its 3 s) takes strictly less total time than interpreting?

Numerical answer: ________

Q4MCQ1 markMachine code

Machine code produced for an x86 processor:

• (A) runs unchanged on any processor
• (B) is specific to the x86 family of processors
• (C) is the same text as the C source program
• (D) must be run by an interpreter such as Python

Q5MCQ1 markClassification

A tool reads a C++ program and writes an equivalent C program. This tool is:

• (A) an interpreter
• (B) a compiler
• (C) a processor
• (D) a text editor

11 · Rule Card

• Compiler: whole source → equivalent target program, before running; reports source errors at compile time.
• Interpreter: source + input → results directly, statement by statement; no target program.
• Compiled code is usually faster; interpreted code is easier to change and test.
• Break-even: compile when the number of runs n > C / (i − r).
• Target can be machine code, assembly-like text, bytecode or another high-level language.

12 · Concept Ledger

programhigh-level languagemachine codeprocessortranslatorsource programtarget programcompilerinterpretercompile timerun timeexecutablebytecodevirtual machine

13 · Answer Vault