SGX543

From Vita Development Wiki
Jump to navigation Jump to search

The PSVita custom SoC (Kermit) embeds the SGX543MP4+ GPU since the model DEM-300xG (2010-09) and SDK 0.920. Older prototype models embeds the SGX541MP.

Instruction set

General Info

It looks like instructions are 8 bytes long. opcode1 field is common to all instructions. However instructions may have additional opcode fields.

Predicate field is common almost to all instructions as well, however it differs in length between different instructions.

Other fields in instruction are usually used to encode operands and their properties such as sign, math operation modifiers, channel masking, swizzling etc.

Some fields of instruction may change its entire behavior by changing data format or applying some modifiers as type cast modifiers etc.

Bit encoding used in this reference:

value meaning
0 bit clear
1 bit set
x dont care
? unknown
see reference

Predicates

Not sure about predicates yet, but they are used to mask execution of certain instructions.

Notation is the following:

<predicate> <instruction>

For example:

!p0 mad.f32

To reduce amount of examples - they are not listed with predicates.

It is assumed that all predicates are applicable to all instructions in the group unless opposite is indicated.

Operands

Different types of operands exist. They are described in further sections:

Instructions may have up to four operands specified.

In this documentation they will be encoded as:

<op0> <op1> <op2> <op3>

Certain instructions allow to adjust position of single normal register operand with opN_sel field.

In this case such operand is encoded as: opN

This should not be confused with Operand N which is used to aggregate operands op1, op2, op3.

Operand 0

Operand 0 R6

Destination operand 0 can be encoded in different ways.

Usually the following fields are used to encode it:

  • alt_opt0 - alter opt0. this bit can be combined with opt0 to produce the following modes for op0:
alt_opt0 opt0 value details
1 0 0 sa
1 0 1 {} op0 encodes CNST6. applicable only with swizzles.
1 1 0 index<N> op0 encodes IDX6
1 1 1 index2 mode op0 encodes RIO6.

or selects other modes for encoding op0 if specified in alt_opt0.

or with Register Index Offset RIO6 using index1 mode if specified in opt0.

or with Constant CNST6 if specified in alt_opt0.

or with Index IDX6 if specified in alt_opt0.

or with Register Index Offset RIO6 using index2 mode if specified in alt_opt0.

Operand 0 R7

Destination operand 0 can be encoded in different ways.

Usually the following fields are used to encode it:

or with Register Index Offset RIO7 using index1 mode if specified in opt0.

Operand N

Source operand <N> (where N > 0) can be encoded in different ways.

Usually the following fields are used to encode it:

  • alt_opt<N> - alter opt<N>. this bit can be combined with opt<N> to produce the following modes for op<N >:
alt_opt<N> opt<N> value details
1 0 0 index1 mode op<N> encodes RIO6.
1 0 1 {} op<N> encodes CNST6. applicable only with swizzles.
1 1 0 immediate op<N> encodes IMM6.
1 1 1 index2 mode op<N> encodes RIO6.

or selects other modes for encoding op<N> if specified in alt_opt<N>.

or with Register Index Offset RIO6 using index1 mode if specified in alt_opt<N>.

or with CNST6 if specified in alt_opt<N>.

or IMM6 if specified in alt_opt<N>.

or with Register Index Offset RIO6 using index2 mode if specified in alt_opt<N>.

Registers

  • pa - primary attribute register. 32 bit long.
  • sa - secondary attribute register. 32 bit long.
  • o - output register. 32 bit long.
  • r - temporary register. 32 bit long.
  • i - internal register. 128 bit long.

Register Selector RS2

This encoding uses 2 bits to encode register type.

selector is encoded as:

1 0 meaning
0 0 r
0 1 o
1 0 pa
1 1 sa

Note that internal registers are not encoded - they are reserved in Register R6

Register Selector Indexable RSI2

This encoding uses 2 bits to encode register type.

selector is encoded as:

1 0 meaning
0 0 r
0 1 o
1 0 pa
1 1 index<N> mode

When index<N> mode is used - there has to be another field that encodes index expression with Register Index Offset RIO6

The way that index expression is buit:

<reg>[index1 * 2 + <offset>]

Example:

r[index1 * 2 + 8]

Register R6

This encoding uses 6 bits to encode register index.

Specific type of register can be selected with Register Selector RS2.

For destination operand op0 specific type of register can be selected with Register Selector Indexable RSI2.

index is calculated as: value * 2

Register expression is built as:

<reg><index>

Example:

r68
Register Encoding - r

register is encoded as:

5 4 3 2 1 0 index
0 0 0 0 0 0 0
0 0 0 0 0 1 2
... ... ... ... ... ...
1 1 1 0 1 0 116
1 1 1 0 1 1 118
1 1 1 1 0 0 i0 (reserved)
1 1 1 1 0 1 i1 (reserved)
1 1 1 1 1 0 i2 (reserved)
1 1 1 1 1 1 i3 (reserved)

Last 4 values are reserved for internal registers i0, i1, i2, i3

Register Encoding - o, pa, sa

register is encoded as:

5 4 3 2 1 0 index
0 0 0 0 0 0 0
0 0 0 0 0 1 2
... ... ... ... ... ...
1 1 1 1 1 0 124
1 1 1 1 1 1 126

Register R7

This encoding uses 7 bits to encode register index.

Specific type of register can be selected with Register Selector RS2.

For destination operand op0 specific type of register can be selected with Register Selector Indexable RSI2.

index is calculated as: value * 2

Register expression is built as:

<reg><index>

Example:

r68
Register Encoding - r

register is encoded as:

6 5 4 3 2 1 0 index
0 0 0 0 0 0 0 0
0 0 0 0 0 0 1 2
... ... ... ... ... ...
1 1 1 1 0 1 0 244
1 1 1 1 0 1 1 246
1 1 1 1 1 0 0 i0 (reserved)
1 1 1 1 1 0 1 i1 (reserved)
1 1 1 1 1 1 0 i2 (reserved)
1 1 1 1 1 1 1 i3 (reserved)

Last 4 values are reserved for internal registers i0, i1, i2, i3

Register Encoding - o, pa, sa

register is encoded as:

6 5 4 3 2 1 0 index
0 0 0 0 0 0 0 0
0 0 0 0 0 0 1 2
... ... ... ... ... ...
1 1 1 1 1 1 0 252
1 1 1 1 1 1 1 254

Register RI2

This encoding uses 2 bits to encode internal register.

1 0 value
0 0 i0
0 1 i1
1 0 i2
1 1 i3

Register Index Offset RIO6

5 4 3 2 1 0
rt offset

rt is encoded as Register Selector RS2

offset is calculated as: value * 2

offset is encoded as:

3 2 1 0 offset
0 0 0 0 0
0 0 0 1 2
... ... ... ... ...
1 1 1 0 28
1 1 1 1 30

Register Index Offset RIO7

6 5 4 3 2 1 0
rt offset

rt is encoded as Register Selector RS2

offset is calculated as: value * 2

offset is encoded as:

4 3 2 1 0 offset
0 0 0 0 0 0
0 0 0 0 1 2
... ... ... ... ...
1 1 1 1 0 60
1 1 1 1 1 62

Immediates

Immediate IMM6

Some operands may act as immediate values which are encoded using 6 bits.

Indexes

Not sure what index<N> expression means.

It can be used in 2 places:

  • Alternative mode for operand 0

Index IDX6

Uses 6 bits to encode index<N> expression.

Index is calculated as value * 2. Max index is 126.

Example:

mad.f32 index24, r0, r0, r0

Constants

Constant CNST6

Some operands may act as constant values which are encoded using 6 bits.

Constants are taken from table below.

Constants differ in 32 and 16 bit mode.

32 bit mode

bank 1 is used for channel 1 of op0

for other operands - only bank 0 is used for each channel

f32 mode - bank0:

operand hex value comment
0x00 0x00000000 0.0
0x01 0x00000000 0.0
0x02 0x3F800000 1.0
0x03 0x3F800000 1.0
0x04 0x40000000 2.0 2 ^ 1
0x05 0x41000000 8.0 2 ^ 3
0x06 0x42000000 32.0 2 ^ 5
0x07 0x43000000 128.0 2 ^ 7
0x08 0x44000000 512.0 2 ^ 9
0x09 0x45000000 2048.0 2 ^ 11
0x0A 0x46000000 8192.0 2 ^ 13
0x0B 0x47000000 32768.0 2 ^ 15
0x0C 0x3F000000 0.5 2 ^ -1
0x0D 0x3E000000 0.125 2 ^ -3
0x0E 0x3D000000 0.03125 2 ^ -5
0x0F 0x3C000000 0.0078125 2 ^ -7
0x10 0x3B000000 0.001953125 2 ^ -9
0x11 0x3A000000 0.00048828125 2 ^ -11
0x12 0x39000000 0.00012207031 2 ^ -13
0x13 0x38000000 0.000030517578 2 ^ -15
0x14 0x402DF854 2.7182817 e
0x15 0x3FB504F3 1.4142135 sqrt(2)
0x16 0x40490FDB 3.1415927 pi
0x17 0x3F490FDB 0.78539819 pi / 4
0x18 0x40C90FDB 6.2831855 pi * 2
0x19 0x41C90FDB 25.132742 pi * 8
0x1A 0x37800000 0.000015258789 1 / 2 ^ 16
0x1B 0x37800080 0.000015259022 1 / (2 ^ 16 - 1)
0x1C 0x35D00D01 0.0000015500992 sin(0.5) - taylor 4th term
0x1D 0x39888889 0.00026041668 sin(0.5) - taylor 3rd term
0x1E 0x3CAAAAAB 0.020833334 sin(0.5) - taylor 2nd term
0x1F 0x3F000000 0.5 sin(0.5) - taylor 1st term
0x20 0x00000000 0.0
0x21 0x00000000 0.0
0x22 0x3C003C00 0.0078268051
0x23 0x44004000 513.0
0x24 0x54005000 2.204392e12
0x25 0x64006000 9.4724031e21
0x26 0x74007000 4.0703468e31
0x27 0x34003800 1.1941302e-7
0x28 0x24002800 2.7789457e-17
0x29 0x14001800 6.4670817e-27
0x2A 0x04000800 1.5050001e-36
0x2B 0x35E2416F 0.0000016857356
0x2C 0x39A83DA8 0.00032089395
0x2D 0x3E484248 0.19556534
0x2E 0x4A484648 3281298.0
0x2F 0x00000000 0.0
0x30 0x00000000 0.0
0x31 0x30002555 4.6619181e-10
0x32 0x00000000 0.0
0x33 0x00000000 0.0
0x34 0x00000000 0.0
0x35 0x00000000 0.0
0x36 0x00000000 0.0
0x37 0x00000000 0.0
0x38 0xFFFFFFFF -1.#QNB -nan
0x39 0xFFFFFFFF -1.#QNB -nan
0x3A 0xFFFFFFFF -1.#QNB -nan
0x3B 0xFFFFFFFF -1.#QNB -nan
0x3C 0x7FFF7FFF 1.#QNB +nan
0x3D 0x7FFF7FFF 1.#QNB +nan
0x3E 0x7FFF7FFF 1.#QNB +nan
0x3F 0x7FFF7FFF 1.#QNB +nan

f32 mode - bank1:

operand hex value comment
0x00 0x00000000 0.0
0x01 0x3F800000 1.0
0x02 0x00000000 0.0
0x03 0x3F800000 1.0
0x04 0x40800000 4.0 2 ^ 2
0x05 0x41800000 16.0 2 ^ 4
0x06 0x42800000 64.0 2 ^ 6
0x07 0x43800000 256.0 2 ^ 8
0x08 0x44800000 1024.0 2 ^ 10
0x09 0x45800000 4096.0 2 ^ 12
0x0A 0x46800000 16384.0 2 ^ 14
0x0B 0x47800000 65536.0 2 ^ 16
0x0C 0x3E800000 0.25 2 ^ -2
0x0D 0x3D800000 0.0625 2 ^ -4
0x0E 0x3C800000 0.015625 2 ^ -6
0x0F 0x3B800000 0.00390625 2 ^ -8
0x10 0x3A800000 0.0009765625 2 ^ 10
0x11 0x39800000 0.00024414062 2 ^ 12
0x12 0x38800000 0.000061035156 2 ^ 14
0x13 0x37800000 0.000015258789 2 ^ -16
0x14 0x3EBC5AB2 0.36787945 e ^ -1
0x15 0x3F3504F3 0.70710677 sqrt(2) ^ -1
0x16 0x3FC90FDB 1.5707964 pi / 2
0x17 0x3EC90FDB 0.39269909 pi / 8
0x18 0x41490FDB 12.566371 pi * 4
0x19 0x00000000 0.0
0x1A 0x38000000 0.000030517578 1 / 2 ^ 15
0x1B 0x38000100 0.000030518509 1 / (2 ^ 15 - 1)
0x1C 0x37B60B61 0.000021701389 cos(0.5) - taylor 4th term
0x1D 0x3B2AAAAB 0.0026041667 cos(0.5) - taylor 3rd term
0x1E 0x3E000000 0.125 cos(0.5) - taylor 2nd term
0x1F 0x3F800000 1.0 cos(0.5) - taylor 1st term
0x20 0x3C000000 0.0078125 1 / 2 ^ 7
0x21 0x00000000 0.0
0x22 0x3C003C00 0.0078268051
0x23 0x4C004800 3.362816e7
0x24 0x5C005800 1.4450222e17
0x25 0x6C006800 6.2093452e26
0x26 0x00007800 4.3048e-41
0x27 0x2C003000 1.8216539e-12
0x28 0x1C002000 4.2393006e-22
0x29 0x0C001000 9.8655761e-32
0x2A 0x00000000 0.0
0x2B 0x00000000 0.0
0x2C 0x00000000 0.0
0x2D 0x36483A48 0.0000029836247
0x2E 0x00004E48 2.8082e-41
0x2F 0x00000000 0.0
0x30 0x19550C44 1.1014319e-23
0x31 0x3C003800 0.0078258514
0x32 0x00000000 0.0
0x33 0x00000000 0.0
0x34 0x00000000 0.0
0x35 0x00000000 0.0
0x36 0x00000000 0.0
0x37 0x00000000 0.0
0x38 0x00000000 0.0
0x39 0x00000000 0.0
0x3A 0x00000000 0.0
0x3B 0x00000000 0.0
0x3C 0x00000000 0.0
0x3D 0x00000000 0.0
0x3E 0x00000000 0.0
0x3F 0x00000000 0.0
16 bit mode

Table for 16 bit mode does not have accurate values.

bank 1, bank 2, bank 3 is used for channel 1, channel 2, channel 3 of op0

for other operands - only bank 0 is used for each channel

f16 mode - bank 0:

operand value comment
0x00 0.0
0x01 0.0
0x02 0.0
0x03 0.0
0x04 0.0
0x05 0.0
0x06 0.0
0x07 0.0
0x08 0.0
0x09 0.0
0x0A 0.0
0x0B 0.0
0x0C 0.0
0x0D 0.0
0x0E 0.0
0x0F 0.0
0x10 0.0
0x11 0.0
0x12 0.0
0x13 0.0
0x14 -35456.0
0x15 7.5519e-05
0x16 0.00047946
0x17 0.00047946
0x18 0.00047946
0x19 0.00047946
0x1A 0.0
0x1B 7.6294e-06
0x1C 0.00030541
0x1D -0.0001384
0x1E -0.052094
0x1F 0.0
0x20 0.0
0x21 0.0
0x22 1.0
0x23 2.0 2 ^ 1
0x24 32.0 2 ^ 5
0x25 512.0 2 ^ 9
0x26 8192.0 2 ^ 13
0x27 0.5 2 ^ -1
0x28 0.03125 2 ^ -5
0x29 0.0019531 2 ^ -9
0x2A 0.00012207 2 ^ - 13
0x2B 2.7168 e
0x2C 1.4141 sqrt(2)
0x2D 3.1406 pi
0x2E 6.2813 pi * 2
0x2F 0.0
0x30 0.0
0x31 0.020828
0x32 0.0
0x33 0.0
0x34 0.0
0x35 0.0
0x36 0.0
0x37 0.0
0x38 -1.#QNB 0xFFFFFFFF
0x39 -1.#QNB 0xFFFFFFFF
0x3A -1.#QNB 0xFFFFFFFF
0x3B -1.#QNB 0xFFFFFFFF
0x3C 1.#QNB 0x7FFF7FFF
0x3D 1.#QNB 0x7FFF7FFF
0x3E 1.#QNB 0x7FFF7FFF
0x3F 1.#QNB 0x7FFF7FFF

f16 mode - bank 1:

operand value comment
0x00 0.0
0x01 0.0
0x02 1.875
0x03 1.875
0x04 2.0
0x05 2.5
0x06 3.0
0x07 3.5
0x08 4.0
0x09 5.0
0x0A 6.0
0x0B 7.0
0x0C 1.75
0x0D 1.5
0x0E 1.25
0x0F 1.0
0x10 0.875
0x11 0.75
0x12 0.625
0x13 0.5
0x14 2.0879
0x15 1.9268
0x16 2.1426
0x17 1.8213
0x18 2.3926
0x19 2.8926
0x1A 0.46875
0x1B 0.46875
0x1C 0.36328
0x1D 0.69141
0x1E 1.166
0x1F 1.75
0x20 0.0
0x21 0.0
0x22 1.0
0x23 4.0
0x24 64.0
0x25 1024.0
0x26 16384.0
0x27 0.25
0x28 0.015625
0x29 0.00097656
0x2A 6.1035e-05
0x2B 0.36768
0x2C 0.70703
0x2D 1.5703
0x2E 12.563
0x2F 0.0
0x30 0.0
0x31 0.125
0x32 0.0
0x33 0.0
0x34 0.0
0x35 0.0
0x36 0.0
0x37 0.0
0x38 -1.#QNB 0xFFFFFFFF
0x39 -1.#QNB 0xFFFFFFFF
0x3A -1.#QNB 0xFFFFFFFF
0x3B -1.#QNB 0xFFFFFFFF
0x3C 1.#QNB 0x7FFF7FFF
0x3D 1.#QNB 0x7FFF7FFF
0x3E 1.#QNB 0x7FFF7FFF
0x3F 1.#QNB 0x7FFF7FFF

f16 mode - bank 2:

operand value comment
0x00 0.0
0x01 0.0
0x02 0.0
0x03 0.0
0x04 0.0
0x05 0.0
0x06 0.0
0x07 0.0
0x08 0.0
0x09 0.0
0x0A 0.0
0x0B 0.0
0x0C 0.0
0x0D 0.0
0x0E 0.0
0x0F 0.0
0x10 0.0
0x11 0.0
0x12 0.0
0x13 0.0
0x14 214.25
0x15 7.5519e-05
0x16 0.00047946
0x17 0.00047946
0x18 0.00047946
0x19 0.0
0x1A 0.0
0x1B 1.5259e-05
0x1C 0.00022519
0x1D -0.052094
0x1E 0.0
0x1F 0.0
0x20 0.0
0x21 0.0
0x22 1.0
0x23 8.0
0x24 128.0
0x25 2048.0
0x26 32768.0
0x27 0.125
0x28 0.0078125
0x29 0.00048828
0x2A 0.0
0x2B 0.0
0x2C 0.0
0x2D 0.78516
0x2E 25.125
0x2F 0.0
0x30 0.00026035
0x31 0.5
0x32 0.0
0x33 0.0
0x34 0.0
0x35 0.0
0x36 0.0
0x37 0.0
0x38 0.0
0x39 0.0
0x3A 0.0
0x3B 0.0
0x3C 0.0
0x3D 0.0
0x3E 0.0
0x3F 0.0

f16 mode - bank 3:

operand value comment
0x00 0.0
0x01 1.875
0x02 0.0
0x03 1.875
0x04 2.25
0x05 2.75
0x06 3.25
0x07 3.75
0x08 4.5
0x09 5.5
0x0A 6.5
0x0B 7.5
0x0C 1.625
0x0D 1.375
0x0E 1.125
0x0F 0.9375
0x10 0.8125
0x11 0.6875
0x12 0.5625
0x13 0.46875
0x14 1.6836
0x15 1.8018
0x16 1.9463
0x17 1.6963
0x18 2.6426
0x19 0.0
0x1A 0.5
0x1B 0.5
0x1C 0.48193
0x1D 0.89551
0x1E 1.5
0x1F 1.875
0x20 1.0
0x21 0.0
0x22 1.0
0x23 16.0
0x24 256.0
0x25 4096.0
0x26 0.0
0x27 0.0625
0x28 0.0039063
0x29 0.00024414
0x2A 0.0
0x2B 0.0
0x2C 0.0
0x2D 0.39258
0x2E 0.0
0x2F 0.0
0x30 0.0026035
0x31 1.0
0x32 0.0
0x33 0.0
0x34 0.0
0x35 0.0
0x36 0.0
0x37 0.0
0x38 0.0
0x39 0.0
0x3A 0.0
0x3B 0.0
0x3C 0.0
0x3D 0.0
0x3E 0.0
0x3F 0.0

Swizzles

Swizzle notation

There are 2 notations:

  • text notation
  • constant notation

When some of channels have constants - text notation is used

mul.f32 r0.xyzw, r0.h1xx, r0.xxxx

When all channels have constants - constant notation is used

mul.f32 r0.xyzw, {0.5, 1, 1, 0.5}, r0.xxxx

When channel is masked in text notation it is marked as -

mul.f32 r0.-y-w, r0.-x-x, r0.-x-x

When channel is masked in constant notation it is replaced with zero

mul.f32 r0.-y-w, {0, 1, 0, 0.5}, r0.-x-x

Register Swizzle RSWZ2

This encoding uses 2 bits to encode the swizzle.

Usually combinations are additinally controlled by 1 or more bits called swz_alt_op

This type of swizzling does not allow precise control on each channel as opposed to RSWZ3

Usually there is a predefined table of swizzles.

Swizzle expression is built as:

<reg><index>.<swizzle>

Example:

r22.x

Register Swizzle RSWZ3

This encoding uses 3 bits to encode the mask.

channel is encoded as:

2 1 0 text notation constant notation
0 0 0 x x
0 0 1 y y
0 1 0 z z
0 1 1 w w
1 0 0 0 0.0
1 0 1 1 1.0
1 1 0 2 2.0
1 1 1 h 0.5

swizzle expression is built as:

<reg><index>.<swizzle>

Example:

r22.x

Modifier and dest data format

At the moment it is not known which of the data format fields is dest and which is source.

This is the reason why term modifier is mixed with term dest data format.

0x00000000 - 0x08000000

Instructions

mad

Encoding

Higher 4 bytes

7 6 5 4 3 2 1 0
opcode1 data_format predicate
0 0 0 0 0
7 6 5 4 3 2 1 0
swz_alt_op1 alt_opt0 abs_op1 alt_opt2 alt_opt3
x x x
7 6 5 4 3 2 1 0
swz_alt_op3 op3_swz swz_alt_op2 swz_mask16 swz_mask32
x x
7 6 5 4 3 2 1 0
swz_en abs_op2 neg_op2 abs_op3 neg_op3 opt1 opt0

Lower 4 bytes

7 6 5 4 3 2 1 0
opt2 opt3 op0
7 6 5 4 3 2 1 0
op0 op2_swz op1_swz op1
7 6 5 4 3 2 1 0
op1 op2
7 6 5 4 3 2 1 0
op2 op3

Notes

  • x bits do not affect instruction or operands. might affect something else?
  • what do index<N> mean. are these registers or something?
  • looks like there is functionality to switch sign of index expression
  • probably can move swizzle masking to generic section? if other instructions use same encodings.

Fields - instruction

data_format:

2 value
0 f32
1 f16

predicate:

1 0 value
0 0
0 1 p0
1 0 !p0
1 1 Pn

Fields - operands

Constants

Specific operand may be used as float constant. This can be achieved with following groups of bits:

  • alt_opt0, opt0, op0
  • alt_opt2, opt2, op2
  • alt_opt3, opt3, op3

Float constants can only be used when swizzling is enabled for particular operand. Consider checking sections Swizzles_f32 and Swizzles_f16.

Constants are taken from tables Constants.

Constants differ between 32 and 16 bit mode.

Swizzle masking

Masking is controled by control bits:

  • control bits: swz_en, swz_mask32, swz_mask16

Each channel can be masked with control bits. Combinations of control bits produce the following masking table.

Encoding used in masking table:

value meaning
0 channel not selected
1 channel selected
x channel masked

Masking table 32 bit mode:

swz_mask32 swz_en ch0 ch1
0 0 0 0
0 1 1 0
1 0 x 1
1 1 1 1

Masking table 16 bit mode:

swz_mask16 swz_en ch0 ch1 ch2 ch3
0 0 0 0 0 0
0 1 1 1 0 0
1 0 x x 1 1
1 1 1 1 1 1

Swizzles f32

Swizzles of operand 1, operand 2 and operand 3 can not be precisely controlled and have predefined combinations.

Swizzles are controlled with bits:

  • swizzle fields: op1_swz, op2_swz, op3_swz
  • control bits: swz_alt_op1, swz_alt_op2, swz_alt_op3

Swizzles of operand 0 can not be controlled.

operand 0 operand 1 operand 2 operand 3
value
xy
swz_alt_op1 op1_swz value
0 0 0 xx
0 0 1 yy
0 1 0 zz
0 1 1 ww
1 0 0 xy
1 0 1 yz
1 1 0 xy
1 1 1 zw
swz_alt_op2 op2_swz value
0 0 0 xx
0 0 1 yy
0 1 0 zz
0 1 1 ww
1 0 0 xy
1 0 1 xy
1 1 0 yy
1 1 1 wy
swz_alt_op3 op3_swz value
0 0 0 xx
0 0 1 yy
0 1 0 zz
0 1 1 ww
1 0 0 xy
1 0 1 xz
1 1 0 xx
1 1 1 xy

Swizzles f16

Swizzles of operand 1, operand 2 and operand 3 can not be precisely controlled and have predefined combinations.

Swizzles are controlled with bits:

  • swizzle fields: op1_swz, op2_swz, op3_swz
  • control bits: swz_alt_op1, swz_alt_op2, swz_alt_op3

Swizzles of operand 0 can not be controlled.

operand 0 operand 1 operand 2 operand 3
value
xyzw
swz_alt_op1 op1_swz value
0 0 0 xxxx
0 0 1 yyyy
0 1 0 zzzz
0 1 1 wwww
1 0 0 xyzw
1 0 1 yzxw
1 1 0 xyww
1 1 1 zwxy
swz_alt_op2 op2_swz value
0 0 0 xxxx
0 0 1 yyyy
0 1 0 zzzz
0 1 1 wwww
1 0 0 xyzw
1 0 1 xyyz
1 1 0 yyww
1 1 1 wyzw
swz_alt_op3 op3_swz value
0 0 0 xxxx
0 0 1 yyyy
0 1 0 zzzz
0 1 1 wwww
1 0 0 xyzw
1 0 1 xzww
1 1 0 xxyz
1 1 1 xyzz

Examples

mad.f32 r0, r0, r0, r0
 
mad.f16 r0, r0, r0, r0

0x08000000 - 0x10000000

Instructions

mul.f32, add.f32, frc.f32, dsx.f32, dsy.f32, min.f32, max.f32, dot.f32

Encoding

Higher 4 bytes

7 6 5 4 3 2 1 0
opcode1 predicate
0 0 0 0 1
7 6 5 4 3 2 1 0
op1_swz_c3x alt_opt0 op1_swz_c30 alt_opt1 alt_opt2
x x
7 6 5 4 3 2 1 0
swz_alt_op2 op2_swz swz_mask3 swz_mask2 swz_mask1
x
7 6 5 4 3 2 1 0
swz_en abs_op1 neg_op1 abs_op2 op1_swz_c2x opt0

Lower 4 bytes

7 6 5 4 3 2 1 0
opt1 opt2 op0
7 6 5 4 3 2 1 0
op0 op1_swz_c20 op1_swz_c1 op1_swz_c0
7 6 5 4 3 2 1 0
op1_swz_c0 opcode2 op1
7 6 5 4 3 2 1 0
op1 op2

Notes

Fields - instruction

predicate:

2 1 0 value
0 0 0
0 0 1 p0
0 1 0 p1
0 1 1 p2
1 0 0 !p0
1 0 1 !p1
1 1 0 !p2
1 1 1 Pn

opcode2:

6 5 4 value
0 0 0 mul.f32
0 0 1 add.f32
0 1 0 frc.f32
0 1 1 dsx.f32
1 0 0 dsy.f32
1 0 1 min.f32
1 1 0 max.f32
1 1 1 dot.f32

Fields - operands

Constants

Specific operand may be used as float constant. This can be achieved with following groups of bits:

  • alt_opt0, opt0, op0
  • alt_opt1, opt1, op1
  • alt_opt2, opt2, op2

Float constants can only be used when swizzling is enabled for particular operand. Consider checking sections Swizzle_masking.

Constants are taken from tables Constants.

Constants correspond to table for 32 bit mode.

Swizzle masking

Masking is controled by control bits:

  • control bits: swz_en, swz_mask1, swz_mask2, swz_mask3

Each channel can be masked with control bits. Combinations of control bits produce the following masking table.

dot.f32 instruction has explicit swizzling in operand 1 and operand 2 so masking does not apply to these operands. number of channels is always 4.

Encoding used in masking table:

value meaning
0 channel not selected
1 channel selected
x channel masked

Masking table:

swz_mask3 swz_mask2 swz_mask1 swz_en ch0 ch1 ch2 ch3
0 0 0 0 0 0 0 0
0 0 0 1 1 0 0 0
0 0 1 0 x 1 0 0
0 0 1 1 1 1 0 0
0 1 0 0 x x 1 0
0 1 0 1 1 x 1 0
0 1 1 0 x 1 1 0
0 1 1 1 1 1 1 0
1 0 0 0 x x x 1
1 0 0 1 1 x x 1
1 0 1 0 x 1 x 1
1 0 1 1 1 1 x 1
1 1 0 0 x x 1 1
1 1 0 1 1 x 1 1
1 1 1 0 x 1 1 1
1 1 1 1 1 1 1 1

Swizzles - operand 0

Swizzles of operand 0 can not be controled and have predefined combinations described below:

value
xyzw

Each channel can be masked with control bits. Masking is described in Swizzle_masking.

  • swz_en, swz_mask1, swz_mask2, swz_mask3

Swizzles - operand 1

Each channel of operand 1 can be precisely controlled with swizzle fields encoded as RSWZ3.

  • op1_swz_c0, op1_swz_c1, op1_swz_c20, op1_swz_c2x, op1_swz_c30, op1_swz_c3x

Each channel can be masked with control bits. Masking is described in Swizzle_masking.

  • swz_en, swz_mask1, swz_mask2, swz_mask3

masking does not apply for dot.f32 instruction

Swizzles - operand 2

Swizzles of operand 2 can not be precisely controlled and have predefined combinations described below and controlled by swizzle fields:

  • op2_swz, swz_alt_op2

Each channel can be masked with control bits. Masking is described in Swizzle_masking.

  • swz_en, swz_mask1, swz_mask2, swz_mask3

masking does not apply for dot.f32 instruction

swz_alt_op2 op2_swz value
0 0 0 0 xxxx
0 0 0 1 yyyy
0 0 1 0 zzzz
0 0 1 1 wwww
0 1 0 0 xyzw
0 1 0 1 yzww
0 1 1 0 xyzz
0 1 1 1 xxyz
1 0 0 0 xyxy
1 0 0 1 xywz
1 0 1 0 zxyw
1 0 1 1 zwzw
1 1 0 0 yzxz
1 1 0 1 xxyy
1 1 1 0 xzww
1 1 1 1 xyz1

Examples

mul.f32 r0, r0, r0
add.f32 r0, r0, r0
frc.f32 r0, r0, r0
dsx.f32 r0, r0, r0
dsy.f32 r0, r0, r0
min.f32 r0, r0, r0
max.f32 r0, r0, r0
dot.f32 r0, r0.xxxx, r0.xxxx

0x10000000 - 0x18000000

Instructions

mul.f16, add.f16, frc.f16, dsx.f16, dsy.f16, min.f16, max.f16, dot.f16

Encoding

Higher 4 bytes

7 6 5 4 3 2 1 0
opcode1 predicate
0 0 0 1 0
7 6 5 4 3 2 1 0
op1_swz_c3x alt_opt0 op1_swz_c30 alt_opt1 alt_opt2
x x
7 6 5 4 3 2 1 0
swz_alt_op2 op2_swz swz_mask3 swz_mask2 swz_mask1
x
7 6 5 4 3 2 1 0
swz_en abs_op1 neg_op1 abs_op2 op1_swz_c2x opt0

Lower 4 bytes

7 6 5 4 3 2 1 0
opt1 opt2 op0
7 6 5 4 3 2 1 0
op0 op1_swz_c20 op1_swz_c1 op1_swz_c0
7 6 5 4 3 2 1 0
op1_swz_c0 opcode2 op1
7 6 5 4 3 2 1 0
op1 op2

Notes

Fields - instruction

predicate:

2 1 0 value
0 0 0
0 0 1 p0
0 1 0 p1
0 1 1 p2
1 0 0 !p0
1 0 1 !p1
1 1 0 !p2
1 1 1 Pn

opcode2:

6 5 4 value
0 0 0 mul.f16
0 0 1 add.f16
0 1 0 frc.f16
0 1 1 dsx.f16
1 0 0 dsy.f16
1 0 1 min.f16
1 1 0 max.f16
1 1 1 dot.f16

Fields - operands

Constants

Specific operand may be used as float constant. This can be achieved with following groups of bits:

  • alt_opt0, opt0, op0
  • alt_opt1, opt1, op1
  • alt_opt2, opt2, op2

Float constants can only be used when swizzling is enabled for particular operand. Consider checking sections Swizzle_masking.

Constants are taken from tables Constants.

Constants correspond to table for 16 bit mode.

Swizzle masking

Masking is controled by control bits:

  • control bits: swz_en, swz_mask1, swz_mask2, swz_mask3

Each channel can be masked with control bits. Combinations of control bits produce the following masking table.

dot.f16 instruction has explicit swizzling in operand 1 and operand 2 so masking does not apply to these operands. number of channels is always 4.

Encoding used in masking table:

value meaning
0 channel not selected
1 channel selected
x channel masked

Masking table:

swz_mask3 swz_mask2 swz_mask1 swz_en ch0 ch1 ch2 ch3
0 0 0 0 0 0 0 0
0 0 0 1 1 0 0 0
0 0 1 0 x 1 0 0
0 0 1 1 1 1 0 0
0 1 0 0 x x 1 0
0 1 0 1 1 x 1 0
0 1 1 0 x 1 1 0
0 1 1 1 1 1 1 0
1 0 0 0 x x x 1
1 0 0 1 1 x x 1
1 0 1 0 x 1 x 1
1 0 1 1 1 1 x 1
1 1 0 0 x x 1 1
1 1 0 1 1 x 1 1
1 1 1 0 x 1 1 1
1 1 1 1 1 1 1 1

Swizzles - operand 0

Swizzles of operand 0 can not be controled and have predefined combinations described below:

value
xyzw

Each channel can be masked with control bits. Masking is described in Swizzle_masking.

  • swz_en, swz_mask1, swz_mask2, swz_mask3

Swizzles - operand 1

Each channel of operand 1 can be precisely controlled with swizzle fields encoded as RSWZ3.

  • op1_swz_c0, op1_swz_c1, op1_swz_c20, op1_swz_c2x, op1_swz_c30, op1_swz_c3x

Each channel can be masked with control bits. Masking is described in Swizzle_masking.

  • swz_en, swz_mask1, swz_mask2, swz_mask3

masking does not apply for dot.f16 instruction

Swizzles - operand 2

Swizzles of operand 2 can not be precisely controlled and have predefined combinations described below and controlled by swizzle fields:

  • op2_swz, swz_alt_op2

Each channel can be masked with control bits. Masking is described in Swizzle_masking.

  • swz_en, swz_mask1, swz_mask2, swz_mask3

masking does not apply for dot.f16 instruction

swz_alt_op2 op2_swz value
0 0 0 0 xxxx
0 0 0 1 yyyy
0 0 1 0 zzzz
0 0 1 1 wwww
0 1 0 0 xyzw
0 1 0 1 yzww
0 1 1 0 xyzz
0 1 1 1 xxyz
1 0 0 0 xyxy
1 0 0 1 xywz
1 0 1 0 zxyw
1 0 1 1 zwzw
1 1 0 0 yzxz
1 1 0 1 xxyy
1 1 1 0 xzww
1 1 1 1 xyz1

Examples

mul.f16 r0, r0, r0
add.f16 r0, r0, r0
frc.f16 r0, r0, r0
dsx.f16 r0, r0, r0
dsy.f16 r0, r0, r0
min.f16 r0, r0, r0
max.f16 r0, r0, r0
dot.f16 r0, r0.xxxx, r0.xxxx

0x18000000 - 0x20000000

Notes

  • Any way to use constants for op2, op3? Apart from swizzle constants.
  • 2 Strange fields for both dot.f32 and mad.f32.

dot.f32

Instructions

dot.f32

Encoding

Higher 4 bytes

7 6 5 4 3 2 1 0
opcode1 predicate
0 0 0 1 1
7 6 5 4 3 2 1 0
opcode2 c3_en alt_opt0 alt_opt1
x x 0 x x
7 6 5 4 3 2 1 0
abs_op2 swz_en_strange1 swz_en_strange0 swz_mask3 swz_mask2 swz_mask1
x x
7 6 5 4 3 2 1 0
swz_en neg_op1 abs_op1 opt0
x x x

Lower 4 bytes

7 6 5 4 3 2 1 0
opt1 op2i op0
7 6 5 4 3 2 1 0
op0 swz_alt_op2 op2_swz op1_swz_c3
7 6 5 4 3 2 1 0
op1_swz_c3 op1_swz_c2 op1_swz_c1 op1_swz_c0
7 6 5 4 3 2 1 0
op1_swz_c0 op1
Fields - instruction

opcode2:

5 value
0 dot.f32
1 mad.f32

predicate:

2 1 0 value
0 0 0
0 0 1 p0
0 1 0 p1
0 1 1 p2
1 0 0 !p0
1 0 1 !p1
1 1 0 !p2
1 1 1 Pn
Fields - operands
Constants

Specific operand may be used as float constant. This can be achieved with following groups of bits:

  • alt_opt0, opt0, op0
  • alt_opt1, opt1, op1

Float constants can only be used when swizzling is enabled for particular operand. Consider checking sections Swizzle_masking.

Constants are taken from tables Constants.

Constants correspond to table for 32 bit mode.

Swizzle masking

Masking is controled by control bits:

  • control bits: swz_en, swz_mask1, swz_mask2, swz_mask3

Each channel can be masked with control bits. Combinations of control bits produce the following masking table.

dot.f32 instruction has explicit swizzling in operand 1 and operand 2 so masking does not apply to these operands. number of channels is 3 or 4 depending on c3_en.

Encoding used in masking table:

value meaning
0 channel not selected
1 channel selected
x channel masked

Masking table operand 0:

swz_mask3 swz_mask2 swz_mask1 swz_en ch0 ch1 ch2 ch3
0 0 0 0 0 0 0 0
0 0 0 1 1 0 0 0
0 0 1 0 x 1 0 0
0 0 1 1 1 1 0 0
0 1 0 0 x x 1 0
0 1 0 1 1 x 1 0
0 1 1 0 x 1 1 0
0 1 1 1 1 1 1 0
1 0 0 0 x x x 1
1 0 0 1 1 x x 1
1 0 1 0 x 1 x 1
1 0 1 1 1 1 x 1
1 1 0 0 x x 1 1
1 1 0 1 1 x 1 1
1 1 1 0 x 1 1 1
1 1 1 1 1 1 1 1
Swizzles - operand 0

Swizzles of operand 0 can not be controled and have predefined combinations described below:

value
xyzw

Each channel can be masked with control bits. Masking is described in Swizzle_masking.

  • swz_en, swz_mask1, swz_mask2, swz_mask3
Swizzles - operand 1

Each channel of operand 1 can be precisely controlled with swizzle fields encoded as RSWZ3.

  • op1_swz_c0, op1_swz_c1, op1_swz_c2, op1_swz_c3

Channel 3 can be enabled with bit:

  • c3_en
Swizzles - operand 2

Swizzles of operand 2 can not be precisely controlled and have predefined combinations described below and controlled by swizzle fields:

  • op2_swz, swz_alt_op2

Channel 3 can be enabled with bit:

  • c3_en

3 channels:

swz_alt_op2 op2_swz value
0 0 0 0 xxx
0 0 0 1 yyy
0 0 1 0 zzz
0 0 1 1 www
0 1 0 0 xyz
0 1 0 1 yzw
0 1 1 0 xxy
0 1 1 1 xyx
1 0 0 0 yyx
1 0 0 1 yyz
1 0 1 0 zxy
1 0 1 1 xzy
1 1 0 0 yzx
1 1 0 1 zyx
1 1 1 0 zzy
1 1 1 1 xy1

4 channels:

swz_alt_op2 op2_swz value
0 0 0 0 xxxx
0 0 0 1 yyyy
0 0 1 0 zzzz
0 0 1 1 wwww
0 1 0 0 xyzw
0 1 0 1 yzww
0 1 1 0 xyzz
0 1 1 1 xxyz
1 0 0 0 xyxy
1 0 0 1 xywz
1 0 1 0 zxyw
1 0 1 1 zwzw
1 1 0 0 yzxz
1 1 0 1 xxyy
1 1 1 0 xzww
1 1 1 1 xyz1
Examples
dot.f32 r0, r0.xxx, i0.xxx

mad.f32

Instructions

mad.f32

Encoding

Higher 4 bytes

7 6 5 4 3 2 1 0
opcode1 predicate
0 0 0 1 1
7 6 5 4 3 2 1 0
swz_alt_op3_2 opcode2 alt_opt0 alt_opt1
x 1 x x x
7 6 5 4 3 2 1 0
abs_op2 op0_strange1 op0_strange0 swz_mask3 swz_mask2 swz_mask1
x
7 6 5 4 3 2 1 0
swz_en neg_op1 abs_op1 neg_op3 abs_op3 swz_alt_op2_2 opt0

Lower 4 bytes

7 6 5 4 3 2 1 0
opt1 op2i op0
7 6 5 4 3 2 1 0
op0 swz_alt_op2_x op2_swz swz_alt_op3_x
7 6 5 4 3 2 1 0
op3_swz op3i swz_alt_op1
x
7 6 5 4 3 2 1 0
op1_swz op1
Fields - instruction

opcode2:

5 value
0 dot.f32
1 mad.f32

predicate:

2 1 0 value
0 0 0
0 0 1 p0
0 1 0 p1
0 1 1 p2
1 0 0 !p0
1 0 1 !p1
1 1 0 !p2
1 1 1 Pn
Fields - operands
Swizzle masking

Masking is controled by control bits:

  • control bits: swz_en, swz_mask1, swz_mask2, swz_mask3

Each channel can be masked with control bits. Combinations of control bits produce the following masking table.

Encoding used in masking table:

value meaning
0 channel not selected
1 channel selected
x channel masked

Masking table:

swz_mask3 swz_mask2 swz_mask1 swz_en ch0 ch1 ch2 ch3
0 0 0 0 0 0 0 0
0 0 0 1 1 0 0 0
0 0 1 0 x 1 0 0
0 0 1 1 1 1 0 0
0 1 0 0 x x 1 0
0 1 0 1 1 x 1 0
0 1 1 0 x 1 1 0
0 1 1 1 1 1 1 0
1 0 0 0 x x x 1
1 0 0 1 1 x x 1
1 0 1 0 x 1 x 1
1 0 1 1 1 1 x 1
1 1 0 0 x x 1 1
1 1 0 1 1 x 1 1
1 1 1 0 x 1 1 1
1 1 1 1 1 1 1 1
Swizzles - operand 0

Swizzles of operand 0 can not be controled and have predefined combinations described below:

value
xyzw

Each channel can be masked with control bits. Masking is described in Swizzle masking.

  • swz_en, swz_mask1, swz_mask2, swz_mask3
Swizzles - operand 1

Swizzles of operand 1 can not be precisely controlled and have predefined combinations described below and controlled by swizzle fields:

  • op1_swz, swz_alt_op1

Each channel can be masked with control bits. Masking is described in Swizzle masking.

  • swz_en, swz_mask1, swz_mask2, swz_mask3
swz_alt_op1 op1_swz value
0 0 0 0 0 xxxx
0 0 0 0 1 yyyx
0 0 0 1 0 zzzx
0 0 0 1 1 wwwx
0 0 1 0 0 xyzx
0 0 1 0 1 yzwx
0 0 1 1 0 xxyx
0 0 1 1 1 xyxx
0 1 0 0 0 yyxx
0 1 0 0 1 yyzx
0 1 0 1 0 zxyx
0 1 0 1 1 xzyx
0 1 1 0 0 yzxx
0 1 1 0 1 zyxx
0 1 1 1 0 zzyx
0 1 1 1 1 xy1x
1 0 0 0 0 xyyx
1 0 0 0 1 yxyx
1 0 0 1 0 xxzx
1 0 0 1 1 yxxx
1 0 1 0 0 xy0x
1 0 1 0 1 x10x
1 0 1 1 0 000x
1 0 1 1 1 111x
1 1 0 0 0 hhhx
1 1 0 0 1 222x
1 1 0 1 0 x00x
1 1 0 1 1 {0.5, 0.5, 0.5, 0.5}
1 1 1 0 0 {0.5, 0.5, 0.5, 0.5}
1 1 1 0 1 {0.5, 0.5, 0.5, 0.5}
1 1 1 1 0 {0.5, 0.5, 0.5, 0.5}
1 1 1 1 1 {0.5, 0.5, 0.5, 0.5}
Swizzles - operand 2

Swizzles of operand 2 can not be precisely controlled and have predefined combinations described below and controlled by swizzle fields:

  • op2_swz, swz_alt_op2_x, swz_alt_op2_2

Each channel can be masked with control bits. Masking is described in Swizzle masking.

  • swz_en, swz_mask1, swz_mask2, swz_mask3
swz_alt_op2_2 swz_alt_op2_x op2_swz value
0 0 0 0 0 xxxx
0 0 0 0 1 yyyx
0 0 0 1 0 zzzx
0 0 0 1 1 wwwx
0 0 1 0 0 xyzx
0 0 1 0 1 yzwx
0 0 1 1 0 xxyx
0 0 1 1 1 xyxx
0 1 0 0 0 yyxx
0 1 0 0 1 yyzx
0 1 0 1 0 zxyx
0 1 0 1 1 xzyx
0 1 1 0 0 yzxx
0 1 1 0 1 zyxx
0 1 1 1 0 zzyx
0 1 1 1 1 xy1x
1 0 0 0 0 xyyx
1 0 0 0 1 yxyx
1 0 0 1 0 xxzx
1 0 0 1 1 yxxx
1 0 1 0 0 xy0x
1 0 1 0 1 x10x
1 0 1 1 0 000x
1 0 1 1 1 111x
1 1 0 0 0 hhhx
1 1 0 0 1 222x
1 1 0 1 0 x00x
1 1 0 1 1 {0.5, 0.5, 0.5, 0.5}
1 1 1 0 0 {0.5, 0.5, 0.5, 0.5}
1 1 1 0 1 {0.5, 0.5, 0.5, 0.5}
1 1 1 1 0 {0.5, 0.5, 0.5, 0.5}
1 1 1 1 1 {0.5, 0.5, 0.5, 0.5}
Swizzles - operand 3

Swizzles of operand 3 can not be precisely controlled and have predefined combinations described below and controlled by swizzle fields:

  • op3_swz, swz_alt_op3_x, swz_alt_op3_2

Each channel can be masked with control bits. Masking is described in Swizzle masking.

  • swz_en, swz_mask1, swz_mask2, swz_mask3
swz_alt_op3_2 swz_alt_op3_x op3_swz value
0 0 0 0 0 xxxx
0 0 0 0 1 yyyx
0 0 0 1 0 zzzx
0 0 0 1 1 wwwx
0 0 1 0 0 xyzx
0 0 1 0 1 yzwx
0 0 1 1 0 xxyx
0 0 1 1 1 xyxx
0 1 0 0 0 yyxx
0 1 0 0 1 yyzx
0 1 0 1 0 zxyx
0 1 0 1 1 xzyx
0 1 1 0 0 yzxx
0 1 1 0 1 zyxx
0 1 1 1 0 zzyx
0 1 1 1 1 xy1x
1 0 0 0 0 xyyx
1 0 0 0 1 yxyx
1 0 0 1 0 xxzx
1 0 0 1 1 yxxx
1 0 1 0 0 xy0x
1 0 1 0 1 x10x
1 0 1 1 0 000x
1 0 1 1 1 111x
1 1 0 0 0 hhhx
1 1 0 0 1 222x
1 1 0 1 0 x00x
1 1 0 1 1 {0.5, 0.5, 0.5, 0.5}
1 1 1 0 0 {0.5, 0.5, 0.5, 0.5}
1 1 1 0 1 {0.5, 0.5, 0.5, 0.5}
1 1 1 1 0 {0.5, 0.5, 0.5, 0.5}
1 1 1 1 1 {0.5, 0.5, 0.5, 0.5}
Examples
mad.f32 r0, r0, i0, i0

0x20000000 - 0x28000000

Instructions

mad, dot, add, mul, subfl, exp, mov, log, rsq, rcp

Encoding

There is total of 10 instructions with 28 variations in this group. However instruction encoding is quite complex and is controlled by following fields:

  • op_sel2 - controls opcode page that is selected. there are 2 pages.
  • opcode2 - controls opcodes in the page. there are 8 opcodes.
  • gr_sel - controls validity of instructions
  • op_sel1 - controls explicit swizzling
  • opcode3 - controls validity of instructions

Higher 4 bytes

7 6 5 4 3 2 1 0
opcode1
0 0 1 0 0 x x x
7 6 5 4 3 2 1 0
op_sel2
x x x x x x x
7 6 5 4 3 2 1 0
opcode2 gr_sel op_sel1
x x x
7 6 5 4 3 2 1 0
x x x x x x x x

Lower 4 bytes

7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
opcode3
x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x
Allowed Instruction Encodings

Allowed encodings are defined by these tables.

opcode2, op_sel2 are a composite key for table Allowed Instructions

Join with this table to fill in instruction gaps.

gr_sel = 0

opcode2 op_sel2 gr_sel opcode3 instruction
0 x 0 0 0 x invalid
0 x 0 0 1 0 instruction
0 x 0 0 1 1 invalid
0 x 0 1 0 0 invalid
0 x 0 1 0 1 instruction
0 x 0 1 1 x instruction
1 x 0 0 0 0 invalid
1 x 0 0 0 1 instruction
1 x 0 0 1 x instruction
1 x 0 1 x x instruction
2 x 0 0 0 0 invalid
2 x 0 0 0 1 instruction
2 x 0 0 1 x instruction
2 x 0 1 x x instruction
3 x 0 0 0 0 invalid
3 x 0 0 0 1 instruction
3 x 0 0 1 x instruction
3 x 0 1 x x instruction
4 x 0 0 0 0 invalid
4 x 0 0 0 1 instruction
4 x 0 0 1 x instruction
4 x 0 1 x x instruction
5 x 0 0 0 0 invalid
5 x 0 0 0 1 instruction
5 x 0 0 1 x instruction
5 x 0 1 x x instruction
6 x 0 0 0 0 invalid
6 x 0 0 0 1 instruction
6 x 0 0 1 x instruction
6 x 0 1 x x instruction
7 x 0 0 0 0 invalid
7 x 0 0 0 1 instruction
7 x 0 0 1 x instruction
7 x 0 1 x x instruction

gr_sel = 1, op_sel2 = 0

opcode2 op_sel2 gr_sel opcode3 instruction
0 0 1 0 x x invalid
0 0 1 1 0 x instruction
0 0 1 1 1 x invalid
1 0 1 0 0 0 invalid
1 0 1 0 0 1 instruction
1 0 1 0 1 x instruction
1 0 1 1 0 x instruction
1 0 1 1 1 x invalid
2 0 1 0 x x instruction
2 0 1 1 0 x instruction
2 0 1 1 1 x invalid
3 0 1 0 0 0 invalid
3 0 1 0 0 1 instruction
3 0 1 0 1 x instruction
3 0 1 1 0 x instruction
3 0 1 1 1 x invalid
4 0 1 0 0 0 invalid
4 0 1 0 0 1 instruction
4 0 1 0 1 x instruction
4 0 1 1 0 x instruction
4 0 1 1 1 x invalid
5 0 1 0 x x instruction
5 0 1 1 0 x instruction
5 0 1 1 1 x invalid
6 0 1 0 x x instruction
6 0 1 1 0 x instruction
6 0 1 1 1 x invalid
7 0 1 0 x x instruction
7 0 1 1 0 x instruction
7 0 1 1 1 x invalid

gr_sel = 1, op_sel2 = 1

opcode2 op_sel2 gr_sel opcode3 instruction
0 1 1 0 x x invalid
0 1 1 1 0 x instruction
0 1 1 1 1 x invalid
1 1 1 0 0 0 invalid
1 1 1 0 0 1 instruction
1 1 1 0 1 x instruction
1 1 1 1 0 x instruction
1 1 1 1 1 x invalid
2 1 1 0 0 0 invalid
2 1 1 0 0 1 instruction
2 1 1 0 1 x instruction
2 1 1 1 0 x instruction
2 1 1 1 1 x invalid
3 1 1 0 0 0 invalid
3 1 1 0 0 1 instruction
3 1 1 0 1 x instruction
3 1 1 1 0 x instruction
3 1 1 1 1 x invalid
4 1 1 0 x x instruction
4 1 1 1 0 x instruction
4 1 1 1 1 x invalid
5 1 1 0 x x instruction
5 1 1 1 0 x instruction
5 1 1 1 1 x invalid
6 1 1 0 x x invalid
6 1 1 1 0 x invalid
6 1 1 1 1 x invalid
7 1 1 0 x x invalid
7 1 1 1 0 x invalid
7 1 1 1 1 x invalid
Allowed Instructions

This table describes 10 instructions with 28 variations.

opcode2, op_sel2 is a composite key.

opcode2 op_sel2 op_sel1 instruction
0 0 0 MAD IRII
0 0 1 MAD RRII EXP1
0 1 0 MAD IRII EXP
0 1 1 MAD RRII EXP2
1 0 0 DOT IRI
1 0 1 DOT RRI EXP
1 1 0 ADD IRI EXP
1 1 1 ADD RRI EXP2
2 0 0 DOT IRR EXP
2 0 1 DOT RRR EXP
2 1 0 MUL IRI EXP
2 1 1 MUL RRI EXP2
3 0 0 MUL IRI
3 0 1 MUL RRI EXP1
3 1 0 SUBFLR IRI
3 1 1 SUBFLR RRI EXP
4 0 0 ADD IRI
4 0 1 ADD RRI EXP1
4 1 0 EXP IR
4 1 1 EXP RR EXP
5 0 0 MOV IR
5 0 1 MOV RR EXP
5 1 0 LOG IR
5 1 1 LOG RR EXP
6 0 0 RSQ IR
6 0 1 RSQ RR EXP
6 1 0 invalid
6 1 1 invalid
7 0 0 RCP IR
7 0 1 RCP RR EXP
7 1 0 invalid
7 1 1 invalid

Notes

  • Only 10 main instructions and 1 variation is documented at the moment.

MAD IRII

Encoding

Higher 4 bytes

7 6 5 4 3 2 1 0
opcode1 neg_op2i predicate
0 0 1 0 0
7 6 5 4 3 2 1 0
op_sel2 data_format swz_alt_op2i_2 swz_alt_opN opN_swz
x 0
7 6 5 4 3 2 1 0
neg_opN opcode2 gr_sel op_sel1 swz_alt_op1i
0 0 0 x 0
7 6 5 4 3 2 1 0
op1i_swz swz_alt_op2i_x op2i_swz opt0
x x

Lower 4 bytes

7 6 5 4 3 2 1 0
optN op0i op0
x x x x
7 6 5 4 3 2 1 0
op0 opcode3 opN_sel
x x x x x x
7 6 5 4 3 2 1 0
op3i op2i op1i swz_mask2 swz_mask1
7 6 5 4 3 2 1 0
swz_en opN
Fields - instruction

data_format:

5 value
0 f32
1 f16

predicate:

1 0 value
0 0
0 1 p0
1 0 !p0
1 1 Pn
Fields - operands Compatibility

One or multiple registers can change type between register R7 and internal register RI2.

Different combinations of operand types are controlled by opN_sel.

This field controls position of opN operand which is a normal register.

Not all types of fields are applicable to each combination of registers.

opN_sel operands example
0 0 IRII mad.f32 i0, r0, i0, i0
0 1 IIRI mad.f32 i0, i0, r0, i0
1 0 IIIR mad.f32 i0, i0, i0, r0
1 1 IIII mad.f32 i0, i0, i0, i0
opN_sel neg_op2i swz_alt_op2i_2 swz_alt_opN opN_swz neg_opN swz_alt_op1i op1i_swz swz_alt_op2i_x op2i_swz optN op0i op3i op2i op1i swz_mask2 swz_mask1 swz_en opN
0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
0 1 1 1 1 1 1 1 1 1 1 1 1 1 1
1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
Fields - operands
Swizzle masking

Masking is controled by control bits:

  • control bits: swz_en, swz_mask1, swz_mask2

Each channel can be masked with control bits. Combinations of control bits produce the following masking table.

Encoding used in masking table:

value meaning
0 channel not selected
1 channel selected
x channel masked

Masking table:

swz_mask2 swz_mask1 swz_en ch0 ch1 ch2
0 0 0 0 0 0
0 0 1 1 0 0
0 1 0 x 1 0
0 1 1 1 1 0
1 0 0 x x 1
1 0 1 1 x 1
1 1 0 x 1 1
1 1 1 1 1 1
Swizzles - operand 0 i

Swizzles of operand 0 i can not be controled and have predefined combinations described below:

value
xyz

Each channel can be masked with control bits. Masking is described in Swizzle masking.

  • swz_en, swz_mask1, swz_mask2
Swizzles - operand 1 i

Swizzles of operand 1 i can not be precisely controlled and have predefined combinations described below and controlled by swizzle fields:

  • op1i_swz, swz_alt_op1i

Each channel can be masked with control bits. Masking is described in Swizzle masking.

  • swz_en, swz_mask1, swz_mask2
swz_alt_op1i op1i_swz value
0 0 0 0 xxx
0 0 0 1 yyy
0 0 1 0 zzz
0 0 1 1 www
0 1 0 0 xyz
0 1 0 1 yzw
0 1 1 0 xxy
0 1 1 1 xyx
1 0 0 0 yyx
1 0 0 1 yyz
1 0 1 0 zxy
1 0 1 1 xzy
1 1 0 0 {0, 0, 0}
1 1 0 1 {0.5, 0.5, 0.5}
1 1 1 0 {1, 1, 1}
1 1 1 1 {2, 2, 2}
Swizzles - operand 2 i

Swizzles of operand 2 i can not be precisely controlled and have predefined combinations described below and controlled by swizzle fields:

  • op2i_swz, swz_alt_op2i_x, swz_alt_op2i_2

Each channel can be masked with control bits. Masking is described in Swizzle masking.

  • swz_en, swz_mask1, swz_mask2
swz_alt_op2i_2 swz_alt_op2i_x op2i_swz value
0 0 0 0 0 xxx
0 0 0 0 1 yyy
0 0 0 1 0 zzz
0 0 0 1 1 www
0 0 1 0 0 xyz
0 0 1 0 1 yzw
0 0 1 1 0 xxy
0 0 1 1 1 xyx
0 1 0 0 0 yyx
0 1 0 0 1 yyz
0 1 0 1 0 zxy
0 1 0 1 1 xzy
0 1 1 0 0 {0, 0, 0}
0 1 1 0 1 {0.5, 0.5, 0.5}
0 1 1 1 0 {1, 1, 1}
0 1 1 1 1 {2, 2, 2}
1 0 0 0 0 xyy
1 0 0 0 1 yxy
1 0 0 1 0 xxz
1 0 0 1 1 yxx
1 0 1 0 0 xy0
1 0 1 0 1 x10
1 0 1 1 0 xzy
1 0 1 1 1 yzx
1 1 0 0 0 zyx
1 1 0 0 1 zzy
1 1 0 1 0 xy1
1 1 0 1 1 {0.5, 0.5, 0.5}
1 1 1 0 0 {0.5, 0.5, 0.5}
1 1 1 0 1 {0.5, 0.5, 0.5}
1 1 1 1 0 {0.5, 0.5, 0.5}
1 1 1 1 1 {0.5, 0.5, 0.5}
Swizzles - operand 3 i

Swizzles of operand 3 i can not be controled and have predefined combinations described below:

value
xyz

Each channel can be masked with control bits. Masking is described in Swizzle masking.

  • swz_en, swz_mask1, swz_mask2
Swizzles - operand N

Swizzles of operand N can not be precisely controlled and have predefined combinations described below and controlled by swizzle fields:

  • opN_swz, swz_alt_opN

Each channel can be masked with control bits. Masking is described in Swizzle masking.

  • swz_en, swz_mask1, swz_mask2
swz_alt_opN opN_swz value
0 0 0 0 xxx
0 0 0 1 yyy
0 0 1 0 zzz
0 0 1 1 www
0 1 0 0 xyz
0 1 0 1 yzw
0 1 1 0 xxy
0 1 1 1 xyx
1 0 0 0 yyx
1 0 0 1 yyz
1 0 1 0 zxy
1 0 1 1 xzy
1 1 0 0 {0, 0, 0}
1 1 0 1 {0.5, 0.5, 0.5}
1 1 1 0 {1, 1, 1}
1 1 1 1 {2, 2, 2}
Examples
mad.f32 i0, r0, i0, i0

mad.f16 i0, r0, i0, i0

MAD RRII EXP1

Encoding

Higher 4 bytes

7 6 5 4 3 2 1 0
opcode1 neg_op2i predicate
0 0 1 0 0
7 6 5 4 3 2 1 0
op_sel2 data_format swz_alt_op2i_2 swz_alt_opN opN_swz
x 0
7 6 5 4 3 2 1 0
neg_opN opcode2 gr_sel op_sel1 swz_alt_op1i
0 0 0 x 1
7 6 5 4 3 2 1 0
op1i_swz swz_alt_op2i_x op2i_swz opt0

Lower 4 bytes

7 6 5 4 3 2 1 0
optN op0i op0
x x
7 6 5 4 3 2 1 0
op0 opcode3 opN_sel
x x x
7 6 5 4 3 2 1 0
op3i op2i op1i swz_mask2 swz_mask1
x x
7 6 5 4 3 2 1 0
swz_en opN
x
Fields - instruction

data_format:

5 value
0 f32
1 f16

predicate:

1 0 value
0 0
0 1 p0
1 0 !p0
1 1 Pn
Fields - operands Compatibility

One or multiple registers can change type between register R7 and internal register RI2.

Different combinations of operand types are controlled by opN_sel.

This field controls position of opN operand which is a normal register.

Not all types of fields are applicable to each combination of registers.

opN_sel operands example
0 0 RRII mad.f32 r0.xy, r0.xx, i0.xx, i0.xy
0 1 RIRI mad.f32 r0.xy, i0.xx, r0.xx, i0.xy
1 0 RIIR mad.f32 r0.xy, i0.xx, i0.xx, r0.xx
1 1 RIII mad.f32 r0.xy, i0.xx, i0.xx, i0.xy
opN_sel neg_op2i swz_alt_op2i_2 swz_alt_opN opN_swz neg_opN swz_alt_op1i op1i_swz swz_alt_op2i_x op2i_swz opt0 optN op0 op3i op2i op1i opN
0 0 1 1 1 1 1 1 1 1 1 1 1 1 1
0 1 1 1 1 1 1 1 1 1 1 1 1
1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1 1 1 1
Fields - operands
Swizzle masking

Swizzle masking does not apply because instruction has explicit swizzling.

Swizzles - operand 0

Swizzles of operand 0 can not be controlled and have predefined combinations described below:

Masking does not apply because instruction has explicit swizzling

data_format = 0

value
xy

data_format = 1

value
xyzw
Swizzles - operand 1 i

Swizzles of operand 1 i can not be precisely controlled and have predefined combinations described below and controlled by swizzle fields:

  • op1i_swz, swz_alt_op1i

Masking does not apply because instruction has explicit swizzling

data_format = 0

swz_alt_op1i op1i_swz value
0 0 0 0 xx
0 0 0 1 yy
0 0 1 0 zz
0 0 1 1 ww
0 1 0 0 xy
0 1 0 1 yz
0 1 1 0 xx
0 1 1 1 xy
1 0 0 0 yy
1 0 0 1 yy
1 0 1 0 zx
1 0 1 1 xz
1 1 0 0 {0, 0}
1 1 0 1 {0.5, 0.5}
1 1 1 0 {1, 1}
1 1 1 1 {2, 2}

data_format = 1

swz_alt_op1i op1i_swz value
0 0 0 0 xxxx
0 0 0 1 yyyx
0 0 1 0 zzzx
0 0 1 1 wwwx
0 1 0 0 xyzx
0 1 0 1 yzwx
0 1 1 0 xxyx
0 1 1 1 xyxx
1 0 0 0 yyxx
1 0 0 1 yyzx
1 0 1 0 zxyx
1 0 1 1 xzyx
1 1 0 0 000x
1 1 0 1 hhhx
1 1 1 0 111x
1 1 1 1 222x
Swizzles - operand 2 i

Swizzles of operand 2 i can not be precisely controlled and have predefined combinations described below and controlled by swizzle fields:

  • op2i_swz, swz_alt_op2i_x, swz_alt_op2i_2

Masking does not apply because instruction has explicit swizzling

data_format = 0

swz_alt_op2i_2 swz_alt_op2i_x op2i_swz value
0 0 0 0 0 xx
0 0 0 0 1 yy
0 0 0 1 0 zz
0 0 0 1 1 ww
0 0 1 0 0 xy
0 0 1 0 1 yz
0 0 1 1 0 xx
0 0 1 1 1 xy
0 1 0 0 0 yy
0 1 0 0 1 yy
0 1 0 1 0 zx
0 1 0 1 1 xz
0 1 1 0 0 {0, 0}
0 1 1 0 1 {0.5, 0.5}
0 1 1 1 0 {1, 1}
0 1 1 1 1 {2, 2}
1 0 0 0 0 xy
1 0 0 0 1 yx
1 0 0 1 0 xx
1 0 0 1 1 yx
1 0 1 0 0 xy
1 0 1 0 1 x1
1 0 1 1 0 xz
1 0 1 1 1 yz
1 1 0 0 0 zy
1 1 0 0 1 zz
1 1 0 1 0 xy
1 1 0 1 1 {0.5, 0.5}
1 1 1 0 0 {0.5, 0.5}
1 1 1 0 1 {0.5, 0.5}
1 1 1 1 0 {0.5, 0.5}
1 1 1 1 1 {0.5, 0.5}

data_format = 1

swz_alt_op2i_2 swz_alt_op2i_x op2i_swz value
0 0 0 0 0 xxxx
0 0 0 0 1 yyyx
0 0 0 1 0 zzzx
0 0 0 1 1 wwwx
0 0 1 0 0 xyzx
0 0 1 0 1 yzwx
0 0 1 1 0 xxyx
0 0 1 1 1 xyxx
0 1 0 0 0 yyxx
0 1 0 0 1 yyzx
0 1 0 1 0 zxyx
0 1 0 1 1 xzyx
0 1 1 0 0 000x
0 1 1 0 1 hhhx
0 1 1 1 0 111x
0 1 1 1 1 222x
1 0 0 0 0 xyyx
1 0 0 0 1 yxyx
1 0 0 1 0 xxzx
1 0 0 1 1 yxxx
1 0 1 0 0 xy0x
1 0 1 0 1 x10x
1 0 1 1 0 xzyx
1 0 1 1 1 yzxx
1 1 0 0 0 zyxx
1 1 0 0 1 zzyx
1 1 0 1 0 xy1x
1 1 0 1 1 {0.5, 0.5, 0.5, 0.5}
1 1 1 0 0 {0.5, 0.5, 0.5, 0.5}
1 1 1 0 1 {0.5, 0.5, 0.5, 0.5}
1 1 1 1 0 {0.5, 0.5, 0.5, 0.5}
1 1 1 1 1 {0.5, 0.5, 0.5, 0.5}
Swizzles - operand 3 i

Swizzles of operand 3 i can not be controled and have predefined combinations described below:

Masking does not apply because instruction has explicit swizzling

data_format = 0

value
xy

data_format = 1

value
xyzw
Swizzles - operand N

Swizzles of operand N can not be precisely controlled and have predefined combinations described below and controlled by swizzle fields:

  • opN_swz, swz_alt_opN

Masking does not apply because instruction has explicit swizzling

data_format = 0

swz_alt_opN opN_swz value
0 0 0 0 xx
0 0 0 1 yy
0 0 1 0 zz
0 0 1 1 ww
0 1 0 0 xy
0 1 0 1 yz
0 1 1 0 xx
0 1 1 1 xy
1 0 0 0 yy
1 0 0 1 yy
1 0 1 0 zx
1 0 1 1 xz
1 1 0 0 {0, 0}
1 1 0 1 {0.5, 0.5}
1 1 1 0 {1, 1}
1 1 1 1 {2, 2}

data_format = 1

swz_alt_opN opN_swz value
0 0 0 0 xxxx
0 0 0 1 yyyx
0 0 1 0 zzzx
0 0 1 1 wwwx
0 1 0 0 xyzx
0 1 0 1 yzwx
0 1 1 0 xxyx
0 1 1 1 xyxx
1 0 0 0 yyxx
1 0 0 1 yyzx
1 0 1 0 zxyx
1 0 1 1 xzyx
1 1 0 0 000x
1 1 0 1 hhhx
1 1 1 0 111x
1 1 1 1 222x
Examples
mad.f32 r0.xy, r0.xx, i0.xx, i0.xy

mad.f16 r0.xyzw, r0.xxxx, i0.xxxx, i0.xyzw

MAD IRII EXP

Encoding

Higher 4 bytes

7 6 5 4 3 2 1 0
opcode1
0 0 1 0 0 x x x
7 6 5 4 3 2 1 0
op_sel2
x 1 x x x x x x
7 6 5 4 3 2 1 0
opcode2 gr_sel op_sel1
x 0 0 0 x 0 x x
7 6 5 4 3 2 1 0
x x x x x x x x

Lower 4 bytes

7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
opcode3
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x

MAD RRII EXP2

Encoding

Higher 4 bytes

7 6 5 4 3 2 1 0
opcode1
0 0 1 0 0 x x x
7 6 5 4 3 2 1 0
op_sel2
x 1 x x x x x x
7 6 5 4 3 2 1 0
opcode2 gr_sel op_sel1
x 0 0 0 x 1 x x
7 6 5 4 3 2 1 0
x x x x x x x x

Lower 4 bytes

7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
opcode3
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x

DOT IRI

Encoding

Higher 4 bytes

7 6 5 4 3 2 1 0
opcode1 neg_op2i predicate
0 0 1 0 0
7 6 5 4 3 2 1 0
op_sel2 data_format swz_alt_op2i_2 swz_alt_opN opN_swz
x 0
7 6 5 4 3 2 1 0
neg_opN opcode2 gr_sel op_sel1 swz_alt_op1i
0 0 1 x 0
7 6 5 4 3 2 1 0
op1i_swz swz_alt_op2i_x op2i_swz opt0
x x

Lower 4 bytes

7 6 5 4 3 2 1 0
optN op0i op0
x x x x
7 6 5 4 3 2 1 0
op0 opcode3 opN_sel
x x x x x x
7 6 5 4 3 2 1 0
op3i op2i op1i swz_mask2 swz_mask1
x x
7 6 5 4 3 2 1 0
swz_en opN
Fields - instruction

data_format:

5 value
0 f32
1 f16

predicate:

1 0 value
0 0
0 1 p0
1 0 !p0
1 1 Pn
Fields - operands Compatibility

One or multiple registers can change type between register R7 and internal register RI2.

Different combinations of operand types are controlled by opN_sel.

This field controls position of opN operand which is a normal register.

Not all types of fields are applicable to each combination of registers.

opN_sel operands example
0 0 IRI dot.f32 i0, r0.xxx, i0.xxx
0 1 IIR dot.f32 i0, i0.xxx, r0.xxx
1 0 III dot.f32 i0, i0.xxx, i0.xxx
1 1 invalid
opN_sel neg_op2i swz_alt_op2i_2 swz_alt_opN opN_swz neg_opN swz_alt_op1i op1i_swz swz_alt_op2i_x op2i_swz optN op0i op2i op1i swz_mask2 swz_mask1 swz_en opN
0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1
0 1 1 1 1 1 1 1 1 1 1 1 1 1
1 0 1 1 1 1 1 1 1 1 1 1 1 1
1 1 invalid invalid invalid invalid invalid invalid invalid invalid invalid invalid invalid invalid invalid invalid invalid invalid invalid
Fields - operands
Swizzle masking

Masking is controled by control bits:

  • control bits: swz_en, swz_mask1, swz_mask2

Each channel can be masked with control bits. Combinations of control bits produce the following masking table.

Masking does not apply to operand 1 i and operand 2 i because they have explicit swizzling.

Encoding used in masking table:

value meaning
0 channel not selected
1 channel selected
x channel masked

Masking table:

swz_mask2 swz_mask1 swz_en ch0 ch1 ch2
0 0 0 0 0 0
0 0 1 1 0 0
0 1 0 x 1 0
0 1 1 1 1 0
1 0 0 x x 1
1 0 1 1 x 1
1 1 0 x 1 1
1 1 1 1 1 1
Swizzles - operand 0 i

Swizzles of operand 0 i can not be controled and have predefined combinations described below:

value
xyz

Each channel can be masked with control bits. Masking is described in Swizzle masking.

  • swz_en, swz_mask1, swz_mask2
Swizzles - operand 1 i

Swizzles of operand 1 i can not be precisely controlled and have predefined combinations described below and controlled by swizzle fields:

  • op1i_swz, swz_alt_op1i

Masking does not apply to operand 1 i because it has explicit swizzling.

swz_alt_op1i op1i_swz value
0 0 0 0 xxx
0 0 0 1 yyy
0 0 1 0 zzz
0 0 1 1 www
0 1 0 0 xyz
0 1 0 1 yzw
0 1 1 0 xxy
0 1 1 1 xyx
1 0 0 0 yyx
1 0 0 1 yyz
1 0 1 0 zxy
1 0 1 1 xzy
1 1 0 0 {0, 0, 0}
1 1 0 1 {0.5, 0.5, 0.5}
1 1 1 0 {1, 1, 1}
1 1 1 1 {2, 2, 2}
Swizzles - operand 2 i

Swizzles of operand 2 i can not be precisely controlled and have predefined combinations described below and controlled by swizzle fields:

  • op2i_swz, swz_alt_op2i_x, swz_alt_op2i_2

Masking does not apply to operand 2 i because it has explicit swizzling.

swz_alt_op2i_2 swz_alt_op2i_x op2i_swz value
0 0 0 0 0 xxx
0 0 0 0 1 yyy
0 0 0 1 0 zzz
0 0 0 1 1 www
0 0 1 0 0 xyz
0 0 1 0 1 yzw
0 0 1 1 0 xxy
0 0 1 1 1 xyx
0 1 0 0 0 yyx
0 1 0 0 1 yyz
0 1 0 1 0 zxy
0 1 0 1 1 xzy
0 1 1 0 0 {0, 0, 0}
0 1 1 0 1 {0.5, 0.5, 0.5}
0 1 1 1 0 {1, 1, 1}
0 1 1 1 1 {2, 2, 2}
1 0 0 0 0 xyy
1 0 0 0 1 yxy
1 0 0 1 0 xxz
1 0 0 1 1 yxx
1 0 1 0 0 xy0
1 0 1 0 1 x10
1 0 1 1 0 xzy
1 0 1 1 1 yzx
1 1 0 0 0 zyx
1 1 0 0 1 zzy
1 1 0 1 0 xy1
1 1 0 1 1 {0.5, 0.5, 0.5}
1 1 1 0 0 {0.5, 0.5, 0.5}
1 1 1 0 1 {0.5, 0.5, 0.5}
1 1 1 1 0 {0.5, 0.5, 0.5}
1 1 1 1 1 {0.5, 0.5, 0.5}
Swizzles - operand N

Swizzles of operand N can not be precisely controlled and have predefined combinations described below and controlled by swizzle fields:

  • opN_swz, swz_alt_opN

Masking does not apply to operand 1 N because it has explicit swizzling.

swz_alt_opN opN_swz value
0 0 0 0 xxx
0 0 0 1 yyy
0 0 1 0 zzz
0 0 1 1 www
0 1 0 0 xyz
0 1 0 1 yzw
0 1 1 0 xxy
0 1 1 1 xyx
1 0 0 0 yyx
1 0 0 1 yyz
1 0 1 0 zxy
1 0 1 1 xzy
1 1 0 0 {0, 0, 0}
1 1 0 1 {0.5, 0.5, 0.5}
1 1 1 0 {1, 1, 1}
1 1 1 1 {2, 2, 2}
Examples
dot.f32 i0, r0.xxx, i0.xxx

dot.f16 i0, r0.xxx, i0.xxx

DOT RRI EXP

Encoding

Higher 4 bytes

7 6 5 4 3 2 1 0
opcode1
0 0 1 0 0 x x x
7 6 5 4 3 2 1 0
op_sel2
x 0 x x x x x x
7 6 5 4 3 2 1 0
opcode2 gr_sel op_sel1
x 0 0 1 x 1 x x
7 6 5 4 3 2 1 0
x x x x x x x x

Lower 4 bytes

7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
opcode3
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x

DOT IRR EXP

Encoding

Higher 4 bytes

7 6 5 4 3 2 1 0
opcode1 neg_op2i predicate
0 0 1 0 0
7 6 5 4 3 2 1 0
op_sel2 data_format swz_alt_op2i_2 swz_alt_opN opN_swz
x 0
7 6 5 4 3 2 1 0
neg_opN opcode2 gr_sel op_sel1 swz_alt_op1i
0 1 0 x 0
7 6 5 4 3 2 1 0
op1i_swz swz_alt_op2i_x op2i_swz opt0
x x

Lower 4 bytes

7 6 5 4 3 2 1 0
optN op0i op0
x x x x
7 6 5 4 3 2 1 0
op0 opcode3 opN_sel
x x x x x x
7 6 5 4 3 2 1 0
op3i op2i op1i swz_mask2 swz_mask1
x x
7 6 5 4 3 2 1 0
swz_en opN
Fields - instruction

data_format:

5 value
0 f32
1 f16

predicate:

1 0 value
0 0
0 1 p0
1 0 !p0
1 1 Pn
Fields - operands Compatibility

One or multiple registers can change type between register R7 and internal register RI2.

Different combinations of operand types are controlled by opN_sel.

This field controls position of opN operand which is a normal register.

Not all types of fields are applicable to each combination of registers.

opN_sel operands example
0 0 IRR dot.f32 i0, r0.xxx, r0.xxx
0 1 III dot.f32 i0, i0.xxx, i0.xxx
1 0 III dot.f32 i0, i0.xxx, i0.xxx
1 1 II2I2 dot.f32 i0, i2.xyz, i2.xyz
opN_sel neg_op2i swz_alt_op2i_2 swz_alt_opN opN_swz neg_opN swz_alt_op1i op1i_swz swz_alt_op2i_x op2i_swz optN op0i op2i op1i swz_mask2 swz_mask1 swz_en opN
0 0 1 1 1 1 1 1 1 1 1
0 1 1 1 1 1 1 1 1
1 0 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1
Fields - operands
  • neg_op2i - negate op2i. applies to both operand 1 i and operand 2 i.
  • swz_alt_op2i_2 - change op2i swizzle. consult Swizzles - operand 2 i. applies to both operand 1 i and operand 2 i.
  • swz_alt_opN - change opN swizzle. consult Swizzles - operand N. encodes 2 operands.
  • opN_swz - opN swizzle encoded with RSWZ2. consult Swizzles - operand N. encodes 2 operands.
  • neg_opN - negate opN.
  • swz_alt_op1i - change op1i swizzle. consult Swizzles - operand 1 i. applies to both operand 1 i and operand 2 i.
  • op1i_swz - op1i swizzle encoded with RSWZ2. consult Swizzles - operand 1 i. applies to both operand 1 i and operand 2 i.
  • swz_alt_op2i_x - change op2i swizzle. consult Swizzles - operand 2 i. applies to both operand 1 and operand 2.
  • op2i_swz - op2i swizzle encoded with RSWZ2. consult Swizzles - operand 2 i. applies to both operand 1 i and operand 2 i.
  • opt0 - unused
  • optN - encoded with RS2. encodes 2 operands.
  • op0i - encoded with RI2.
  • op0 - unused
  • op3i - unused
  • op2i - encoded with RI2. encodes both operand 1 i and operand 2 i.
  • op1i - encoded with RI2. encodes both operand 1 i and operand 2 i.
  • swz_mask2 - mask swizzle. consult Swizzle masking.
  • swz_mask1 - mask swizzle. consult Swizzle masking.
  • swz_en - enables usage of swizzling. consult Swizzle masking.
  • opN - encode opN operand selected by opN_sel as R7. encodes 2 operands.
Swizzle masking

Masking is controled by control bits:

  • control bits: swz_en, swz_mask1, swz_mask2

Each channel can be masked with control bits. Combinations of control bits produce the following masking table.

