MIPS Assembly/Instruction Formats

This page describes the implementation details of the MIPS instruction formats.

1 R Instructions

1.1 R Format

2.1 I Format

3.1 J Format

R Instructions

R instructions are used when all the data values used by the instruction are located in registers.

All R-type instructions have the following format:

OP rd, rs, rt

Where "OP" is the mnemonic for the particular instruction. R instructions all use the opcode 0, with the function in the funct field. rs, and rt are the source registers, and rd is the destination register. As an example, the add mnemonic can be used as:

add $s1, $s2, $s3

Where the values in $s2 and $s3 are added together, and the result is stored in $s1. In the main narrative of this book, the operands will be denoted by these names.

R Format

Converting an R mnemonic into the equivalent binary machine code is performed in the following way:

opcode	rs	rt	rd	shift (shamt)	funct
6 bits	5 bits	5 bits	5 bits	5 bits	6 bits

opcode The opcode is the machinecode representation of the instruction mnemonic. The opcode field is 6 bits long (bit 26 to bit 31) but always set to 0 in the R format. The actual instruction to perform is placed in the funct field. rs, rt, rd The numeric representations of the source registers and the destination register. These numbers correspond to the $X representation of a register, such as $0 or $31. Each of these fields is 5 bits long. (25 to 21, 20 to 16, and 15 to 11, respectively). Interestingly, rather than rs and rt being named r1 and r2 (for source register 1 and 2), the registers were named "rs" and "rt" for register source, register target and register destination. Shift (shamt) Used with the shift and rotate instructions, this is the amount by which the source operand rs is rotated/shifted. This field is 5 bits long (6 to 10). Funct In R format instructions, the opcode is always zero, and the funct parameter contains the code for the different instructions. 6 bits long (0 to 5). For instance, to add numbers in two registers and place the result in a third, the opcode will be set to zero and the function to 0x20.

I Instructions

I instructions are used when the instruction must operate on an immediate value and a register value. Immediate values may be a maximum of 16 bits long. Larger numbers may not be manipulated by immediate instructions.

I instructions are called in the following way:

OP rt, rs, IMM

However, sw and lw instructions are called in the following way:

OP rt, IMM(rs)

Where rt is the target register, rs is the source register, and IMM is the immediate value. The immediate value can be up to 16 bits long. For instance, the addi instruction can be called as:

addi $s1, $s2, 100

Where the value of $s2 plus 100 is stored in $s1.

I Format

I instructions are converted into machine code words in the following format:

opcode	rs	rt	IMM
6 bits	5 bits	5 bits	16 bits

Opcode The 6-bit opcode of the instruction. In I instructions, all mnemonics have a one-to-one correspondence with the underlying opcodes. This is because there is no funct parameter to differentiate instructions with an identical opcode. 6 bits (26 to 31) rs, rt The source and target register operands, respectively. 5 bits each (21 to 25 and 16 to 20, respectively). [ 1 ] IMM The 16 bit immediate value. 16 bits (0 to 15). This value is usually used as the offset value in various instructions, and depending on the instruction, may be expressed in two's complement.

J Instructions

J instructions are used when a jump needs to be performed. The J instruction has the most space for an immediate value, because addresses are large numbers.

J instructions are called in the following way:

OP LABEL

Where OP is the mnemonic for the particular jump instruction, and LABEL is the target address to jump to.

J Format

J instructions have the following machine-code format:

Opcode	Pseudo-Address
6 bits	26 bits

Opcode The 6 bit opcode corresponding to the particular jump command. (26 to 31). Address A 26-bit shortened address of the destination. (0 to 25). The full 32-bit destination address is formed by concatenating the highest 4 bits of the PC (the address of the instruction following the jump), the 26-bit pseudo-address, and 2 zero bits (since instructions are always aligned on a 32-bit word).

FR Instructions

FR instructions are similar to the R instructions described above, except they are reserved for use with floating-point numbers:

Opcode

fmt

funct

FI Instructions

FI instructions are similar to the I instructions described above, except they are reserved for use with floating-point numbers: