# Syntax of AMDGPU Instruction Operands¶

## Conventions¶

The following notation is used throughout this document:

Notation Description
{0..N} Any integer value in the range from 0 to N (inclusive).
<x> Syntax and meaning of x is explained elsewhere.

## Operands¶

### v¶

Vector registers. There are 256 32-bit vector registers.

A sequence of vector registers may be used to operate with more than 32 bits of data.

Assembler currently supports sequences of 1, 2, 3, 4, 8 and 16 vector registers.

Syntax Description
v<N>

A single 32-bit vector register.

N must be a decimal integer number.

v[<N>]

A single 32-bit vector register.

N may be specified as an integer number or an absolute expression.

v[<N>:<K>]

A sequence of (K-N+1) vector registers.

N and K may be specified as integer numbers or absolute expressions.

[v<N>, v<N+1>, … v<K>]

A sequence of (K-N+1) vector registers.

Register indices must be specified as decimal integer numbers.

Note. N and K must satisfy the following conditions:

• N <= K.
• 0 <= N <= 255.
• 0 <= K <= 255.
• K-N+1 must be equal to 1, 2, 3, 4, 8 or 16.

Examples:

```v255
v
v[0:1]
v[1:1]
v[0:3]
v[2*2]
v[1-1:2-1]
[v252]
[v252,v253,v254,v255]
```

### s¶

Scalar 32-bit registers. The number of available scalar registers depends on GPU:

GPU Number of scalar registers
GFX7 104
GFX8 102
GFX9 102

A sequence of scalar registers may be used to operate with more than 32 bits of data. Assembler currently supports sequences of 1, 2, 4, 8 and 16 scalar registers.

Pairs of scalar registers must be even-aligned (the first register must be even). Sequences of 4 and more scalar registers must be quad-aligned.

Syntax Description
s<N>

A single 32-bit scalar register.

N must be a decimal integer number.

s[<N>]

A single 32-bit scalar register.

N may be specified as an integer number or an absolute expression.

s[<N>:<K>]

A sequence of (K-N+1) scalar registers.

N and K may be specified as integer numbers or absolute expressions.

[s<N>, s<N+1>, … s<K>]

A sequence of (K-N+1) scalar registers.

Register indices must be specified as decimal integer numbers.

Note. N and K must satisfy the following conditions:

• N must be properly aligned based on sequence size.
• N <= K.
• 0 <= N < SMAX, where SMAX is the number of available scalar registers.
• 0 <= K < SMAX, where SMAX is the number of available scalar registers.
• K-N+1 must be equal to 1, 2, 4, 8 or 16.

Examples:

```s0
s
s[0:1]
s[1:1]
s[0:3]
s[2*2]
s[1-1:2-1]
[s4]
[s4,s5,s6,s7]
```

Examples of scalar registers with an invalid alignment:

```s[1:2]
s[2:5]
```

### trap¶

A set of trap handler registers:

### ttmp¶

Trap handler temporary scalar registers, 32-bits wide. The number of available ttmp registers depends on GPU:

GPU Number of ttmp registers
GFX7 12
GFX8 12
GFX9 16

A sequence of ttmp registers may be used to operate with more than 32 bits of data. Assembler currently supports sequences of 1, 2, 4, 8 and 16 ttmp registers.

Pairs of ttmp registers must be even-aligned (the first register must be even). Sequences of 4 and more ttmp registers must be quad-aligned.

Syntax Description
ttmp<N>

A single 32-bit ttmp register.

N must be a decimal integer number.

ttmp[<N>]

A single 32-bit ttmp register.

N may be specified as an integer number or an absolute expression.

ttmp[<N>:<K>]

A sequence of (K-N+1) ttmp registers.

N and K may be specified as integer numbers or absolute expressions.

[ttmp<N>, ttmp<N+1>, … ttmp<K>]

A sequence of (K-N+1) ttmp registers.

Register indices must be specified as decimal integer numbers.

Note. N and K must satisfy the following conditions:

• N must be properly aligned based on sequence size.
• N <= K.
• 0 <= N < TMAX, where TMAX is the number of available ttmp registers.
• 0 <= K < TMAX, where TMAX is the number of available ttmp registers.
• K-N+1 must be equal to 1, 2, 4, 8 or 16.