Masking does not apply to operand 1 i and operand 2 i because they have explicit swizzling.

Encoding used in masking table:

value meaning
0 channel not selected
1 channel selected
x channel masked

Masking table:

swz_mask2 swz_mask1 swz_en ch0 ch1 ch2
0 0 0 0 0 0
0 0 1 1 0 0
0 1 0 x 1 0
0 1 1 1 1 0
1 0 0 x x 1
1 0 1 1 x 1
1 1 0 x 1 1
1 1 1 1 1 1
Swizzles - operand 0 i

Swizzles of operand 0 i can not be controled and have predefined combinations described below:

value
xyz

Each channel can be masked with control bits. Masking is described in Swizzle masking.

  • swz_en, swz_mask1, swz_mask2
Swizzles - operand 1 i

Swizzles of operand 1 i can not be precisely controlled and have predefined combinations described below and controlled by swizzle fields:

  • op1i_swz, swz_alt_op1i

Masking does not apply to operand 1 i because it has explicit swizzling.

swz_alt_op1i op1i_swz value
0 0 0 0 xxx
0 0 0 1 yyy
0 0 1 0 zzz
0 0 1 1 www
0 1 0 0 xyz
0 1 0 1 yzw
0 1 1 0 xxy
0 1 1 1 xyx
1 0 0 0 yyx
1 0 0 1 yyz
1 0 1 0 zxy
1 0 1 1 xzy
1 1 0 0 {0, 0, 0}
1 1 0 1 {0.5, 0.5, 0.5}
1 1 1 0 {1, 1, 1}
1 1 1 1 {2, 2, 2}

When opN_sel is equal to 3 - operand 1 i swizzle can not be controlled and equals the following value:

value
xyz
Swizzles - operand 2 i

Swizzles of operand 2 i can not be precisely controlled and have predefined combinations described below and controlled by swizzle fields:

  • op2i_swz, swz_alt_op2i_x, swz_alt_op2i_2

Masking does not apply to operand 2 i because it has explicit swizzling.

swz_alt_op2i_2 swz_alt_op2i_x op2i_swz value
0 0 0 0 0 xxx
0 0 0 0 1 yyy
0 0 0 1 0 zzz
0 0 0 1 1 www
0 0 1 0 0 xyz
0 0 1 0 1 yzw
0 0 1 1 0 xxy
0 0 1 1 1 xyx
0 1 0 0 0 yyx
0 1 0 0 1 yyz
0 1 0 1 0 zxy
0 1 0 1 1 xzy
0 1 1 0 0 {0, 0, 0}
0 1 1 0 1 {0.5, 0.5, 0.5}
0 1 1 1 0 {1, 1, 1}
0 1 1 1 1 {2, 2, 2}
1 0 0 0 0 xyy
1 0 0 0 1 yxy
1 0 0 1 0 xxz
1 0 0 1 1 yxx
1 0 1 0 0 xy0
1 0 1 0 1 x10
1 0 1 1 0 xzy
1 0 1 1 1 yzx
1 1 0 0 0 zyx
1 1 0 0 1 zzy
1 1 0 1 0 xy1
1 1 0 1 1 {0.5, 0.5, 0.5}
1 1 1 0 0 {0.5, 0.5, 0.5}
1 1 1 0 1 {0.5, 0.5, 0.5}
1 1 1 1 0 {0.5, 0.5, 0.5}
1 1 1 1 1 {0.5, 0.5, 0.5}

When opN_sel is equal to 3 - operand 2 i swizzle can not be controlled and equals the following value:

value
xyz
Swizzles - operand N

Swizzles of operand N can not be precisely controlled and have predefined combinations described below and controlled by swizzle fields:

  • opN_swz, swz_alt_opN

Masking does not apply to operand 1 N because it has explicit swizzling.

swz_alt_opN opN_swz value
0 0 0 0 xxx
0 0 0 1 yyy
0 0 1 0 zzz
0 0 1 1 www
0 1 0 0 xyz
0 1 0 1 yzw
0 1 1 0 xxy
0 1 1 1 xyx
1 0 0 0 yyx
1 0 0 1 yyz
1 0 1 0 zxy
1 0 1 1 xzy
1 1 0 0 {0, 0, 0}
1 1 0 1 {0.5, 0.5, 0.5}
1 1 1 0 {1, 1, 1}
1 1 1 1 {2, 2, 2}
Examples
dot.f32 i0, r0.xxx, r0.xxx

dot.f16 i0, r0.xxx, r0.xxx

DOT RRR EXP

Encoding

Higher 4 bytes

7 6 5 4 3 2 1 0
opcode1
0 0 1 0 0 x x x
7 6 5 4 3 2 1 0
op_sel2
x 0 x x x x x x
7 6 5 4 3 2 1 0
opcode2 gr_sel op_sel1
x 0 1 0 x 1 x x
7 6 5 4 3 2 1 0
x x x x x x x x

Lower 4 bytes

7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
opcode3
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x

ADD IRI

Encoding

Higher 4 bytes

7 6 5 4 3 2 1 0
opcode1 neg_op2i predicate
0 0 1 0 0
7 6 5 4 3 2 1 0
op_sel2 data_format swz_alt_op2i_2 swz_alt_opN opN_swz
x 0
7 6 5 4 3 2 1 0
neg_opN opcode2 gr_sel op_sel1 swz_alt_op1i
1 0 0 x 0
7 6 5 4 3 2 1 0
op1i_swz swz_alt_op2i_x op2i_swz opt0
x x

Lower 4 bytes

7 6 5 4 3 2 1 0
optN op0i op0
x x x x
7 6 5 4 3 2 1 0
op0 opcode3 opN_sel
x x x x x x
7 6 5 4 3 2 1 0
op3i op2i op1i swz_mask2 swz_mask1
x x
7 6 5 4 3 2 1 0
swz_en opN
Fields - instruction

data_format:

5 value
0 f32
1 f16

predicate:

1 0 value
0 0
0 1 p0
1 0 !p0
1 1 Pn
Fields - operands Compatibility

One or multiple registers can change type between register R7 and internal register RI2.

Different combinations of operand types are controlled by opN_sel.

This field controls position of opN operand which is a normal register.

Not all types of fields are applicable to each combination of registers.

opN_sel operands example
0 0 IRI add.f32 i0, r0, i0
0 1 IIR add.f32 i0, i0, r0
1 0 III add.f32 i0, i0, i0
1 1 invalid
opN_sel neg_op2i swz_alt_op2i_2 swz_alt_opN opN_swz neg_opN swz_alt_op1i op1i_swz swz_alt_op2i_x op2i_swz optN op0i op2i op1i swz_mask2 swz_mask1 swz_en opN
0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1
0 1 1 1 1 1 1 1 1 1 1 1 1 1
1 0 1 1 1 1 1 1 1 1 1 1 1 1
1 1 invalid invalid invalid invalid invalid invalid invalid invalid invalid invalid invalid invalid invalid invalid invalid invalid invalid
Fields - operands
Swizzle masking

Masking is controled by control bits:

  • control bits: swz_en, swz_mask1, swz_mask2

Each channel can be masked with control bits. Combinations of control bits produce the following masking table.

Encoding used in masking table:

value meaning
0 channel not selected
1 channel selected
x channel masked

Masking table:

swz_mask2 swz_mask1 swz_en ch0 ch1 ch2
0 0 0 0 0 0
0 0 1 1 0 0
0 1 0 x 1 0
0 1 1 1 1 0
1 0 0 x x 1
1 0 1 1 x 1
1 1 0 x 1 1
1 1 1 1 1 1
Swizzles - operand 0 i

Swizzles of operand 0 i can not be controled and have predefined combinations described below:

value
xyz

Each channel can be masked with control bits. Masking is described in Swizzle masking.

  • swz_en, swz_mask1, swz_mask2
Swizzles - operand 1 i

Swizzles of operand 1 i can not be precisely controlled and have predefined combinations described below and controlled by swizzle fields:

  • op1i_swz, swz_alt_op1i

Each channel can be masked with control bits. Masking is described in Swizzle masking.

  • swz_en, swz_mask1, swz_mask2
swz_alt_op1i op1i_swz value
0 0 0 0 xxx
0 0 0 1 yyy
0 0 1 0 zzz
0 0 1 1 www
0 1 0 0 xyz
0 1 0 1 yzw
0 1 1 0 xxy
0 1 1 1 xyx
1 0 0 0 yyx
1 0 0 1 yyz
1 0 1 0 zxy
1 0 1 1 xzy
1 1 0 0 {0, 0, 0}
1 1 0 1 {0.5, 0.5, 0.5}
1 1 1 0 {1, 1, 1}
1 1 1 1 {2, 2, 2}
Swizzles - operand 2 i

Swizzles of operand 2 i can not be precisely controlled and have predefined combinations described below and controlled by swizzle fields:

  • op2i_swz, swz_alt_op2i_x, swz_alt_op2i_2

Each channel can be masked with control bits. Masking is described in Swizzle masking.

  • swz_en, swz_mask1, swz_mask2
swz_alt_op2i_2 swz_alt_op2i_x op2i_swz value
0 0 0 0 0 xxx
0 0 0 0 1 yyy
0 0 0 1 0 zzz
0 0 0 1 1 www
0 0 1 0 0 xyz
0 0 1 0 1 yzw
0 0 1 1 0 xxy
0 0 1 1 1 xyx
0 1 0 0 0 yyx
0 1 0 0 1 yyz
0 1 0 1 0 zxy
0 1 0 1 1 xzy
0 1 1 0 0 {0, 0, 0}
0 1 1 0 1 {0.5, 0.5, 0.5}
0 1 1 1 0 {1, 1, 1}
0 1 1 1 1 {2, 2, 2}
1 0 0 0 0 xyy
1 0 0 0 1 yxy
1 0 0 1 0 xxz
1 0 0 1 1 yxx
1 0 1 0 0 xy0
1 0 1 0 1 x10
1 0 1 1 0 xzy
1 0 1 1 1 yzx
1 1 0 0 0 zyx
1 1 0 0 1 zzy
1 1 0 1 0 xy1
1 1 0 1 1 {0.5, 0.5, 0.5}
1 1 1 0 0 {0.5, 0.5, 0.5}
1 1 1 0 1 {0.5, 0.5, 0.5}
1 1 1 1 0 {0.5, 0.5, 0.5}
1 1 1 1 1 {0.5, 0.5, 0.5}
Swizzles - operand N

Swizzles of operand N can not be precisely controlled and have predefined combinations described below and controlled by swizzle fields:

  • opN_swz, swz_alt_opN

Each channel can be masked with control bits. Masking is described in Swizzle masking.

  • swz_en, swz_mask1, swz_mask2
swz_alt_opN opN_swz value
0 0 0 0 xxx
0 0 0 1 yyy
0 0 1 0 zzz
0 0 1 1 www
0 1 0 0 xyz
0 1 0 1 yzw
0 1 1 0 xxy
0 1 1 1 xyx
1 0 0 0 yyx
1 0 0 1 yyz
1 0 1 0 zxy
1 0 1 1 xzy
1 1 0 0 {0, 0, 0}
1 1 0 1 {0.5, 0.5, 0.5}
1 1 1 0 {1, 1, 1}
1 1 1 1 {2, 2, 2}
Examples
add.f32 i0, r0, i0

add.f16 i0, r0, i0

ADD RRI EXP1

Encoding

Higher 4 bytes

7 6 5 4 3 2 1 0
opcode1
0 0 1 0 0 x x x
7 6 5 4 3 2 1 0
op_sel2
x 0 x x x x x x
7 6 5 4 3 2 1 0
opcode2 gr_sel op_sel1
x 1 0 0 x 1 x x
7 6 5 4 3 2 1 0
x x x x x x x x

Lower 4 bytes

7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
opcode3
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x

ADD IRI EXP

Encoding

Higher 4 bytes

7 6 5 4 3 2 1 0
opcode1
0 0 1 0 0 x x x
7 6 5 4 3 2 1 0
op_sel2
x 1 x x x x x x
7 6 5 4 3 2 1 0
opcode2 gr_sel op_sel1
x 0 0 1 x 0 x x
7 6 5 4 3 2 1 0
x x x x x x x x

Lower 4 bytes

7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
opcode3
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x

ADD RRI EXP2

Encoding

Higher 4 bytes

7 6 5 4 3 2 1 0
opcode1
0 0 1 0 0 x x x
7 6 5 4 3 2 1 0
op_sel2
x 1 x x x x x x
7 6 5 4 3 2 1 0
opcode2 gr_sel op_sel1
x 0 0 1 x 1 x x
7 6 5 4 3 2 1 0
x x x x x x x x

Lower 4 bytes

7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
opcode3
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x

MUL IRI

Encoding

Higher 4 bytes

7 6 5 4 3 2 1 0
opcode1 neg_op2i predicate
0 0 1 0 0
7 6 5 4 3 2 1 0
op_sel2 data_format swz_alt_op2i_2 swz_alt_opN opN_swz
x 0
7 6 5 4 3 2 1 0
neg_opN opcode2 gr_sel op_sel1 swz_alt_op1i
0 1 1 x 0
7 6 5 4 3 2 1 0
op1i_swz swz_alt_op2i_x op2i_swz opt0
x x

Lower 4 bytes

7 6 5 4 3 2 1 0
optN op0i op0
x x x x
7 6 5 4 3 2 1 0
op0 opcode3 opN_sel
x x x x x x
7 6 5 4 3 2 1 0
op3i op2i op1i swz_mask2 swz_mask1
x x
7 6 5 4 3 2 1 0
swz_en opN
Fields - instruction

data_format:

5 value
0 f32
1 f16

predicate:

1 0 value
0 0
0 1 p0
1 0 !p0
1 1 Pn
Fields - operands Compatibility

One or multiple registers can change type between register R7 and internal register RI2.

Different combinations of operand types are controlled by opN_sel.

This field controls position of opN operand which is a normal register.

Not all types of fields are applicable to each combination of registers.

opN_sel operands example
0 0 IRI mul.f32 i0, r0, i0
0 1 IIR mul.f32 i0, i0, r0
1 0 III mul.f32 i0, i0, i0
1 1 invalid
opN_sel neg_op2i swz_alt_op2i_2 swz_alt_opN opN_swz neg_opN swz_alt_op1i op1i_swz swz_alt_op2i_x op2i_swz optN op0i op2i op1i swz_mask2 swz_mask1 swz_en opN
0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1
0 1 1 1 1 1 1 1 1 1 1 1 1 1
1 0 1 1 1 1 1 1 1 1 1 1 1 1
1 1 invalid invalid invalid invalid invalid invalid invalid invalid invalid invalid invalid invalid invalid invalid invalid invalid invalid
Fields - operands
Swizzle masking

Masking is controled by control bits:

  • control bits: swz_en, swz_mask1, swz_mask2

Each channel can be masked with control bits. Combinations of control bits produce the following masking table.

Encoding used in masking table:

value meaning
0 channel not selected
1 channel selected
x channel masked

Masking table:

swz_mask2 swz_mask1 swz_en ch0 ch1 ch2
0 0 0 0 0 0
0 0 1 1 0 0
0 1 0 x 1 0
0 1 1 1 1 0
1 0 0 x x 1
1 0 1 1 x 1
1 1 0 x 1 1
1 1 1 1 1 1
Swizzles - operand 0 i

Swizzles of operand 0 i can not be controled and have predefined combinations described below:

value
xyz

Each channel can be masked with control bits. Masking is described in Swizzle masking.

  • swz_en, swz_mask1, swz_mask2
Swizzles - operand 1 i

Swizzles of operand 1 i can not be precisely controlled and have predefined combinations described below and controlled by swizzle fields:

  • op1i_swz, swz_alt_op1i

Each channel can be masked with control bits. Masking is described in Swizzle masking.

  • swz_en, swz_mask1, swz_mask2
swz_alt_op1i op1i_swz value
0 0 0 0 xxx
0 0 0 1 yyy
0 0 1 0 zzz
0 0 1 1 www
0 1 0 0 xyz
0 1 0 1 yzw
0 1 1 0 xxy
0 1 1 1 xyx
1 0 0 0 yyx
1 0 0 1 yyz
1 0 1 0 zxy
1 0 1 1 xzy
1 1 0 0 {0, 0, 0}
1 1 0 1 {0.5, 0.5, 0.5}
1 1 1 0 {1, 1, 1}
1 1 1 1 {2, 2, 2}
Swizzles - operand 2 i

Swizzles of operand 2 i can not be precisely controlled and have predefined combinations described below and controlled by swizzle fields:

  • op2i_swz, swz_alt_op2i_x, swz_alt_op2i_2

Each channel can be masked with control bits. Masking is described in Swizzle masking.

  • swz_en, swz_mask1, swz_mask2
swz_alt_op2i_2 swz_alt_op2i_x op2i_swz value
0 0 0 0 0 xxx
0 0 0 0 1 yyy
0 0 0 1 0 zzz
0 0 0 1 1 www
0 0 1 0 0 xyz
0 0 1 0 1 yzw
0 0 1 1 0 xxy
0 0 1 1 1 xyx
0 1 0 0 0 yyx
0 1 0 0 1 yyz
0 1 0 1 0 zxy
0 1 0 1 1 xzy
0 1 1 0 0 {0, 0, 0}
0 1 1 0 1 {0.5, 0.5, 0.5}
0 1 1 1 0 {1, 1, 1}
0 1 1 1 1 {2, 2, 2}
1 0 0 0 0 xyy
1 0 0 0 1 yxy
1 0 0 1 0 xxz
1 0 0 1 1 yxx
1 0 1 0 0 xy0
1 0 1 0 1 x10
1 0 1 1 0 xzy
1 0 1 1 1 yzx
1 1 0 0 0 zyx
1 1 0 0 1 zzy
1 1 0 1 0 xy1
1 1 0 1 1 {0.5, 0.5, 0.5}
1 1 1 0 0 {0.5, 0.5, 0.5}
1 1 1 0 1 {0.5, 0.5, 0.5}
1 1 1 1 0 {0.5, 0.5, 0.5}
1 1 1 1 1 {0.5, 0.5, 0.5}
Swizzles - operand N

Swizzles of operand N can not be precisely controlled and have predefined combinations described below and controlled by swizzle fields:

  • opN_swz, swz_alt_opN

Each channel can be masked with control bits. Masking is described in Swizzle masking.

  • swz_en, swz_mask1, swz_mask2
swz_alt_opN opN_swz value
0 0 0 0 xxx
0 0 0 1 yyy
0 0 1 0 zzz
0 0 1 1 www
0 1 0 0 xyz
0 1 0 1 yzw
0 1 1 0 xxy
0 1 1 1 xyx
1 0 0 0 yyx
1 0 0 1 yyz
1 0 1 0 zxy
1 0 1 1 xzy
1 1 0 0 {0, 0, 0}
1 1 0 1 {0.5, 0.5, 0.5}
1 1 1 0 {1, 1, 1}
1 1 1 1 {2, 2, 2}
Examples
mul.f32 i0, r0, i0

mul.f16 i0, r0, i0

MUL RRI EXP1

Encoding

Higher 4 bytes

7 6 5 4 3 2 1 0
opcode1
0 0 1 0 0 x x x
7 6 5 4 3 2 1 0
op_sel2
x 0 x x x x x x
7 6 5 4 3 2 1 0
opcode2 gr_sel op_sel1
x 0 1 1 x 1 x x
7 6 5 4 3 2 1 0
x x x x x x x x

Lower 4 bytes

7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
opcode3
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x

MUL IRI EXP

Encoding

Higher 4 bytes

7 6 5 4 3 2 1 0
opcode1
0 0 1 0 0 x x x
7 6 5 4 3 2 1 0
op_sel2
x 1 x x x x x x
7 6 5 4 3 2 1 0
opcode2 gr_sel op_sel1
x 0 1 0 x 0 x x
7 6 5 4 3 2 1 0
x x x x x x x x

Lower 4 bytes

7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
opcode3
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x

MUL RRI EXP2

Encoding

Higher 4 bytes

7 6 5 4 3 2 1 0
opcode1
0 0 1 0 0 x x x
7 6 5 4 3 2 1 0
op_sel2
x 1 x x x x x x
7 6 5 4 3 2 1 0
opcode2 gr_sel op_sel1
x 0 1 0 x 1 x x
7 6 5 4 3 2 1 0
x x x x x x x x

Lower 4 bytes

7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
opcode3
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x

SUBFLR IRI

Encoding

Higher 4 bytes

7 6 5 4 3 2 1 0
opcode1 neg_op2i predicate
0 0 1 0 0
7 6 5 4 3 2 1 0
op_sel2 data_format swz_alt_op2i_2 swz_alt_opN opN_swz
x 1
7 6 5 4 3 2 1 0
neg_opN opcode2 gr_sel op_sel1 swz_alt_op1i
0 1 1 x 0
7 6 5 4 3 2 1 0
op1i_swz swz_alt_op2i_x op2i_swz opt0
x x

Lower 4 bytes

7 6 5 4 3 2 1 0
optN op0i op0
x x x x
7 6 5 4 3 2 1 0
op0 opcode3 opN_sel
x x x x x x
7 6 5 4 3 2 1 0
op3i op2i op1i swz_mask2 swz_mask1
x x
7 6 5 4 3 2 1 0
swz_en opN
Fields - instruction

data_format:

5 value
0 f32
1 f16

predicate:

1 0 value
0 0
0 1 p0
1 0 !p0
1 1 Pn
Fields - operands Compatibility

One or multiple registers can change type between register R7 and internal register RI2.

