GPU/Shader Instruction Set

Overview

A compiled shader binary is comprised of two parts : the main instruction sequence and the operand descriptor table. These are both sent to the GPU around the same time but using separate GPU Commands. Instructions (such as format 1 instruction) may reference operand descriptors. When such is the case, the operand descriptor ID is the offset, in words, of the descriptor within the table. Both instructions and descriptors are coded in little endian. Basic implementations of the following specification can be found at [1] and [2]. The instruction set seems to have been heavily inspired by Microsoft's vs_3_0 [3]. Please note that this page is being written as the instruction set is reverse engineered; as such it may very well contain mistakes.

Nomenclature

• opcode names with I appended to them are the same as their non-I version, except they use the inverted instruction format, giving 7 bits to SRC2 (and access to uniforms) and 5 bits to SRC1

• opcode names with U appended to them are the same as their non-U version, except they are executed conditionally based on the value of a uniform boolean.

• opcode names with C appended to them are the same as their non-C version, except they are executed conditionally based on a logical expression specified in the instruction.

Instruction formats

Format 1 : (used for register operations)

Offset	Size (bits)	Description
0x0	0x7	Operand descriptor ID (DESC)
0x7	0x5	Source 2 register (SRC2)
0xC	0x7	Source 1 register (SRC1)
0x13	0x2	Address register index for SRC1 (IDX_1)
0x15	0x5	Destination register (DST)
0x1A	0x6	Opcode

Format 1i : (used for register operations)

Offset	Size (bits)	Description
0x0	0x7	Operand descriptor ID (DESC)
0x7	0x7	Source 2 register (SRC2)
0xE	0x5	Source 1 register (SRC1)
0x13	0x2	Address register index for SRC2 (IDX_2)
0x15	0x5	Destination register (DST)
0x1A	0x6	Opcode

Format 1u : (used for unary register operations)

Offset	Size (bits)	Description
0x0	0x7	Operand descriptor ID (DESC)
0xC	0x7	Source 1 register (SRC1)
0x13	0x2	Address register index for SRC1 (IDX_1)
0x15	0x5	Destination register (DST)
0x1A	0x6	Opcode

Format 1c : (used for comparison operations)

Offset	Size (bits)	Description
0x0	0x7	Operand descriptor ID (DESC)
0x7	0x7	Source 1 register (SRC1)
0xE	0x5	Source 2 register (SRC2)
0x13	0x2	Address register index for SRC1 (IDX_1)
0x15	0x3	Comparison operator for Y (CMPY)
0x18	0x3	Comparison operator for X (CMPX)
0x1B	0x5	Opcode

Format 2 : (used for flow control instructions)

Offset	Size (bits)	Description
0x0	0x8	Number of instructions (NUM)
0xA	0xC	Destination offset (in words) (DST)
0x16	0x2	Condition boolean operator (CONDOP)
0x18	0x1	Y reference bit (REFY)
0x19	0x1	X reference bit (REFX)
0x1A	0x6	Opcode

Format 3 : (used for uniform-based conditional flow control instructions)

Offset	Size (bits)	Description
0x0	0x8	Number of instructions ? (NUM)
0xA	0xC	Destination offset (in words) (DST)
0x16	0x4	Uniform ID (BOOL/INT)
0x1A	0x6	Opcode

Format 4 : (used for SETEMIT)

Offset	Size (bits)	Description
0x16	0x1	Winding flag (FLAG_WINDING)
0x17	0x1	Primitive emit flag (FLAG_PRIMEMIT)
0x18	0x2	Vertex ID (VTXID)
0x1A	0x6	Opcode

Format 5 : (used for MAD)

Offset	Size (bits)	Description
0x0	0x5	Operand descriptor ID (DESC)
0x5	0x5	Source 3 register (SRC3)
0xA	0x7	Source 2 register (SRC2)
0x11	0x7	Source 1 register (SRC1)
0x18	0x5	Destination register (DST)
0x1D	0x3	Opcode

Format 5i : (used for MADI)

Offset	Size (bits)	Description
0x0	0x5	Operand descriptor ID (DESC)
0x5	0x7	Source 3 register (SRC3)
0xC	0x5	Source 2 register (SRC2)
0x11	0x7	Source 1 register (SRC1)
0x18	0x5	Destination register (DST)
0x1D	0x3	Opcode

Instructions

Unless noted otherwise, SRC1 and SRC2 refer to their respectively indexed float[4] registers (after swizzling). Similarly, DST refers to its indexed register modulo destination component masking, i.e. an expression like DST=SRC1 might actually just set DST.y to SRC1.y.

Opcode	Format	Name	Description
0x00	1	ADD	Adds two vectors component by component; DST[i] = SRC1[i]+SRC2[i] for all i
0x01	1	DP3	Computes dot product on 3-component vectors; DST = SRC1.SRC2
0x02	1	DP4	Computes dot product on 4-component vectors; DST = SRC1.SRC2
0x03	1	DPH	Computes dot product on a 4-component vector and a 3-component one with 1.0 appended to it; DST = SRC1.SRC2 (with SRC2 homogenous)
0x04	1	???	?
0x05	1u	EX2	Computes SRC1's per-component exponent with base 2; DST[i] = EXP(SRC1[i]) for all i
0x06	1u	LG2	Computes SRC1's log2 component by component; DST[i] = LOG2(SRC1[i]) for all i
0x07	1u	???	?
0x08	1	MUL	Multiplies two vectors component by component; DST[i] = SRC1[i].SRC2[i] for all i
0x09	1	SGE	Sets output if SRC1 is greater than or equal to SRC2; DST[i] = (SRC1[i] >= SRC2[i]) ? 1.0 : 0.0 for all i
0x0A	1	SLT	Sets output if SRC1 is strictly less than SRC2; DST[i] = (SRC1[i] < SRC2[i]) ? 1.0 : 0.0 for all i
0x0B	1u	FLR	Computes SRC1's floor component by component; DST[i] = FLOOR(SRC1[i]) for all i
0x0C	1	MAX	Takes the max of two vectors, component by component; DST[i] = MAX(SRC1[i], SRC2[i]) for all i
0x0D	1	MIN	Takes the min of two vectors, component by component; DST[i] = MIN(SRC1[i], SRC2[i]) for all i
0x0E	1u	RCP	Computes the reciprocal of the vector, component by component; DST[i] = 1/SRC1[i] for all i
0x0F	1u	RSQ	Computes the reciprocal of the square root of the vector, component by component; DST[i] = 1/sqrt(SRC1[i]) for all i
0x10	?	???	?
0x11	?	???	?
0x12	1u	MOVA	Move to address register; Casts the float uniform given by SRC1 to an integer (truncating the fractional part) and assigns the result to (a0.x, a0.y, _, _), respecting the destination component mask.
0x13	1u	MOV	Moves value from one register to another; DST = SRC1.
0x14	?	???	?
0x15	?	???	?
0x16	?	???	?
0x17	?	???	?
0x18	1i	DPHI	Computes dot product on a 4-component vector and a 3-component one with 1.0 appended to it; DST = SRC1.SRC2 (with SRC2 homogenous)
0x19	1i	???	?
0x1A	1i	SGEI	Sets output if SRC1 is greater than or equal to SRC2; DST[i] = (SRC1[i] >= SRC2[i]) ? 1.0 : 0.0 for all i
0x1B	1i	SLTI	Sets output if SRC1 is strictly less than SRC2; DST[i] = (SRC1[i] < SRC2[i]) ? 1.0 : 0.0 for all i
0x1C	?	???	?
0x1D	?	???	?
0x1E	?	???	?
0x1F	?	???	?
0x20	?	???	?
0x21	0	NOP	Does literally nothing.
0x22	0	END	Signals the shader unit that processing for this vertex/primitive is done.
0x23	2	BREAKC	If condition (see below for details) is true, then breaks out of LOOP block.
0x24	2	CALL	Jumps to DST and executes instructions until it reaches DST+NUM instructions
0x25	2	CALLC	If condition (see below for details) is true, then jumps to DST and executes instructions until it reaches DST+NUM instructions, else does nothing.
0x26	3	CALLU	Jumps to DST and executes instructions until it reaches DST+NUM instructions if BOOL is true
0x27	3	IFU	If condition BOOL is true, then executes instructions until DST, then jumps to DST+NUM; else, jumps to DST.
0x28	2	IFC	If condition (see below for details) is true, then executes instructions until DST, then jumps to DST+NUM; else, jumps to DST
0x29	3	LOOP	Loops over the code between itself and DST (inclusive), performing INT.x+1 iterations in total. First, aL is initialized to INT.y. After each iteration, aL is incremented by INT.z.
0x2A	0 (no param)	EMIT	(geometry shader only) Emits a vertex (and primitive if FLAG_PRIMEMIT was set in the corresponding SETEMIT). SETEMIT must be called before this.
0x2B	4	SETEMIT	(geometry shader only) Sets VTXID, FLAG_WINDING and FLAG_PRIMEMIT for the next EMIT instruction. VTXID is the ID of the vertex about to be emitted within the primitive, while FLAG_PRIMEMIT is zero if we are just emitting a single vertex and non-zero if are emitting a vertex and primitive simultaneously. FLAG_WINDING controls the output primitive's winding. Note that the output vertex buffer (which holds 4 vertices) is not cleared when the primitive is emitted, meaning that vertices from the previous primitive can be reused for the current one. (this is still a working hypothesis and unconfirmed)
0x2C	2	JMPC	If condition (see below for details) is true, then jumps to DST, else does nothing.
0x2D	3	JMPU	If condition BOOL is true, then jumps to DST, else does nothing. It seems possible that having NUM = 1 will jump if BOOL is false instead, though this is unconfirmed.
0x2E-0x2F	1c	CMP	Sets booleans cmp.x and cmp.y based on the operand's x and y components and the CMPX and CMPY comparison operators respectively. See below for details about operators. It's unknown whether CMP respects the destination component mask or not.
0x30-0x37	5i	MADI	Multiplies two vectors and adds a third one component by component; DST[i] = SRC3[i] + SRC2[i].SRC1[i] for all i
0x38-0x3F	5	MAD	Multiplies two vectors and adds a third one component by component; DST[i] = SRC3[i] + SRC2[i].SRC1[i] for all i

