MIPS Assembly Language Elements Concepts

Machine code

The binary instructions a CPU actually executes. Each CPU architecture has its own machine code.

Why can't a CPU run C++ directly?

C++ is a universal, high-level language. It has to be compiled into the CPU's own machine code (binary) first.

Assembly language

Text mnemonics that stand for binary machine code, so humans can read and write it.

Mnemonic

A short text name for an instruction or operand, e.g. lw, add, t0.

Assembler

The tool that translates assembly language into machine code.

Assembling vs. disassembling

Assembling: assembly text → machine code (binary). Disassembling: machine code → assembly text.

4 things assembly language defines

1) Operation mnemonics (instructions) 2) Operand mnemonics (data) 3) Number/type of operands 4) Operand ordering (source/destination)

RISC

Reduced Instruction Set Computer. Fewer instructions, fixed instruction size. MIPS is RISC.

CISC

Complex Instruction Set Computer. Many instructions of varying size. Example: Intel architecture.

MIPS instruction size

Fixed: every instruction is exactly 32 bits.

Load/store architecture

Data must be loaded into registers before it can be processed and stored back to memory afterward. MIPS cannot operate directly on data in memory.

Load vs. Store

Load = memory (or a constant/address) → register. Store = register → memory.

Pipelined architecture

The CPU works on several instructions at once, each in a different stage. In MIPS this causes the delayed branch.

Delayed branch / branch delay slot

The instruction right after a branch or jump always executes, even when the branch is taken.

nop

'No operation.' An instruction that does nothing. A safe filler for a branch delay slot.

How many general-purpose registers does MIPS have?

32

Register zero ($0)

Always holds the value 0 and cannot be changed.

Register t0

Temporary register 0. It is register number 8, NOT register 0.

Register at

Assembler temporary. The assembler uses it when expanding macro instructions (e.g. lui at, 0x1234).

Register names in MPLAB

Leave off the $. Write t0, not $t0.

MPLAB X IDE

Microchip's development environment, used with the PIC32 microcontroller (a MIPS variant).

Required file extension in MPLAB X

".S" (dot capital S). It allows C-preprocessor directives.

#include <xc.h>

Required C-preprocessor directive. Defines built-in labels.

#define century 100

Optional C-preprocessor directive that defines a named constant.

Comment styles to use in .S files

// for single-line comments and /* ... */ for multi-line comments. AVOID the native MIPS # comment.

Label

A name followed by a colon (e.g. string:) that represents an address. Used for variables, constants, and branch/jump targets.

Assembler directive

A keyword starting with "." that gives the assembler instructions (e.g. .word, .data). It is not a CPU instruction.

.word

Reserves one 32-bit word (4 bytes) per value, initialized to the given value. Example: .word 3456

.byte

Reserves 1 byte per value. You can list more than one value per line. Example: .byte 0x41, 'B'

.space

Reserves the given number of bytes (uninitialized). Example: .space 24

.data

Segment directive. Allocates data storage. Variables go here.

.text

Segment directive. Allocates instruction code. Instructions (and constants, at the end) go here.

.global main

Makes the label main visible to other files. Used for any label referenced in another file.

.set noreorder

Tells the assembler NOT to rearrange instructions or fill delay slots. You must handle the delay slots yourself.

What decides whether a label is a variable or a constant?

Its segment: .data = variable, .text = constant.

Variable (in assembly terms)

A memory location. The name is a label (= an address), and the value is the contents of that memory location.

Address vs. value

Address = WHERE something is in memory (the drawer number). Value = WHAT is stored there (the drawer's contents).

Sequential execution

Instructions run one after another, in order, unless a jump or branch changes the flow.

Control transfer

Jumps and branches change which instruction runs next. A label marks the target.

if/else flow in assembly

A branch skips forward past instructions to a label.

for/while loop flow in assembly

A branch goes backward to a label at the top of the loop.

Function call flow in assembly

A jump to the function's label, then a return back.

Do register and memory values reset on their own?

No. They persist until an instruction explicitly changes them.

la

Load Address. Puts a label's address into a register. It is a macro instruction.

li

Load Immediate. Puts a constant value into a register. It is a macro instruction.

lw

Load Word. Copies a 32-bit word from memory into a register. It is a real machine instruction.

sw

Store Word. Copies a 32-bit word from a register into memory. It is a real machine instruction.

Machine instruction vs. macro instruction

Machine instruction = a real CPU instruction (lw, sw). Macro instruction = assembler shorthand that expands into one or more machine instructions (la, li).

Why does li t9, 0x1234abcd need two instructions?

Each instruction is 32 bits, but only 16 bits are left for a constant. A 32-bit constant is split into an upper half (lui) and a lower half (ori).

What li t9, 0x1234abcd expands to

lui at, 0x1234 | ori t9, at, 0xabcd

Offset in lw t1, 8(t0)

The 8. Address used = offset + address in register (8 + t0).

Byte addressing

Each memory address holds 1 byte. A word (4 bytes) takes up 4 addresses.

Size of a MIPS word

32 bits = 4 bytes

Offset of an array element (word array)

offset = index × 4

Which register gets changed in lw t1, 0(t0)?

t1 (the left/destination register). t0 only supplies the address.

MIPS / PIC32 environment challenges

I/O limitations, managing a complex environment, virtual vs. physical addresses, no explicit stack instructions, and a verbose function-calling structure.