GPU/Shader Instruction Set
Contents
Overview[edit]
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] and the Direct3D shader code [4]. Please note that this page is being written as the instruction set is reverse engineered; as such it may very well contain mistakes.
Debug information found in the code.bin of "Ironfall: Invasion" suggests that there may not be more than 512 instructions and 128 operand descriptors in a shader.
Nomenclature[edit]
- 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[edit]
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 | 0x5 | Source 2 register (SRC2) |
0xC | 0x7 | Source 1 register (SRC1) |
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 | 0x5 | Source 1 register (SRC1) |
0x16 | 0x2 | Address register index for SRC2 (IDX_2) |
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 | 0x5 | Source 1 register (SRC1) |
0x16 | 0x2 | Address register index for SRC3 (IDX_3) |
0x18 | 0x5 | Destination register (DST) |
0x1D | 0x3 | Opcode |
Instructions[edit]
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 3-component vector with 1.0 appended to it and a 4-component vector; DST = SRC1.SRC2 (with SRC1 homogenous) |
0x04 | 1 | DST | Equivalent to Microsoft's dst instruction: DST = {1, SRC1[1]*SRC2[1], SRC1[2], SRC2[3]} |
0x05 | 1u | EX2 | Computes SRC1's first component exponent with base 2; DST[i] = EXP2(SRC1[0]) for all i |
0x06 | 1u | LG2 | Computes SRC1's first component logarithm with base 2; DST[i] = LOG2(SRC1[0]) for all i |
0x07 | 1u | LITP | Appears to be related to Microsoft's lit instruction; DST = clamp(SRC1, min={0, -127.9961, 0, 0}, max={inf, 127.9961, 0, inf}); n.b.: 127.9961 = 0x7FFF / 0x100 |
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's first component; DST[i] = 1/SRC1[0] for all i |
0x0F | 1u | RSQ | Computes the reciprocal of the square root of the vector's first component; DST[i] = 1/sqrt(SRC1[0]) 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 3-component vector with 1.0 appended to it and a 4-component vector; DST = SRC1.SRC2 (with SRC1 homogenous) |
0x19 | 1i | DSTI | DST with sources swapped. |
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 | 0 | BREAK | Breaks out of LOOP block; do not use while in nested IF/CALL block inside LOOP block. |
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. Having bit 0 of NUM = 1 will invert the test, jumping if BOOL is false instead. |
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; this is not an FMA, the intermediate result is rounded |
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; this is not an FMA, the intermediate result is rounded |
Operand descriptors[edit]
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.
Depending on the current shader opcode, source components are disabled implicitly by setting the destination component mask. For example, ADD o0.xy, r0.xyzw, r1.xyzw will not make use of r0's or r1's z/w components, while DP4 o0.xy, r0.xyzw, r1.xyzw will use all input components regardless of the used destination component mask.
Relative addressing[edit]
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. It is only possible to use address registers with vector uniform registers, attempting to use them with input attribute or temporary registers results in the address register being ignored (i.e. read as zero).
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[edit]
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[edit]
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[edit]
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. |
Floating-Point Behavior[edit]
The PICA200 is not IEEE-compliant. It has positive and negative infinities and NaN, but does not seem to have negative 0. Input and output subnormals are flushed to +0. The internal floating point format seems to be the same as used in shader binaries: 1 sign bit, 7 exponent bits, 16 (explicit) mantissa bits. Several instructions also have behavior that differs from the IEEE functions. Here are the results from some tests done on hardware (s = largest subnormal, n = smallest positive normal):
Computation | Result | Notes |
---|---|---|
inf * 0 | 0 | Including inside MUL, MAD, DP4, etc. |
NaN * 0 | NaN | |
+inf - +inf | NaN | Indicates +inf is real inf, not FLT_MAX |
rsq(rcp(-inf)) | +inf | Indicates that there isn't -0.0. |
rcp(-0) | +inf | no -0 so differs from IEEE where rcp(-0) = -inf |
rcp(0) | +inf | |
rcp(+inf) | 0 | |
rcp(NaN) | NaN | |
rsq(-0) | +inf | no -0 so differs from IEEE where rsq(-0) = -inf |
rsq(-2) | NaN | |
rsq(+inf) | 0 | |
rsq(-inf) | NaN | |
rsq(NaN) | NaN | |
max(0, +inf) | +inf | |
max(0, -inf) | -inf | |
max(0, NaN) | NaN | max violates IEEE but match GLSL spec |
max(NaN, 0) | 0 | |
max(-inf, +inf) | +inf | |
min(0, +inf) | 0 | |
min(0, -inf) | -inf | |
min(0, NaN) | NaN | min violates IEEE but match GLSL spec |
min(NaN, 0) | 0 | |
min(-inf, +inf) | -inf | |
cmp(s, 0) | false | cmp does not flush input subnormals |
max(s, 0) | s | max does not flush input or output subnormals |
mul(s, 2) | 0 | input subnormals are flushed in arithmetic instructions |
mul(n, 0.5) | 0 | output subnormals are flushed in arithmetic instructions |
1.0 can be multiplied 63 times by 0.5 until the result compares equal zero. This is consistent with a 7-bit exponent and output subnormal flushing.
Control Flow[edit]
Control flow is implemented using four independent stacks:
- 4-deep CALL stack
- 8-deep IF stack
- 4-deep LOOP stack
All stacks are initially empty. After every instruction but before JMP takes effect, the PC is incremented and a copy is sent to each stack. Each stack is checked against its copy of the PC. If an entry is popped from the stack, the copied PC is updated and used for the next check of this stack, although the IF/LOOP stacks can each only pop one entry per instruction, whereas the CALL stack is checked again until it doesn't match or the stack is empty. The updated PC copy with the highest priority wins: LOOP (highest), IF, CALL, JMP, original PC (lowest).
Special cases:
- JMP overwrites the PC *after* the stacks checks (and only if no stack was popped).
- Executing a BREAK on an empty LOOP stack hangs the GPU.
- A stack overflow discards the oldest element, so you could think of it as a queue or a ring buffer.
- If the CALL stack is popped four times in a row, the fourth update to its copy of the PC is missed (the third PC update will be propagated). Probably a hardware bug.