Examples:

```ttmp0
ttmp
ttmp[0:1]
ttmp[1:1]
ttmp[0:3]
ttmp[2*2]
ttmp[1-1:2-1]
[ttmp4]
[ttmp4,ttmp5,ttmp6,ttmp7]
```

Examples of ttmp registers with an invalid alignment:

```ttmp[1:2]
ttmp[2:5]
```

### tba¶

Trap base address, 64-bits wide. Holds the pointer to the current trap handler program.

Syntax Description Availability
tba 64-bit trap base address register. GFX7, GFX8
[tba] 64-bit trap base address register (an alternative syntax). GFX7, GFX8
[tba_lo,tba_hi] 64-bit trap base address register (an alternative syntax). GFX7, GFX8

High and low 32 bits of trap base address may be accessed as separate registers:

Syntax Description Availability
tba_lo Low 32 bits of trap base address register. GFX7, GFX8
tba_hi High 32 bits of trap base address register. GFX7, GFX8
[tba_lo] Low 32 bits of trap base address register (an alternative syntax). GFX7, GFX8
[tba_hi] High 32 bits of trap base address register (an alternative syntax). GFX7, GFX8

Note that tba, tba_lo and tba_hi are not accessible as assembler registers in GFX9, but tba is readable/writable with the help of s_get_reg and s_set_reg instructions.

### tma¶

Syntax Description Availability
tma 64-bit trap memory address register. GFX7, GFX8
[tma] 64-bit trap memory address register (an alternative syntax). GFX7, GFX8
[tma_lo,tma_hi] 64-bit trap memory address register (an alternative syntax). GFX7, GFX8

High and low 32 bits of trap memory address may be accessed as separate registers:

Syntax Description Availability
tma_lo Low 32 bits of trap memory address register. GFX7, GFX8
tma_hi High 32 bits of trap memory address register. GFX7, GFX8
[tma_lo] Low 32 bits of trap memory address register (an alternative syntax). GFX7, GFX8
[tma_hi] High 32 bits of trap memory address register (an alternative syntax). GFX7, GFX8

Note that tma, tma_lo and tma_hi are not accessible as assembler registers in GFX9, but tma is readable/writable with the help of s_get_reg and s_set_reg instructions.

### flat_scratch¶

Flat scratch address, 64-bits wide. Holds the base address of scratch memory.

Syntax Description
flat_scratch 64-bit flat scratch address register.
[flat_scratch] 64-bit flat scratch address register (an alternative syntax).
[flat_scratch_lo,flat_scratch_hi] 64-bit flat scratch address register (an alternative syntax).

High and low 32 bits of flat scratch address may be accessed as separate registers:

Syntax Description
flat_scratch_lo Low 32 bits of flat scratch address register.
flat_scratch_hi High 32 bits of flat scratch address register.
[flat_scratch_lo] Low 32 bits of flat scratch address register (an alternative syntax).
[flat_scratch_hi] High 32 bits of flat scratch address register (an alternative syntax).

### xnack¶

Warning

GFX7 does not support xnack feature. Not all GFX8 and GFX9 processors support xnack feature.

Syntax Description

High and low 32 bits of xnack mask may be accessed as separate registers:

Syntax Description

### vcc¶

Vector condition code, 64-bits wide. A bit mask with one bit per thread; it holds the result of a vector compare operation.

Syntax Description
vcc 64-bit vector condition code register.
[vcc] 64-bit vector condition code register (an alternative syntax).
[vcc_lo,vcc_hi] 64-bit vector condition code register (an alternative syntax).

High and low 32 bits of vector condition code may be accessed as separate registers:

Syntax Description
vcc_lo Low 32 bits of vector condition code register.
vcc_hi High 32 bits of vector condition code register.
[vcc_lo] Low 32 bits of vector condition code register (an alternative syntax).
[vcc_hi] High 32 bits of vector condition code register (an alternative syntax).

### m0¶

A 32-bit memory register. It has various uses, including register indexing and bounds checking.

Syntax Description
m0 A 32-bit memory register.
[m0] A 32-bit memory register (an alternative syntax).

### exec¶

Execute mask, 64-bits wide. A bit mask with one bit per thread, which is applied to vector instructions and controls which threads execute and which ignore the instruction.

Syntax Description
[exec] 64-bit execute mask register (an alternative syntax).
[exec_lo,exec_hi] 64-bit execute mask register (an alternative syntax).

High and low 32 bits of execute mask may be accessed as separate registers:

Syntax Description
exec_lo Low 32 bits of execute mask register.
exec_hi High 32 bits of execute mask register.
[exec_lo] Low 32 bits of execute mask register (an alternative syntax).
[exec_hi] High 32 bits of execute mask register (an alternative syntax).