Operand descriptors

Sizes below are in bits, not bytes.

Offset	Size	Description
0x0	0x4	Destination component mask. Bit 3 = x, 2 = y, 1 = z, 0 = w.
0x4	0x1	Source 1 negation bit
0x5	0x8	Source 1 component selector
0xD	0x1	Source 2 negation bit
0xE	0x8	Source 2 component selector
0x16	0x1	Source 3 negation bit
0x17	0x8	Source 3 component selector

Component selector :

Offset	Size	Description
0x0	0x2	Component 3 value
0x2	0x2	Component 2 value
0x4	0x2	Component 1 value
0x6	0x2	Component 0 value

Value	Component
0x0	x
0x1	y
0x2	z
0x3	w

The component selector enables swizzling. For example, component selector 0x1B is equivalent to .xyzw, while 0x55 is equivalent to .yyyy.

Relative addressing

There are 3 address registers: a0.x, a0.y and aL (loop counter). For format 1 instructions, when IDX != 0, the value of the corresponding address register is added to SRC1's value. For example, if IDX = 2, a0.y = 3 and SRC1 = c8, then instead SRC1+a0.y = c11 will be used for the instruction.

a0.x and a0.y are set manually through the MOVA instruction by rounding a float value to integer precision. Hence, they may take negative values.

aL can only be set indirectly by the LOOP instruction. It is still accessible and valid after exiting a LOOP block, though.

Comparison operator

CMPX/CMPY raw value	Operator name	Expression
0x0	EQ	src1 == src2
0x1	NE	src1 != src2
0x2	LT	src1 < src2
0x3	LE	src1 <= src2
0x4	GT	src1 > src2
0x5	GE	src1 >= src2
0x6	??	true ?
0x7	??	true ?

6 and 7 seem to always return true.

Conditions

A number of format 2 instructions are executed conditionally. These conditions are based on two boolean registers which can be set with CMP : cmp.x and cmp.y.

Conditional instructions include 3 parameters : CONDOP, REFX and REFY. REFX and REFY are reference values which are tested for equality against cmp.x and cmp.y, respectively. CONDOP describes how the final truth value is constructed from the results of the two tests. There are four conditional expression formats :

CONDOP raw value	Expression	Description
0x0	cmp.x == REFX || cmp.y == REFY	OR
0x1	cmp.x == REFX && cmp.y == REFY	AND
0x2	cmp.x == REFX	X
0x3	cmp.y == REFY	Y

Registers

Input attribute registers (v0-v7?) store the per-vertex data given by the CPU and hence are read-only.

Output attribute registers (o0-o6) hold the data to be passed to the later GPU stages and are write-only. Each of the output attribute register components is assigned a semantic by setting the corresponding GPU_Internal_Registers.

Uniform registers hold user-specified data which is constant throughout all processed vertices. There are 96 float[4] uniform registers (c0-c95), eight boolean registers (b0-b7), and four int[4] registers (i0-i3).

Temporary registers (r0-r15) can be used for intermediate calculations and can both be read and written.

Many shader instructions which take float arguments have only 5 bits available for the second argument. They may hence only refer to input attributes or temporary registers. In particular, it's not possible to pass two float[4] uniforms to these instructions.

It appears that writing twice to the same output register can cause problems (e.g. GPU hangs).

DST mapping :

DST raw value	Register name	Description
0x0-0x6	o0-o6	Output registers.
0x10-0x1F	r0-r15	Temporary registers.

SRC mapping :

SRC1 raw value	Register name	Description
0x0-0x7	v0-v7	Input attribute registers.
0x10-0x1F	r0-r15	Temporary registers.
0x20-0x7F	c0-c95	Vector uniform registers.