Different combinations of operand types are controlled by opN_sel.

This field controls position of opN operand which is a normal register.

Not all types of fields are applicable to each combination of registers.

opN_sel operands example
0 0 IRI subflr.f32 i0, r0.x, i0.x
0 1 IIR subflr.f32 i0, i0.x, r0.x
1 0 III subflr.f32 i0, i0.x, i0.x
1 1 invalid
opN_sel neg_op2i swz_alt_op2i_2 swz_alt_opN opN_swz neg_opN swz_alt_op1i op1i_swz swz_alt_op2i_x op2i_swz optN op0i op2i op1i swz_mask2 swz_mask1 swz_en opN
0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1
0 1 1 1 1 1 1 1 1 1 1 1 1 1
1 0 1 1 1 1 1 1 1 1 1 1 1 1
1 1 invalid invalid invalid invalid invalid invalid invalid invalid invalid invalid invalid invalid invalid invalid invalid invalid invalid
Fields - operands
Swizzle masking

Masking is controled by control bits:

  • control bits: swz_en, swz_mask1, swz_mask2

Each channel can be masked with control bits. Combinations of control bits produce the following masking table.

Masking does not apply to operand 1 i and operand 2 i because they have explicit swizzling.

Encoding used in masking table:

value meaning
0 channel not selected
1 channel selected
x channel masked

Masking table:

swz_mask2 swz_mask1 swz_en ch0 ch1 ch2
0 0 0 0 0 0
0 0 1 1 0 0
0 1 0 x 1 0
0 1 1 1 1 0
1 0 0 x x 1
1 0 1 1 x 1
1 1 0 x 1 1
1 1 1 1 1 1
Swizzles - operand 0 i

Swizzles of operand 0 i can not be controled and have predefined combinations described below:

value
xyz

Each channel can be masked with control bits. Masking is described in Swizzle masking.

  • swz_en, swz_mask1, swz_mask2
Swizzles - operand 1 i

Swizzles of operand 1 i can not be precisely controlled and have predefined combinations described below and controlled by swizzle fields:

  • op1i_swz, swz_alt_op1i

Masking does not apply to operand 1 i because it has explicit swizzling.

This instruction has only one channel for operand 1 i

swz_alt_op1i op1i_swz value
0 0 0 0 x
0 0 0 1 y
0 0 1 0 z
0 0 1 1 w
0 1 0 0 x
0 1 0 1 y
0 1 1 0 x
0 1 1 1 x
1 0 0 0 y
1 0 0 1 y
1 0 1 0 z
1 0 1 1 x
1 1 0 0 x
1 1 0 1 w
1 1 1 0 y
1 1 1 1 z
Swizzles - operand 2 i

Swizzles of operand 2 i can not be precisely controlled and have predefined combinations described below and controlled by swizzle fields:

  • op2i_swz, swz_alt_op2i_x, swz_alt_op2i_2

Masking does not apply to operand 2 i because it has explicit swizzling.

This instruction has only one channel for operand 2 i

swz_alt_op2i_2 swz_alt_op2i_x op2i_swz value
0 0 0 0 0 x
0 0 0 0 1 y
0 0 0 1 0 z
0 0 0 1 1 w
0 0 1 0 0 x
0 0 1 0 1 y
0 0 1 1 0 x
0 0 1 1 1 x
0 1 0 0 0 y
0 1 0 0 1 y
0 1 0 1 0 z
0 1 0 1 1 x
0 1 1 0 0 x
0 1 1 0 1 w
0 1 1 1 0 y
0 1 1 1 1 z
1 0 0 0 0 x
1 0 0 0 1 y
1 0 0 1 0 x
1 0 0 1 1 y
1 0 1 0 0 x
1 0 1 0 1 x
1 0 1 1 0 x
1 0 1 1 1 y
1 1 0 0 0 z
1 1 0 0 1 z
1 1 0 1 0 x
1 1 0 1 1 w
1 1 1 0 0 w
1 1 1 0 1 w
1 1 1 1 0 w
1 1 1 1 1 w
Swizzles - operand N

Swizzles of operand N can not be precisely controlled and have predefined combinations described below and controlled by swizzle fields:

  • opN_swz, swz_alt_opN

Masking does not apply to operand 1 N because it has explicit swizzling.

This instruction has only one channel for operand N

swz_alt_opN opN_swz value
0 0 0 0 x
0 0 0 1 y
0 0 1 0 z
0 0 1 1 w
0 1 0 0 x
0 1 0 1 y
0 1 1 0 x
0 1 1 1 x
1 0 0 0 y
1 0 0 1 y
1 0 1 0 z
1 0 1 1 x
1 1 0 0 x
1 1 0 1 w
1 1 1 0 y
1 1 1 1 z
Examples
subflr.f32 i0, r0.x, i0.x

subflr.f16 i0, r0.x, i0.x

SUBFLR RRI EXP

Encoding

Higher 4 bytes

7 6 5 4 3 2 1 0
opcode1
0 0 1 0 0 x x x
7 6 5 4 3 2 1 0
op_sel2
x 1 x x x x x x
7 6 5 4 3 2 1 0
opcode2 gr_sel op_sel1
x 0 1 1 x 1 x x
7 6 5 4 3 2 1 0
x x x x x x x x

Lower 4 bytes

7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
opcode3
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x

EXP IR

Encoding

Higher 4 bytes

7 6 5 4 3 2 1 0
opcode1 neg_op2i predicate
0 0 1 0 0
7 6 5 4 3 2 1 0
op_sel2 data_format swz_alt_op2i_2 swz_alt_opN opN_swz
x 1
7 6 5 4 3 2 1 0
neg_opN opcode2 gr_sel op_sel1 swz_alt_op1i
1 0 0 x 0
7 6 5 4 3 2 1 0
op1i_swz swz_alt_op2i_x op2i_swz opt0
x x

Lower 4 bytes

7 6 5 4 3 2 1 0
optN op0i op0
x x x x
7 6 5 4 3 2 1 0
op0 opcode3 opN_sel
x x x x x x
7 6 5 4 3 2 1 0
unused abs_opN op2i op1i swz_mask2 swz_mask1
x
7 6 5 4 3 2 1 0
swz_en opN
Fields - instruction

data_format:

5 value
0 f32
1 f16

predicate:

1 0 value
0 0
0 1 p0
1 0 !p0
1 1 Pn
Fields - operands Compatibility

One or multiple registers can change type between register R7 and internal register RI2.

Different combinations of operand types are controlled by opN_sel.

This field controls position of opN operand which is a normal register.

Not all types of fields are applicable to each combination of registers.

opN_sel operands example
0 0 IR exp.f32 i0, r0.x
0 1 II exp.f32 i0, i0.x
1 0 II exp.f32 i0, i0.x
1 1 II2 exp.f32 i0, i2.x
opN_sel neg_op2i swz_alt_op2i_2 swz_alt_opN opN_swz neg_opN swz_alt_op1i op1i_swz swz_alt_op2i_x op2i_swz optN op0i abs_opN op2i op1i swz_mask2 swz_mask1 swz_en opN
0 0 1 1 1 1 1 1 1 1 1 1
0 1 1 1 1 1 1 1 1
1 0 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1
Fields - operands
  • neg_op2i - negate op2i. encodes operand 1 i.
  • swz_alt_op2i_2 - not a standard swizzling for op2i. consult Swizzles - operand 2 i. encodes operand 1 i.
  • swz_alt_opN - not a standard swizzling for opN. consult Swizzles - operand N.
  • opN_swz - not a standard swizzling for opN. consult Swizzles - operand N.
  • neg_opN - negate opN.
  • swz_alt_op1i - not a standard swizzling for op1i. consult Swizzles - operand 1 i.
  • op1i_swz - not a standard swizzling for op1i. consult Swizzles - operand 1 i.
  • swz_alt_op2i_x - not a standard swizzling for op2i. consult Swizzles - operand 2 i. encodes operand 1 i.
  • op2i_swz - not a standard swizzling for op2i. consult Swizzles - operand 2 i. encodes operand 1 i.
  • opt0 - unused
  • optN - encoded with RS2.
  • op0i - encoded with RI2.
  • op0 - unused
  • abs_opN - add abs modifier to opN.
  • op2i - encoded with RI2. encodes operand 1 i.
  • op1i - encoded with RI2.
  • swz_mask2 - mask swizzle. consult Swizzle masking.
  • swz_mask1 - mask swizzle. consult Swizzle masking.
  • swz_en - enables usage of swizzling. consult Swizzle masking.
  • opN - encode opN operand selected by opN_sel as R7.
Swizzle masking

Masking is controled by control bits:

  • control bits: swz_en, swz_mask1, swz_mask2

Each channel can be masked with control bits. Combinations of control bits produce the following masking table.

Masking does not apply to operand 1 since it has explicit swizzling.

Encoding used in masking table:

value meaning
0 channel not selected
1 channel selected
x channel masked

Masking table:

swz_mask2 swz_mask1 swz_en ch0 ch1 ch2
0 0 0 0 0 0
0 0 1 1 0 0
0 1 0 x 1 0
0 1 1 1 1 0
1 0 0 x x 1
1 0 1 1 x 1
1 1 0 x 1 1
1 1 1 1 1 1
Swizzles - operand 0 i

Swizzles of operand 0 i can not be controled and have predefined combinations described below:

value
xyz

Each channel can be masked with control bits. Masking is described in Swizzle masking.

  • swz_en, swz_mask1, swz_mask2
Swizzles - operand 1 i

Swizzles of operand 1 i can not be controlled. Instead masking combinations are described below and controlled by swizzle fields:

  • op1i_swz, swz_alt_op1i

Standard masking does not apply to operand 1 since it has explicit swizzling.

swz_alt_op1i op1i_swz value
0 0 0 0 x
0 0 0 1 -y
0 0 1 0 --z
0 0 1 1 ---w
0 1 0 0 x
0 1 0 1 -y
0 1 1 0 x
0 1 1 1 x
1 0 0 0 -y
1 0 0 1 -y
1 0 1 0 --z
1 0 1 1 x
1 1 0 0 x
1 1 0 1 ---w
1 1 1 0 -y
1 1 1 1 --z

When opN_sel is equal to 3 - operand 1 i swizzle can not be controlled and equals the following value:

value
x
Swizzles - operand 2 i

Control bits usually used for encoding operand 2 are used for encoding operand 1 since this instruction only has 2 operands.

Swizzles of operand 1 i can not be controlled. Instead masking combinations are described below and controlled by swizzle fields:

  • op2i_swz, swz_alt_op2i_x, swz_alt_op2i_2

Standard masking does not apply to operand 1 since it has explicit swizzling.

swz_alt_op2i_2 swz_alt_op2i_x op2i_swz value
0 0 0 0 0 x
0 0 0 0 1 -y
0 0 0 1 0 --z
0 0 0 1 1 ---w
0 0 1 0 0 x
0 0 1 0 1 -y
0 0 1 1 0 x
0 0 1 1 1 x
0 1 0 0 0 -y
0 1 0 0 1 -y
0 1 0 1 0 --z
0 1 0 1 1 x
0 1 1 0 0 x
0 1 1 0 1 ---w
0 1 1 1 0 -y
0 1 1 1 1 --z
1 0 0 0 0 x
1 0 0 0 1 -y
1 0 0 1 0 x
1 0 0 1 1 -y
1 0 1 0 0 x
1 0 1 0 1 x
1 0 1 1 0 x
1 0 1 1 1 -y
1 1 0 0 0 --z
1 1 0 0 1 --z
1 1 0 1 0 x
1 1 0 1 1 ---w
1 1 1 0 0 ---w
1 1 1 0 1 ---w
1 1 1 1 0 ---w
1 1 1 1 1 ---w

When opN_sel is equal to 3 - operand 1 i swizzle can not be controlled and equals the following value:

value
x
Swizzles - operand N

Swizzles of operand N i can not be controlled. Instead masking combinations are described below and controlled by swizzle fields:

  • opN_swz, swz_alt_opN

Standard masking does not apply to operand N since it has explicit swizzling.

swz_alt_opN opN_swz value
0 0 0 0 x
0 0 0 1 -y
0 0 1 0 --z
0 0 1 1 ---w
0 1 0 0 x
0 1 0 1 -y
0 1 1 0 x
0 1 1 1 x
1 0 0 0 -y
1 0 0 1 -y
1 0 1 0 --z
1 0 1 1 x
1 1 0 0 x
1 1 0 1 ---w
1 1 1 0 -y
1 1 1 1 --z
Examples
exp.f32 i0, r0.x

exp.f16 i0, r0.x

EXP RR EXP

Encoding

Higher 4 bytes

7 6 5 4 3 2 1 0
opcode1
0 0 1 0 0 x x x
7 6 5 4 3 2 1 0
op_sel2
x 1 x x x x x x
7 6 5 4 3 2 1 0
opcode2 gr_sel op_sel1
x 1 0 0 x 1 x x
7 6 5 4 3 2 1 0
x x x x x x x x

Lower 4 bytes

7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
opcode3
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x

MOV IR

Encoding

Higher 4 bytes

7 6 5 4 3 2 1 0
opcode1 neg_op2i predicate
0 0 1 0 0
7 6 5 4 3 2 1 0
op_sel2 data_format swz_alt_op2i_2 swz_alt_opN opN_swz
x 0
7 6 5 4 3 2 1 0
neg_opN opcode2 gr_sel op_sel1 swz_alt_op1i
1 0 1 x 0
7 6 5 4 3 2 1 0
op1i_swz swz_alt_op2i_x op2i_swz opt0
x x

Lower 4 bytes

7 6 5 4 3 2 1 0
optN op0i op0
x x x x
7 6 5 4 3 2 1 0
op0 opcode3 opN_sel
x x x x x x
7 6 5 4 3 2 1 0
unused abs_opN op2i op1i swz_mask2 swz_mask1
x
7 6 5 4 3 2 1 0
swz_en opN
Fields - instruction

data_format:

5 value
0 f32
1 f16

predicate:

1 0 value
0 0
0 1 p0
1 0 !p0
1 1 Pn
Fields - operands Compatibility

One or multiple registers can change type between register R7 and internal register RI2.

Different combinations of operand types are controlled by opN_sel.

This field controls position of opN operand which is a normal register.

Not all types of fields are applicable to each combination of registers.

opN_sel operands example
0 0 IR mov.f32 i0, r0
0 1 II mov.f32 i0, i0
1 0 II mov.f32 i0, i0
1 1 II2 mov.f32 i0, i2
opN_sel neg_op2i swz_alt_op2i_2 swz_alt_opN opN_swz neg_opN swz_alt_op1i op1i_swz swz_alt_op2i_x op2i_swz optN op0i abs_opN op2i op1i swz_mask2 swz_mask1 swz_en opN
0 0 1 1 1 1 1 1 1 1 1 1
0 1 1 1 1 1 1 1 1
1 0 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1
Fields - operands
Swizzle masking

Masking is controled by control bits:

  • control bits: swz_en, swz_mask1, swz_mask2

Each channel can be masked with control bits. Combinations of control bits produce the following masking table.

Encoding used in masking table:

value meaning
0 channel not selected
1 channel selected
x channel masked

Masking table:

swz_mask2 swz_mask1 swz_en ch0 ch1 ch2
0 0 0 0 0 0
0 0 1 1 0 0
0 1 0 x 1 0
0 1 1 1 1 0
1 0 0 x x 1
1 0 1 1 x 1
1 1 0 x 1 1
1 1 1 1 1 1
Swizzles - operand 0 i

Swizzles of operand 0 i can not be controled and have predefined combinations described below:

value
xyz

Each channel can be masked with control bits. Masking is described in Swizzle masking.

  • swz_en, swz_mask1, swz_mask2
Swizzles - operand 1 i

Swizzles of operand 1 i can not be precisely controlled and have predefined combinations described below and controlled by swizzle fields:

  • op1i_swz, swz_alt_op1i

Each channel can be masked with control bits. Masking is described in Swizzle masking.

  • swz_en, swz_mask1, swz_mask2
swz_alt_op1i op1i_swz value
0 0 0 0 xxx
0 0 0 1 yyy
0 0 1 0 zzz
0 0 1 1 www
0 1 0 0 xyz
0 1 0 1 yzw
0 1 1 0 xxy
0 1 1 1 xyx
1 0 0 0 yyx
1 0 0 1 yyz
1 0 1 0 zxy
1 0 1 1 xzy
1 1 0 0 {0, 0, 0}
1 1 0 1 {0.5, 0.5, 0.5}
1 1 1 0 {1, 1, 1}
1 1 1 1 {2, 2, 2}

When opN_sel is equal to 3 - operand 1 i swizzle can not be controlled and equals the following value:

value
xyz
Swizzles - operand 2 i

Control bits usually used for encoding operand 2 are used for encoding operand 1 since this instruction only has 2 operands.

Swizzles of operand 1 i can not be precisely controlled and have predefined combinations described below and controlled by swizzle fields:

  • op2i_swz, swz_alt_op2i_x, swz_alt_op2i_2

Each channel can be masked with control bits. Masking is described in Swizzle masking.

  • swz_en, swz_mask1, swz_mask2
swz_alt_op2i_2 swz_alt_op2i_x op2i_swz value
0 0 0 0 0 xxx
0 0 0 0 1 yyy
0 0 0 1 0 zzz
0 0 0 1 1 www
0 0 1 0 0 xyz
0 0 1 0 1 yzw
0 0 1 1 0 xxy
0 0 1 1 1 xyx
0 1 0 0 0 yyx
0 1 0 0 1 yyz
0 1 0 1 0 zxy
0 1 0 1 1 xzy
0 1 1 0 0 {0, 0, 0}
0 1 1 0 1 {0.5, 0.5, 0.5}
0 1 1 1 0 {1, 1, 1}
0 1 1 1 1 {2, 2, 2}
1 0 0 0 0 xyy
1 0 0 0 1 yxy
1 0 0 1 0 xxz
1 0 0 1 1 yxx
1 0 1 0 0 xy0
1 0 1 0 1 x10
1 0 1 1 0 xzy
1 0 1 1 1 yzx
1 1 0 0 0 zyx
1 1 0 0 1 zzy
1 1 0 1 0 xy1
1 1 0 1 1 {0.5, 0.5, 0.5}
1 1 1 0 0 {0.5, 0.5, 0.5}
1 1 1 0 1 {0.5, 0.5, 0.5}
1 1 1 1 0 {0.5, 0.5, 0.5}
1 1 1 1 1 {0.5, 0.5, 0.5}

When opN_sel is equal to 3 - operand 1 i swizzle can not be controlled and equals the following value:

value
xyz
Swizzles - operand N

Swizzles of operand N can not be precisely controlled and have predefined combinations described below and controlled by swizzle fields:

  • opN_swz, swz_alt_opN

Each channel can be masked with control bits. Masking is described in Swizzle masking.