### vccz¶

A single bit-flag indicating that the vcc is all zeros.

Warning

This operand is not currently supported by AMDGPU assembler.

### execz¶

A single bit flag indicating that the exec is all zeros.

Warning

This operand is not currently supported by AMDGPU assembler.

### scc¶

A single bit flag indicating the result of a scalar compare operation.

Warning

This operand is not currently supported by AMDGPU assembler.

### lds_direct¶

A special operand which supplies a 32-bit value fetched from LDS memory using m0 as an address.

Warning

This operand is not currently supported by AMDGPU assembler.

### constant¶

A set of integer and floating-point inline constants:

These operands are encoded as a part of instruction.

If a number may be encoded as either a literal or an inline constant, assembler selects the latter encoding as more efficient.

### iconst¶

An integer number encoded as an inline constant.

Only a small fraction of integer numbers may be encoded as inline constants. They are enumerated in the table below. Other integer numbers have to be encoded as literals.

Integer inline constants are converted to expected operand type as described here.

Value Note
{0..64} Positive integer inline constants.
{-16..-1} Negative integer inline constants.

Warning

GFX7 does not support inline constants for f16 operands.

Warning

These inline constants are not currently supported by AMDGPU assembler.

Syntax Note Availability
shared_base Base address of shared memory region. GFX9
shared_limit Address of the end of shared memory region. GFX9
private_base Base address of private memory region. GFX9
private_limit Address of the end of private memory region. GFX9
pops_exiting_wave_id A dedicated counter for POPS. GFX9

### fconst¶

A floating-point number encoded as an inline constant.

Only a small fraction of floating-point numbers may be encoded as inline constants. They are enumerated in the table below. Other floating-point numbers have to be encoded as literals.

Floating-point inline constants are converted to expected operand type as described here.

Value Note Availability
0.0 The same as integer constant 0. All GPUs
0.5 Floating-point constant 0.5 All GPUs
1.0 Floating-point constant 1.0 All GPUs
2.0 Floating-point constant 2.0 All GPUs
4.0 Floating-point constant 4.0 All GPUs
-0.5 Floating-point constant -0.5 All GPUs
-1.0 Floating-point constant -1.0 All GPUs
-2.0 Floating-point constant -2.0 All GPUs
-4.0 Floating-point constant -4.0 All GPUs
0.1592 1.0/(2.0*pi). Use only for 16-bit operands. GFX8, GFX9
0.15915494 1.0/(2.0*pi). Use only for 16- and 32-bit operands. GFX8, GFX9
0.159154943091895317852646485335 1.0/(2.0*pi). GFX8, GFX9

Warning

GFX7 does not support inline constants for f16 operands.

### literal¶

A literal is a 64-bit value which is encoded as a separate 32-bit dword in the instruction stream.

If a number may be encoded as either a literal or an inline constant, assembler selects the latter encoding as more efficient.

Literals may be specified as integer numbers, floating-point numbers or expressions (expressions are currently supported for 32-bit operands only).

A 64-bit literal value is converted by assembler to an expected operand type as described here.

An instruction may use only one literal but several operands may refer the same literal.

### uimm8¶

A 8-bit positive integer number. The value is encoded as part of the opcode so it is free to use.

### uimm32¶

A 32-bit positive integer number. The value is stored as a separate 32-bit dword in the instruction stream.

### uimm20¶

A 20-bit positive integer number.

### uimm21¶

A 21-bit positive integer number.

Warning

Assembler currently supports 20-bit offsets only. Use uimm20 as a replacement.

### simm21¶

A 21-bit integer number.

Warning

Assembler currently supports 20-bit unsigned offsets only .Use uimm20 as a replacement.

### off¶

A special entity which indicates that the value of this operand is not used.

Syntax Description
off Indicates an unused operand.

## Numbers¶

### Integer Numbers¶

Integer numbers are 64 bits wide. They may be specified in binary, octal, hexadecimal and decimal formats:

Format Syntax
Decimal [-]?[1-9][0-9]*
Binary [-]?0b+
Octal [-]?0[0-7]+
[-]?[0x]?[0-9][0-9a-fA-F]*[hH]

Examples:

```-1234
0b1010
010
0xff
0ffh
```

### Floating-Point Numbers¶

All floating-point numbers are handled as double (64 bits wide).

Floating-point numbers may be specified in hexadecimal and decimal formats:

Format Syntax Note
Decimal [-]?[0-9]*[.][0-9]*([eE][+-]?[0-9]*)? Must include either a decimal separator or an exponent.

Examples:

```-1.234
234e2
-0x1afp-10
0x.1afp10
```

## Expressions¶

An expression specifies an address or a numeric value. There are two kinds of expressions:

### Absolute Expressions¶

The value of an absolute expression remains the same after program relocation. Absolute expressions must not include unassigned and relocatable values such as labels.

Examples:

```x = -1
y = x + 10
```

### Relocatable Expressions¶

The value of a relocatable expression depends on program relocation.

Note that use of relocatable expressions is limited with branch targets and 32-bit literals.

Examples:

```y = x + 10 // x is not yet defined. Undefined symbols are assumed to be PC-relative.
z = .
```

### Expression Data Type¶

Expressions and operands of expressions are interpreted as 64-bit integers.

Expressions may include 64-bit floating-point numbers (double). However these operands are also handled as 64-bit integers using binary representation of specified floating-point numbers. No conversion from floating-point to integer is performed.

Examples:

```x = 0.1    // x is assigned an integer 4591870180066957722 which is a binary representation of 0.1.
y = x + x  // y is a sum of two integer values; it is not equal to 0.2!
```

### Syntax¶

Expressions are composed of symbols, integer numbers, floating-point numbers, binary operators, unary operators and subexpressions.

Expressions may also use “.” which is a reference to the current PC (program counter).

The syntax of expressions is shown below:

```expr ::= expr binop expr | primaryexpr ;

primaryexpr ::= '(' expr ')' | symbol | number | '.' | unop primaryexpr ;

binop ::= '&&'
| '||'
| '|'
| '^'
| '&'
| '!'
| '=='
| '!='
| '<>'
| '<'
| '<='
| '>'
| '>='
| '<<'
| '>>'
| '+'
| '-'
| '*'
| '/'
| '%' ;

unop ::= '~'
| '+'
| '-'
| '!' ;
```

### Binary Operators¶

Binary operators are described in the following table. They operate on and produce 64-bit integers. Operators with higher priority are performed first.

Operator Priority Meaning
* 5 Integer multiplication.
/ 5 Integer division.
% 5 Integer signed remainder.
- 4 Integer subtraction.
<< 3 Integer shift left.
>> 3 Logical shift right.
== 2 Equality comparison.
!= 2 Inequality comparison.
<> 2 Inequality comparison.
< 2 Signed less than comparison.
<= 2 Signed less than or equal comparison.
> 2 Signed greater than comparison.
>= 2 Signed greater than or equal comparison.
| 1 Bitwise or.
^ 1 Bitwise xor.
& 1 Bitwise and.
&& 0 Logical and.
|| 0 Logical or.

### Unary Operators¶

Unary operators are described in the following table. They operate on and produce 64-bit integers.

Operator Meaning
! Logical negation.
~ Bitwise negation.
+ Integer unary plus.
- Integer unary minus.

### Symbols¶

A symbol is a named 64-bit value, representing a relocatable address or an absolute (non-relocatable) number.

Symbol names have the following syntax:
`[a-zA-Z_.][a-zA-Z0-9_\$.@]*`

The table below provides several examples of syntax used for symbol definition.

Syntax Meaning
.globl <S> Declares a global symbol S without assigning it a value.
.set <S>, <E> Assigns the value of an expression E to a symbol S.
<S> = <E> Assigns the value of an expression E to a symbol S.
<S>: Declares a label S and assigns it the current PC value.

A symbol may be used before it is declared or assigned; unassigned symbols are assumed to be PC-relative.

## Conversions¶

This section describes what happens when a 64-bit integer number, a floating-point numbers or a symbol is used for an operand which has a different type or size.

Depending on operand kind, this conversion is performed by either assembler or AMDGPU H/W:

### Inline Constants¶

#### Integer Inline Constants¶

Integer inline constants may be thought of as 64-bit integer numbers; when used as operands they are truncated to the size of expected operand type. No data type conversions are performed.

Examples:

```// GFX9

v_add_u16 v0, -1, 0    // v0 = 0xFFFF
v_add_f16 v0, -1, 0    // v0 = 0xFFFF (NaN)

v_add_u32 v0, -1, 0    // v0 = 0xFFFFFFFF
v_add_f32 v0, -1, 0    // v0 = 0xFFFFFFFF (NaN)
```

#### Floating-Point Inline Constants¶

Floating-point inline constants may be thought of as 64-bit floating-point numbers; when used as operands they are converted to a floating-point number of expected operand size.