  • swz_en, swz_mask1, swz_mask2
swz_alt_opN opN_swz value
0 0 0 0 xxx
0 0 0 1 yyy
0 0 1 0 zzz
0 0 1 1 www
0 1 0 0 xyz
0 1 0 1 yzw
0 1 1 0 xxy
0 1 1 1 xyx
1 0 0 0 yyx
1 0 0 1 yyz
1 0 1 0 zxy
1 0 1 1 xzy
1 1 0 0 {0, 0, 0}
1 1 0 1 {0.5, 0.5, 0.5}
1 1 1 0 {1, 1, 1}
1 1 1 1 {2, 2, 2}
Examples
mov.f32 i0, r0

mov.f16 i0, r0

MOV RR EXP

Encoding

Higher 4 bytes

7 6 5 4 3 2 1 0
opcode1
0 0 1 0 0 x x x
7 6 5 4 3 2 1 0
op_sel2
x 0 x x x x x x
7 6 5 4 3 2 1 0
opcode2 gr_sel op_sel1
x 1 0 1 x 1 x x
7 6 5 4 3 2 1 0
x x x x x x x x

Lower 4 bytes

7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
opcode3
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x

LOG IR

Encoding

Higher 4 bytes

7 6 5 4 3 2 1 0
opcode1 neg_op2i predicate
0 0 1 0 0
7 6 5 4 3 2 1 0
op_sel2 data_format swz_alt_op2i_2 swz_alt_opN opN_swz
x 1
7 6 5 4 3 2 1 0
neg_opN opcode2 gr_sel op_sel1 swz_alt_op1i
1 0 1 x 0
7 6 5 4 3 2 1 0
op1i_swz swz_alt_op2i_x op2i_swz opt0
x x

Lower 4 bytes

7 6 5 4 3 2 1 0
optN op0i op0
x x x x
7 6 5 4 3 2 1 0
op0 opcode3 opN_sel
x x x x x x
7 6 5 4 3 2 1 0
unused abs_opN op2i op1i swz_mask2 swz_mask1
x
7 6 5 4 3 2 1 0
swz_en opN
Fields - instruction

data_format:

5 value
0 f32
1 f16

predicate:

1 0 value
0 0
0 1 p0
1 0 !p0
1 1 Pn
Fields - operands Compatibility

One or multiple registers can change type between register R7 and internal register RI2.

Different combinations of operand types are controlled by opN_sel.

This field controls position of opN operand which is a normal register.

Not all types of fields are applicable to each combination of registers.

opN_sel operands example
0 0 IR log.f32 i0, r0.x
0 1 II log.f32 i0, i0.x
1 0 II log.f32 i0, i0.x
1 1 II2 log.f32 i0, i2.x
opN_sel neg_op2i swz_alt_op2i_2 swz_alt_opN opN_swz neg_opN swz_alt_op1i op1i_swz swz_alt_op2i_x op2i_swz optN op0i abs_opN op2i op1i swz_mask2 swz_mask1 swz_en opN
0 0 1 1 1 1 1 1 1 1 1 1
0 1 1 1 1 1 1 1 1
1 0 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1
Fields - operands
  • neg_op2i - negate op2i. encodes operand 1 i.
  • swz_alt_op2i_2 - not a standard swizzling for op2i. consult Swizzles - operand 2 i. encodes operand 1 i.
  • swz_alt_opN - not a standard swizzling for opN. consult Swizzles - operand N.
  • opN_swz - not a standard swizzling for opN. consult Swizzles - operand N.
  • neg_opN - negate opN.
  • swz_alt_op1i - not a standard swizzling for op1i. consult Swizzles - operand 1 i.
  • op1i_swz - not a standard swizzling for op1i. consult Swizzles - operand 1 i.
  • swz_alt_op2i_x - not a standard swizzling for op2i. consult Swizzles - operand 2 i. encodes operand 1 i.
  • op2i_swz - not a standard swizzling for op2i. consult Swizzles - operand 2 i. encodes operand 1 i.
  • opt0 - unused
  • optN - encoded with RS2.
  • op0i - encoded with RI2.
  • op0 - unused
  • abs_opN - add abs modifier to opN.
  • op2i - encoded with RI2. encodes operand 1 i.
  • op1i - encoded with RI2.
  • swz_mask2 - mask swizzle. consult Swizzle masking.
  • swz_mask1 - mask swizzle. consult Swizzle masking.
  • swz_en - enables usage of swizzling. consult Swizzle masking.
  • opN - encode opN operand selected by opN_sel as R7.
Swizzle masking

Masking is controled by control bits:

  • control bits: swz_en, swz_mask1, swz_mask2

Each channel can be masked with control bits. Combinations of control bits produce the following masking table.

Masking does not apply to operand 1 since it has explicit swizzling.

Encoding used in masking table:

value meaning
0 channel not selected
1 channel selected
x channel masked

Masking table:

swz_mask2 swz_mask1 swz_en ch0 ch1 ch2
0 0 0 0 0 0
0 0 1 1 0 0
0 1 0 x 1 0
0 1 1 1 1 0
1 0 0 x x 1
1 0 1 1 x 1
1 1 0 x 1 1
1 1 1 1 1 1
Swizzles - operand 0 i

Swizzles of operand 0 i can not be controled and have predefined combinations described below:

value
xyz

Each channel can be masked with control bits. Masking is described in Swizzle masking.

  • swz_en, swz_mask1, swz_mask2
Swizzles - operand 1 i

Swizzles of operand 1 i can not be controlled. Instead masking combinations are described below and controlled by swizzle fields:

  • op1i_swz, swz_alt_op1i

Standard masking does not apply to operand 1 since it has explicit swizzling.

swz_alt_op1i op1i_swz value
0 0 0 0 x
0 0 0 1 -y
0 0 1 0 --z
0 0 1 1 ---w
0 1 0 0 x
0 1 0 1 -y
0 1 1 0 x
0 1 1 1 x
1 0 0 0 -y
1 0 0 1 -y
1 0 1 0 --z
1 0 1 1 x
1 1 0 0 x
1 1 0 1 ---w
1 1 1 0 -y
1 1 1 1 --z

When opN_sel is equal to 3 - operand 1 i swizzle can not be controlled and equals the following value:

value
x
Swizzles - operand 2 i

Control bits usually used for encoding operand 2 are used for encoding operand 1 since this instruction only has 2 operands.

Swizzles of operand 1 i can not be controlled. Instead masking combinations are described below and controlled by swizzle fields:

  • op2i_swz, swz_alt_op2i_x, swz_alt_op2i_2

Standard masking does not apply to operand 1 since it has explicit swizzling.

swz_alt_op2i_2 swz_alt_op2i_x op2i_swz value
0 0 0 0 0 x
0 0 0 0 1 -y
0 0 0 1 0 --z
0 0 0 1 1 ---w
0 0 1 0 0 x
0 0 1 0 1 -y
0 0 1 1 0 x
0 0 1 1 1 x
0 1 0 0 0 -y
0 1 0 0 1 -y
0 1 0 1 0 --z
0 1 0 1 1 x
0 1 1 0 0 x
0 1 1 0 1 ---w
0 1 1 1 0 -y
0 1 1 1 1 --z
1 0 0 0 0 x
1 0 0 0 1 -y
1 0 0 1 0 x
1 0 0 1 1 -y
1 0 1 0 0 x
1 0 1 0 1 x
1 0 1 1 0 x
1 0 1 1 1 -y
1 1 0 0 0 --z
1 1 0 0 1 --z
1 1 0 1 0 x
1 1 0 1 1 ---w
1 1 1 0 0 ---w
1 1 1 0 1 ---w
1 1 1 1 0 ---w
1 1 1 1 1 ---w

When opN_sel is equal to 3 - operand 1 i swizzle can not be controlled and equals the following value:

value
x
Swizzles - operand N

Swizzles of operand N i can not be controlled. Instead masking combinations are described below and controlled by swizzle fields:

  • opN_swz, swz_alt_opN

Standard masking does not apply to operand N since it has explicit swizzling.

swz_alt_opN opN_swz value
0 0 0 0 x
0 0 0 1 -y
0 0 1 0 --z
0 0 1 1 ---w
0 1 0 0 x
0 1 0 1 -y
0 1 1 0 x
0 1 1 1 x
1 0 0 0 -y
1 0 0 1 -y
1 0 1 0 --z
1 0 1 1 x
1 1 0 0 x
1 1 0 1 ---w
1 1 1 0 -y
1 1 1 1 --z
Examples
log.f32 i0, r0.x

log.f16 i0, r0.x

LOG RR EXP

Encoding

Higher 4 bytes

7 6 5 4 3 2 1 0
opcode1
0 0 1 0 0 x x x
7 6 5 4 3 2 1 0
op_sel2
x 1 x x x x x x
7 6 5 4 3 2 1 0
opcode2 gr_sel op_sel1
x 1 0 1 x 1 x x
7 6 5 4 3 2 1 0
x x x x x x x x

Lower 4 bytes

7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
opcode3
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x

RSQ IR

Encoding

Higher 4 bytes

7 6 5 4 3 2 1 0
opcode1 neg_op2i predicate
0 0 1 0 0
7 6 5 4 3 2 1 0
op_sel2 data_format swz_alt_op2i_2 swz_alt_opN opN_swz
x 0
7 6 5 4 3 2 1 0
neg_opN opcode2 gr_sel op_sel1 swz_alt_op1i
1 1 0 x 0
7 6 5 4 3 2 1 0
op1i_swz swz_alt_op2i_x op2i_swz opt0
x x

Lower 4 bytes

7 6 5 4 3 2 1 0
optN op0i op0
x x x x
7 6 5 4 3 2 1 0
op0 opcode3 opN_sel
x x x x x x
7 6 5 4 3 2 1 0
unused abs_opN op2i op1i swz_mask2 swz_mask1
x
7 6 5 4 3 2 1 0
swz_en opN
Fields - instruction

data_format:

5 value
0 f32
1 f16

predicate:

1 0 value
0 0
0 1 p0
1 0 !p0
1 1 Pn
Fields - operands Compatibility

One or multiple registers can change type between register R7 and internal register RI2.

Different combinations of operand types are controlled by opN_sel.

This field controls position of opN operand which is a normal register.

Not all types of fields are applicable to each combination of registers.

opN_sel operands example
0 0 IR rsq.f32 i0, r0.x
0 1 II rsq.f32 i0, i0.x
1 0 II rsq.f32 i0, i0.x
1 1 II2 rsq.f32 i0, i2.x
opN_sel neg_op2i swz_alt_op2i_2 swz_alt_opN opN_swz neg_opN swz_alt_op1i op1i_swz swz_alt_op2i_x op2i_swz optN op0i abs_opN op2i op1i swz_mask2 swz_mask1 swz_en opN
0 0 1 1 1 1 1 1 1 1 1 1
0 1 1 1 1 1 1 1 1
1 0 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1
Fields - operands
  • neg_op2i - negate op2i. encodes operand 1 i.
  • swz_alt_op2i_2 - not a standard swizzling for op2i. consult Swizzles - operand 2 i. encodes operand 1 i.
  • swz_alt_opN - not a standard swizzling for opN. consult Swizzles - operand N.
  • opN_swz - not a standard swizzling for opN. consult Swizzles - operand N.
  • neg_opN - negate opN.
  • swz_alt_op1i - not a standard swizzling for op1i. consult Swizzles - operand 1 i.
  • op1i_swz - not a standard swizzling for op1i. consult Swizzles - operand 1 i.
  • swz_alt_op2i_x - not a standard swizzling for op2i. consult Swizzles - operand 2 i. encodes operand 1 i.
  • op2i_swz - not a standard swizzling for op2i. consult Swizzles - operand 2 i. encodes operand 1 i.
  • opt0 - unused
  • optN - encoded with RS2.
  • op0i - encoded with RI2.
  • op0 - unused
  • abs_opN - add abs modifier to opN.
  • op2i - encoded with RI2. encodes operand 1 i.
  • op1i - encoded with RI2.
  • swz_mask2 - mask swizzle. consult Swizzle masking.
  • swz_mask1 - mask swizzle. consult Swizzle masking.
  • swz_en - enables usage of swizzling. consult Swizzle masking.
  • opN - encode opN operand selected by opN_sel as R7.
Swizzle masking

Masking is controled by control bits:

  • control bits: swz_en, swz_mask1, swz_mask2

Each channel can be masked with control bits. Combinations of control bits produce the following masking table.

Masking does not apply to operand 1 since it has explicit swizzling.

Encoding used in masking table:

value meaning
0 channel not selected
1 channel selected
x channel masked

Masking table:

swz_mask2 swz_mask1 swz_en ch0 ch1 ch2
0 0 0 0 0 0
0 0 1 1 0 0
0 1 0 x 1 0
0 1 1 1 1 0
1 0 0 x x 1
1 0 1 1 x 1
1 1 0 x 1 1
1 1 1 1 1 1
Swizzles - operand 0 i

Swizzles of operand 0 i can not be controled and have predefined combinations described below:

value
xyz

Each channel can be masked with control bits. Masking is described in Swizzle masking.

  • swz_en, swz_mask1, swz_mask2
Swizzles - operand 1 i

Swizzles of operand 1 i can not be controlled. Instead masking combinations are described below and controlled by swizzle fields:

  • op1i_swz, swz_alt_op1i

Standard masking does not apply to operand 1 since it has explicit swizzling.

swz_alt_op1i op1i_swz value
0 0 0 0 x
0 0 0 1 -y
0 0 1 0 --z
0 0 1 1 ---w
0 1 0 0 x
0 1 0 1 -y
0 1 1 0 x
0 1 1 1 x
1 0 0 0 -y
1 0 0 1 -y
1 0 1 0 --z
1 0 1 1 x
1 1 0 0 x
1 1 0 1 ---w
1 1 1 0 -y
1 1 1 1 --z

When opN_sel is equal to 3 - operand 1 i swizzle can not be controlled and equals the following value:

value
x
Swizzles - operand 2 i

Control bits usually used for encoding operand 2 are used for encoding operand 1 since this instruction only has 2 operands.

Swizzles of operand 1 i can not be controlled. Instead masking combinations are described below and controlled by swizzle fields:

  • op2i_swz, swz_alt_op2i_x, swz_alt_op2i_2

Standard masking does not apply to operand 1 since it has explicit swizzling.

swz_alt_op2i_2 swz_alt_op2i_x op2i_swz value
0 0 0 0 0 x
0 0 0 0 1 -y
0 0 0 1 0 --z
0 0 0 1 1 ---w
0 0 1 0 0 x
0 0 1 0 1 -y
0 0 1 1 0 x
0 0 1 1 1 x
0 1 0 0 0 -y
0 1 0 0 1 -y
0 1 0 1 0 --z
0 1 0 1 1 x
0 1 1 0 0 x
0 1 1 0 1 ---w
0 1 1 1 0 -y
0 1 1 1 1 --z
1 0 0 0 0 x
1 0 0 0 1 -y
1 0 0 1 0 x
1 0 0 1 1 -y
1 0 1 0 0 x
1 0 1 0 1 x
1 0 1 1 0 x
1 0 1 1 1 -y
1 1 0 0 0 --z
1 1 0 0 1 --z
1 1 0 1 0 x
1 1 0 1 1 ---w
1 1 1 0 0 ---w
1 1 1 0 1 ---w
1 1 1 1 0 ---w
1 1 1 1 1 ---w

When opN_sel is equal to 3 - operand 1 i swizzle can not be controlled and equals the following value:

value
x
Swizzles - operand N

Swizzles of operand N i can not be controlled. Instead masking combinations are described below and controlled by swizzle fields:

  • opN_swz, swz_alt_opN

Standard masking does not apply to operand N since it has explicit swizzling.

swz_alt_opN opN_swz value
0 0 0 0 x
0 0 0 1 -y
0 0 1 0 --z
0 0 1 1 ---w
0 1 0 0 x
0 1 0 1 -y
0 1 1 0 x
0 1 1 1 x
1 0 0 0 -y
1 0 0 1 -y
1 0 1 0 --z
1 0 1 1 x
1 1 0 0 x
1 1 0 1 ---w
1 1 1 0 -y
1 1 1 1 --z
Examples
rsq.f32 i0, r0.x

rsq.f16 i0, r0.x

RSQ RR EXP

Encoding

Higher 4 bytes

7 6 5 4 3 2 1 0
opcode1
0 0 1 0 0 x x x
7 6 5 4 3 2 1 0
op_sel2
x 0 x x x x x x
7 6 5 4 3 2 1 0
opcode2 gr_sel op_sel1
x 1 1 0 x 1 x x
7 6 5 4 3 2 1 0
x x x x x x x x

Lower 4 bytes

7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
opcode3
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x

RCP IR

Encoding

Higher 4 bytes

7 6 5 4 3 2 1 0
opcode1 neg_op2i predicate
0 0 1 0 0
7 6 5 4 3 2 1 0
op_sel2 data_format swz_alt_op2i_2 swz_alt_opN opN_swz
x 0
7 6 5 4 3 2 1 0
neg_opN opcode2 gr_sel op_sel1 swz_alt_op1i
1 1 1 x 0
7 6 5 4 3 2 1 0
op1i_swz swz_alt_op2i_x op2i_swz opt0
x x

Lower 4 bytes

7 6 5 4 3 2 1 0
optN op0i op0
x x x x
7 6 5 4 3 2 1 0
op0 opcode3 opN_sel
x x x x x x
7 6 5 4 3 2 1 0
unused abs_opN op2i op1i swz_mask2 swz_mask1
x
7 6 5 4 3 2 1 0
swz_en opN
Fields - instruction

data_format:

5 value
0 f32
1 f16

predicate:

1 0 value
0 0
0 1 p0
1 0 !p0
1 1 Pn
Fields - operands Compatibility

One or multiple registers can change type between register R7 and internal register RI2.

Different combinations of operand types are controlled by opN_sel.

This field controls position of opN operand which is a normal register.

Not all types of fields are applicable to each combination of registers.

opN_sel operands example
0 0 IR rcp.f32 i0, r0.x
0 1 II rcp.f32 i0, i0.x
1 0 II rcp.f32 i0, i0.x
1 1 II2 rcp.f32 i0, i2.x
opN_sel neg_op2i swz_alt_op2i_2 swz_alt_opN opN_swz neg_opN swz_alt_op1i op1i_swz swz_alt_op2i_x op2i_swz optN op0i abs_opN op2i op1i swz_mask2 swz_mask1 swz_en opN
0 0 1 1 1 1 1 1 1 1 1 1
0 1 1 1 1 1 1 1 1
1 0 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1
Fields - operands
  • neg_op2i - negate op2i. encodes operand 1 i.
  • swz_alt_op2i_2 - not a standard swizzling for op2i. consult Swizzles - operand 2 i. encodes operand 1 i.
  • swz_alt_opN - not a standard swizzling for opN. consult Swizzles - operand N.
  • opN_swz - not a standard swizzling for opN. consult Swizzles - operand N.
  • neg_opN - negate opN.
  • swz_alt_op1i - not a standard swizzling for op1i. consult Swizzles - operand 1 i.
  • op1i_swz - not a standard swizzling for op1i. consult Swizzles - operand 1 i.
  • swz_alt_op2i_x - not a standard swizzling for op2i. consult Swizzles - operand 2 i. encodes operand 1 i.
  • op2i_swz - not a standard swizzling for op2i. consult Swizzles - operand 2 i. encodes operand 1 i.
  • opt0 - unused
  • optN - encoded with RS2.
  • op0i - encoded with RI2.
  • op0 - unused
  • abs_opN - add abs modifier to opN.
  • op2i - encoded with RI2. encodes operand 1 i.
  • op1i - encoded with RI2.
  • swz_mask2 - mask swizzle. consult Swizzle masking.
  • swz_mask1 - mask swizzle. consult Swizzle masking.
  • swz_en - enables usage of swizzling. consult Swizzle masking.
  • opN - encode opN operand selected by opN_sel as R7.
Swizzle masking

Masking is controled by control bits:

  • control bits: swz_en, swz_mask1, swz_mask2

Each channel can be masked with control bits. Combinations of control bits produce the following masking table.

Masking does not apply to operand 1 since it has explicit swizzling.

Encoding used in masking table:

value meaning
0 channel not selected
1 channel selected
x channel masked

Masking table:

swz_mask2 swz_mask1 swz_en ch0 ch1 ch2
0 0 0 0 0 0
0 0 1 1 0 0
0 1 0 x 1 0
0 1 1 1 1 0
1 0 0 x x 1
1 0 1 1 x 1
1 1 0 x 1 1
1 1 1 1 1 1
Swizzles - operand 0 i

Swizzles of operand 0 i can not be controled and have predefined combinations described below:

value
xyz

Each channel can be masked with control bits. Masking is described in Swizzle masking.