Examples:

```// GFX9

v_add_f16 v0, 1.0, 0    // v0 = 0x3C00 (1.0)
v_add_u16 v0, 1.0, 0    // v0 = 0x3C00

v_add_f32 v0, 1.0, 0    // v0 = 0x3F800000 (1.0)
v_add_u32 v0, 1.0, 0    // v0 = 0x3F800000
```

### Literals¶

#### Integer Literals¶

Integer literals are specified as 64-bit integer numbers.

When used as operands they are converted to expected operand type as described below.

Expected type Condition Result Note
i16, u16, b16 cond(num,16) num.u16 Truncate to 16 bits.
i32, u32, b32 cond(num,32) num.u32 Truncate to 32 bits.
i64 cond(num,32) {-1,num.i32} Truncate to 32 bits and then sign-extend the result to 64 bits.
u64, b64 cond(num,32) { 0,num.u32} Truncate to 32 bits and then zero-extend the result to 64 bits.
f16 cond(num,16) num.u16 Use low 16 bits as an f16 value.
f32 cond(num,32) num.u32 Use low 32 bits as an f32 value.
f64 cond(num,32) {num.u32,0} Use low 32 bits of the number as high 32 bits of the result; low 32 bits of the result are zeroed.

The condition cond(X,S) indicates if a 64-bit number X can be converted to a smaller size S by truncation of upper bits. There are two cases when the conversion is possible:

• The truncated bits are all 0.
• The truncated bits are all 1 and the value after truncation has its MSB bit set.

Examples of valid literals:

```// GFX9
// Literal value after conversion:
v_add_u16 v0, 0xff00, v0                 //   0xff00
v_add_u16 v0, 0xffffffffffffff00, v0     //   0xff00
v_add_u16 v0, -256, v0                   //   0xff00
// Literal value after conversion:
s_bfe_i64 s[0:1], 0xffefffff, s3         //   0xffffffffffefffff
s_bfe_u64 s[0:1], 0xffefffff, s3         //   0x00000000ffefffff
v_ceil_f64_e32 v[0:1], 0xffefffff        //   0xffefffff00000000 (-1.7976922776554302e308)
```

Examples of invalid literals:

```// GFX9

v_add_u16 v0, 0x1ff00, v0               // truncated bits are not all 0 or 1
v_add_u16 v0, 0xffffffffffff00ff, v0    // truncated bits do not match MSB of the result
```

#### Floating-Point Literals¶

Floating-point literals are specified as 64-bit floating-point numbers.

When used as operands they are converted to expected operand type as described below.

Expected type Condition Result Note
i16, u16, b16 cond(num,16) f16(num) Convert to f16 and use bits of the result as an integer value.
i32, u32, b32 cond(num,32) f32(num) Convert to f32 and use bits of the result as an integer value.
i64, u64, b64 false - Conversion disabled because of an unclear semantics.
f16 cond(num,16) f16(num) Convert to f16.
f32 cond(num,32) f32(num) Convert to f32.
f64 true {num.u32.hi,0}

Use high 32 bits of the number as high 32 bits of the result; zero-fill low 32 bits of the result.

Note that the result may differ from the original number.

The condition cond(X,S) indicates if an f64 number X can be converted to a smaller S-bit floating-point type without overflow or underflow. Precision lost is allowed.

Examples of valid literals:

```// GFX9

// Literal value before conversion: 1.7976931348623157e308 (0x7fefffffffffffff)
// Literal value after conversion:  1.7976922776554302e308 (0x7fefffff00000000)
v_ceil_f64 v[0:1], 1.7976931348623157e308
```

Examples of invalid literals:

```// GFX9

v_add_f16 v1, 65600.0, v2    // overflow
```

#### Expressions¶

Expressions operate with and result in 64-bit integers.

When used as operands they are truncated to expected operand size. No data type conversions are performed.

Examples:

```// GFX9

x = 0.1
v_sqrt_f32 v0, x           // v0 = [low 32 bits of 0.1 (double)]
v_sqrt_f32 v0, (0.1 + 0)   // the same as above
v_sqrt_f32 v0, 0.1         // v0 = [0.1 (double) converted to float]
```