  • swz_en, swz_mask1, swz_mask2
Swizzles - operand 1 i

Swizzles of operand 1 i can not be controlled. Instead masking combinations are described below and controlled by swizzle fields:

  • op1i_swz, swz_alt_op1i

Standard masking does not apply to operand 1 since it has explicit swizzling.

swz_alt_op1i op1i_swz value
0 0 0 0 x
0 0 0 1 -y
0 0 1 0 --z
0 0 1 1 ---w
0 1 0 0 x
0 1 0 1 -y
0 1 1 0 x
0 1 1 1 x
1 0 0 0 -y
1 0 0 1 -y
1 0 1 0 --z
1 0 1 1 x
1 1 0 0 x
1 1 0 1 ---w
1 1 1 0 -y
1 1 1 1 --z

When opN_sel is equal to 3 - operand 1 i swizzle can not be controlled and equals the following value:

value
x
Swizzles - operand 2 i

Control bits usually used for encoding operand 2 are used for encoding operand 1 since this instruction only has 2 operands.

Swizzles of operand 1 i can not be controlled. Instead masking combinations are described below and controlled by swizzle fields:

  • op2i_swz, swz_alt_op2i_x, swz_alt_op2i_2

Standard masking does not apply to operand 1 since it has explicit swizzling.

swz_alt_op2i_2 swz_alt_op2i_x op2i_swz value
0 0 0 0 0 x
0 0 0 0 1 -y
0 0 0 1 0 --z
0 0 0 1 1 ---w
0 0 1 0 0 x
0 0 1 0 1 -y
0 0 1 1 0 x
0 0 1 1 1 x
0 1 0 0 0 -y
0 1 0 0 1 -y
0 1 0 1 0 --z
0 1 0 1 1 x
0 1 1 0 0 x
0 1 1 0 1 ---w
0 1 1 1 0 -y
0 1 1 1 1 --z
1 0 0 0 0 x
1 0 0 0 1 -y
1 0 0 1 0 x
1 0 0 1 1 -y
1 0 1 0 0 x
1 0 1 0 1 x
1 0 1 1 0 x
1 0 1 1 1 -y
1 1 0 0 0 --z
1 1 0 0 1 --z
1 1 0 1 0 x
1 1 0 1 1 ---w
1 1 1 0 0 ---w
1 1 1 0 1 ---w
1 1 1 1 0 ---w
1 1 1 1 1 ---w

When opN_sel is equal to 3 - operand 1 i swizzle can not be controlled and equals the following value:

value
x
Swizzles - operand N

Swizzles of operand N i can not be controlled. Instead masking combinations are described below and controlled by swizzle fields:

  • opN_swz, swz_alt_opN

Standard masking does not apply to operand N since it has explicit swizzling.

swz_alt_opN opN_swz value
0 0 0 0 x
0 0 0 1 -y
0 0 1 0 --z
0 0 1 1 ---w
0 1 0 0 x
0 1 0 1 -y
0 1 1 0 x
0 1 1 1 x
1 0 0 0 -y
1 0 0 1 -y
1 0 1 0 --z
1 0 1 1 x
1 1 0 0 x
1 1 0 1 ---w
1 1 1 0 -y
1 1 1 1 --z
Examples
rcp.f32 i0, r0.x

rcp.f16 i0, r0.x

RCP RR EXP

Encoding

Higher 4 bytes

7 6 5 4 3 2 1 0
opcode1
0 0 1 0 0 x x x
7 6 5 4 3 2 1 0
op_sel2
x 0 x x x x x x
7 6 5 4 3 2 1 0
opcode2 gr_sel op_sel1
x 1 1 1 x 1 x x
7 6 5 4 3 2 1 0
x x x x x x x x

Lower 4 bytes

7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
opcode3
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x

0x28000000 - 0x30000000

Instructions

mad, dot, mul, add, mov, rsq, rsp

Encoding

There is total of 8 instructions with 8 variations in this group. However instruction encoding is quite complex and is controlled by following fields:

  • opcode2 - controls opcodes in the page. there are 8 opcodes.
  • gr_sel - controls validity of instructions
  • op_sel1 - controls explicit swizzling
  • opcode3 - controls validity of instructions

Higher 4 bytes

7 6 5 4 3 2 1 0
opcode1
0 0 1 0 1 x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
opcode2 gr_sel op_sel1
x x x
7 6 5 4 3 2 1 0
x x x x x x x x

Lower 4 bytes

7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
opcode3
x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x
Allowed Instruction Encodings

Allowed encodings are defined by these tables.

opcode2 is a key for table Allowed Instructions

Join with this table to fill in instruction gaps.

gr_sel = 0

opcode2 gr_sel opcode3 instruction
0 0 0 0 x invalid
0 0 0 1 0 instruction
0 0 0 1 1 invalid
0 0 1 0 0 invalid
0 0 1 0 1 instruction
0 0 1 1 x instruction
1 0 0 0 0 invalid
1 0 0 0 1 instruction
1 0 0 1 x instruction
1 0 1 x x instruction
2 0 0 0 0 invalid
2 0 0 0 1 instruction
2 0 0 1 x instruction
2 0 1 x x instruction
3 0 0 0 0 invalid
3 0 0 0 1 instruction
3 0 0 1 x instruction
3 0 1 x x instruction
4 0 0 0 0 invalid
4 0 0 0 1 instruction
4 0 0 1 x instruction
4 0 1 x x instruction
5 0 0 0 0 invalid
5 0 0 0 1 instruction
5 0 0 1 x instruction
5 0 1 x x instruction
6 0 0 0 0 invalid
6 0 0 0 1 instruction
6 0 0 1 x instruction
6 0 1 x x instruction
7 0 0 0 0 invalid
7 0 0 0 1 instruction
7 0 0 1 x instruction
7 0 1 x x instruction

gr_sel = 1

opcode2 gr_sel opcode3 instruction
0 1 0 x x invalid
0 1 1 0 x instruction
0 1 1 1 x invalid
1 1 0 0 0 invalid
1 1 0 0 1 instruction
1 1 0 1 x instruction
1 1 1 0 x instruction
1 1 1 1 x invalid
2 1 0 x x instruction
2 1 1 0 x instruction
2 1 1 1 x invalid
3 1 0 0 0 invalid
3 1 0 0 1 instruction
3 1 0 1 x instruction
3 1 1 0 x instruction
3 1 1 1 x invalid
4 1 0 0 0 invalid
4 1 0 0 1 instruction
4 1 0 1 x instruction
4 1 1 0 x instruction
4 1 1 1 x invalid
5 1 0 x x instruction
5 1 1 0 x instruction
5 1 1 1 x invalid
6 1 0 x x instruction
6 1 1 0 x instruction
6 1 1 1 x invalid
7 1 0 x x instruction
7 1 1 0 x instruction
7 1 1 1 x invalid
Allowed Instructions

This table describes 8 instructions with 8 variations.

opcode2 is a key.

opcode2 op_sel1 instruction
0 0 MAD IRII
0 1 MAD RRII EXP
1 0 DOT IRI
1 1 DOT RRI EXP
2 0 DOT IRR
2 1 DOT RRR EXP
3 0 MUL IRI
3 1 MUL RRI EXP
4 0 ADD IRI
4 1 ADD RRI EXP
5 0 MOV IR
5 1 MOV RR EXP
6 0 RSQ IR
6 1 RSQ RR EXP
7 0 RCP IR
7 1 RCP RR EXP

Notes

None of 16 instructions are documented

MAD IRII

Encoding

Higher 4 bytes

7 6 5 4 3 2 1 0
opcode1 neg_op2i predicate
0 0 1 0 1
7 6 5 4 3 2 1 0
mask3 data_format swz_alt_op2i_2 swz_alt_opN opN_swz
x
7 6 5 4 3 2 1 0
neg_opN opcode2 gr_sel op_sel1 swz_alt_op1i
0 0 0 x 0
7 6 5 4 3 2 1 0
op1i_swz swz_alt_op2i_x op2i_swz opt0
x x

Lower 4 bytes

7 6 5 4 3 2 1 0
optN op0i op0
x x x x
7 6 5 4 3 2 1 0
op0 opcode3 opN_sel
x x x x x x
7 6 5 4 3 2 1 0
op3i op2i op1i swz_mask2 swz_mask1
7 6 5 4 3 2 1 0
swz_en opN
Fields - instruction

data_format:

5 value
0 f32
1 f16

predicate:

1 0 value
0 0
0 1 p0
1 0 !p0
1 1 Pn
Examples
mad.f32	i0, r0, i0, i0

MAD RRII EXP

Fields - instruction

data_format:

5 value
0 f32
1 f16

predicate:

1 0 value
0 0
0 1 p0
1 0 !p0
1 1 Pn
Examples
mad.f32	r0.xy, r0.xx, i0.xx, i0.xy

DOT IRI

Fields - instruction

data_format:

5 value
0 f32
1 f16

predicate:

1 0 value
0 0
0 1 p0
1 0 !p0
1 1 Pn
Examples
dot.f32	i0, r0.xxxx, i0.xxxx

DOT RRI EXP

Fields - instruction

data_format:

5 value
0 f32
1 f16

predicate:

1 0 value
0 0
0 1 p0
1 0 !p0
1 1 Pn
Examples
dot.f32	r0.x, r0.xxxx, i0.xxxx

DOT IRR

Fields - instruction

data_format:

5 value
0 f32
1 f16

predicate:

1 0 value
0 0
0 1 p0
1 0 !p0
1 1 Pn
Examples
dot.f32	i0, r0.xxxx, r0.xxxx

DOT RRR EXP

Fields - instruction

data_format:

5 value
0 f32
1 f16

predicate:

1 0 value
0 0
0 1 p0
1 0 !p0
1 1 Pn
Examples
dot.f32	r0.x, r0.xxxx, r0.xxxx

MUL IRI

Fields - instruction

data_format:

5 value
0 f32
1 f16

predicate:

1 0 value
0 0
0 1 p0
1 0 !p0
1 1 Pn
Examples
mul.f32	i0, r0, i0

MUL RRI EXP

Fields - instruction

data_format:

5 value
0 f32
1 f16

predicate:

1 0 value
0 0
0 1 p0
1 0 !p0
1 1 Pn
Examples
mul.f32	r0.xy, r0.xx, i0.xx

ADD IRI

Fields - instruction

data_format:

5 value
0 f32
1 f16

predicate:

1 0 value
0 0
0 1 p0
1 0 !p0
1 1 Pn
Examples
add.f32	i0, r0, i0

ADD RRI EXP

Fields - instruction

data_format:

5 value
0 f32
1 f16

predicate:

1 0 value
0 0
0 1 p0
1 0 !p0
1 1 Pn
Examples
add.f32	r0.xy, r0.xx, i0.xx

MOV IR

Fields - instruction

data_format:

5 value
0 f32
1 f16

predicate:

1 0 value
0 0
0 1 p0
1 0 !p0
1 1 Pn
Examples
mov.f32	i0, r0

MOV RR EXP

Fields - instruction

data_format:

5 value
0 f32
1 f16

predicate:

1 0 value
0 0
0 1 p0
1 0 !p0
1 1 Pn
Examples
mov.f32	r0.xy, r0.xx

RSQ IR

Fields - instruction

data_format:

5 value
0 f32
1 f16

predicate:

1 0 value
0 0
0 1 p0
1 0 !p0
1 1 Pn
Examples
rsq.f32	i0, r0.x

RSQ RR EXP

Fields - instruction

data_format:

5 value
0 f32
1 f16

predicate:

1 0 value
0 0
0 1 p0
1 0 !p0
1 1 Pn
Examples
rsq.f32	r0.x, r0.x

RCP IR

Fields - instruction

data_format:

5 value
0 f32
1 f16

predicate:

1 0 value
0 0
0 1 p0
1 0 !p0
1 1 Pn
Examples
rcp.f32	i0, r0.x

RCP RR EXP

Fields - instruction

data_format:

5 value
0 f32
1 f16

predicate:

1 0 value
0 0
0 1 p0
1 0 !p0
1 1 Pn
Examples
rcp.f32	r0.x, r0.x

0x30000000 - 0x38000000

Instructions: rcp, rsq, log, exp

Encoding:

7 6 5 4 3 2 1 0
opcode1 predicate
0 0 1 1 0
7 6 5 4 3 2 1 0
data_format
x x x x x x
7 6 5 4 3 2 1 0
opcode2 modifier
x x x x x
7 6 5 4 3 2 1 0
? ? ? ? ? ? ? ?

Notes:

modifier should be omitted if data_format matches modifier.

Fields:

opcode2:

2 1 value
0 0 rcp
0 1 rsq
1 0 log
1 1 exp

data_format:

6 5 value
0 0 f32
0 1 f16
1 0 fx10
1 1 invalid

modifier:

0 value
0 f32
1 fx10

predicate:

2 1 0 value
0 0 0
0 0 1 p0
0 1 0 p1
0 1 1 p2
1 0 0 p3
1 0 1 !p0
1 1 0 !p1
1 1 1 Pn

Examples:

rcp.f32
rsq.f32
log.f32
exp.f32

rcp.f32.fx10
rsq.f32.fx10
log.f32.fx10
exp.f32.fx10

rcp.f16.f32
rsq.f16.f32
log.f16.f32
exp.f16.f32

rcp.f16.fx10
rsq.f16.fx10
log.f16.fx10
exp.f16.fx10

rcp.fx10.f32
rsq.fx10.f32
log.fx10.f32
exp.fx10.f32

rcp.fx10
rsq.fx10
log.fx10
exp.fx10

0x38000000 - 0x40000000

Instructions: mov, cmov, cmov8

Encoding:

7 6 5 4 3 2 1 0
opcode1 predicate
0 0 1 1 1
7 6 5 4 3 2 1 0
cond
x x x x x x x
7 6 5 4 3 2 1 0
opcode2 data_format
x x x
7 6 5 4 3 2 1 0
? ? ? ? ? ? ? ?

Notes:

cond is only applicable to cmov and cmov8 since this is conditional move.

Fields:

opcode2:

7 6 value
0 0 mov
0 1 cmov
1 0 cmov8
1 1 invalid

cond:

6 value
0 eqzero
1 ltzero

data_format:

2 1 0 value
0 0 0 i8
0 0 1 i16
0 1 0 i32
0 1 1 fx10
1 0 0 f16
1 0 1 f32
1 1 0 invalid
1 1 1 invalid

predicate:

2 1 0 value
0 0 0
0 0 1 p0
0 1 0 p1
0 1 1 p2
1 0 0 p3
1 0 1 !p0
1 1 0 !p1
1 1 1 Pn

Examples:

mov.i8
mov.i16
mov.i32
mov.fx10
mov.f16
mov.f32

cmov.eqzero.i8
cmov.eqzero.i16
cmov.eqzero.i32
cmov.eqzero.fx10
cmov.eqzero.f16
cmov.eqzero.f32

cmov8.eqzero.i8
cmov8.eqzero.i16
cmov8.eqzero.i32
cmov8.eqzero.fx10
cmov8.eqzero.f16
cmov8.eqzero.f32

cmov.ltzero.i8
cmov.ltzero.i16
cmov.ltzero.i32
cmov.ltzero.fx10
cmov.ltzero.f16
cmov.ltzero.f32

cmov8.ltzero.i8
cmov8.ltzero.i16
cmov8.ltzero.i32
cmov8.ltzero.fx10
cmov8.ltzero.f16
cmov8.ltzero.f32

0x40000000 - 0x48000000

Instructions: pack, (mov)

Encoding:

7 6 5 4 3 2 1 0
opcode1 predicate
0 1 0 0 0
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
modifier data_format
x x x x
7 6 5 4 3 2 1 0
? ? ? ? ? ? ? ?

Notes:

when modifier matches data_format it shall be omitted since it has no effect in terms of packing.

furthermore instruction mnemonic shall be replaced to mov

Fields:

data_format:

0 value
0 u8
1 s16

modifier:

3 2 1 value
0 0 0 u8
0 0 1 s8
0 1 0 o8
0 1 1 u16
1 0 0 s16
1 0 1 f16
1 1 0 f32
1 1 1 invalid

predicate:

2 1 0 value
0 0 0
0 0 1 p0
0 1 0 p1
0 1 1 p2
1 0 0 p3
1 0 1 !p0
1 1 0 !p1
1 1 1 Pn

Examples:

mov.u8
pack.s16.u8
pack.u8.s8
pack.s16.s8
pack.u8.o8
pack.s16.o8
pack.u8.u16
pack.s16.u16
pack.u8.s16
mov.s16
pack.u8.f16
pack.s16.f16
pack.u8.f32
pack.s16.f32

0x48000000 - 0x50000000

Instructions: this group only contains illegal instructions

Encoding:

7 6 5 4 3 2 1 0
opcode1
0 1 0 0 1 x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
? ? ? ? ? ? ? ?

0x50000000 - 0x58000000

Instructions: and.u32

Encoding:

7 6 5 4 3 2 1 0
opcode1 predicate
0 1 0 1 0
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
? ? ? ? ? ? ? ?

Fields:

predicate:

2 1 0 value
0 0 0
0 0 1 p0
0 1 0 p1
0 1 1 p2
1 0 0 p3
1 0 1 !p0
1 1 0 !p1
1 1 1 Pn

Examples:

and.u32

0x58000000 - 0x60000000

Instructions: xor.u32

Encoding:

7 6 5 4 3 2 1 0
opcode1 predicate
0 1 0 1 1
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
? ? ? ? ? ? ? ?

Fields:

predicate:

2 1 0 value
0 0 0
0 0 1 p0
0 1 0 p1
0 1 1 p2
1 0 0 p3
1 0 1 !p0
1 1 0 !p1
1 1 1 Pn

Examples:

xor.u32

0x60000000 - 0x68000000

Instructions: shl.u32

Encoding:

7 6 5 4 3 2 1 0
opcode1 predicate
0 1 1 0 0
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
? ? ? ? ? ? ? ?

Fields:

predicate:

2 1 0 value
0 0 0
0 0 1 p0
0 1 0 p1
0 1 1 p2
1 0 0 p3
1 0 1 !p0
1 1 0 !p1
1 1 1 Pn

Examples:

shl.u32

0x68000000 - 0x70000000

Instructions: shr.u32

Encoding:

7 6 5 4 3 2 1 0
opcode1 predicate
0 1 1 0 1 x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
? ? ? ? ? ? ? ?

Fields:

predicate:

2 1 0 value
0 0 0
0 0 1 p0
0 1 0 p1
0 1 1 p2
1 0 0 p3
1 0 1 !p0
1 1 0 !p1
1 1 1 Pn

Examples:

shr.u32

0x70000000 - 0x78000000

Instructions: rlp.u32

Encoding:

7 6 5 4 3 2 1 0
opcode1 predicate
0 1 1 1 0 x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
? ? ? ? ? ? ? ?

Fields:

predicate:

2 1 0 value
0 0 0
0 0 1 p0
0 1 0 p1
0 1 1 p2
1 0 0 p3
1 0 1 !p0
1 1 0 !p1
1 1 1 Pn

Examples:

rlp.u32

0x78000000 - 0x80000000

Instructions: this group only contains illegal instructions

Encoding:

7 6 5 4 3 2 1 0
opcode1
0 1 1 1 1 x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
? ? ? ? ? ? ? ?

0x80000000 - 0x88000000

Instructions: add.fx8

Encoding:

7 6 5 4 3 2 1 0
opcode1 predicate
1 0 0 0 0 x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
? ? ? ? ? ? ? ?

Fields:

predicate:

2 1 value
0 0
0 1 p0
1 0 p1
1 1 !p0

Examples:

add.fx8

0x88000000 - 0x90000000

Instructions: add.fx8, sub.fx8

Encoding:

7 6 5 4 3 2 1 0
opcode1 predicate
1 0 0 0 1 x
7 6 5 4 3 2 1 0
opcode2
x x x x x x
7 6 5 4 3 2 1 0
x x x x 0 0 x x
7 6 5 4 3 2 1 0
? ? ? ? ? ? ? ?

Notes:

Having bits 2, 3 in byte 2 set to 1 produces invalid instruction

Fields:

opcode2:

5 4 value
0 0 add.fx8
0 1 sub.fx8
1 0 invalid
1 1 invalid

predicate:

2 1 value
0 0
0 1 p0
1 0 p1
1 1 !p0

Examples:

add.fx8

sub.fx8

0x90000000 - 0x98000000

Instructions: add.fx8, sub.fx8, min.fx8, max.fx8

Encoding:

7 6 5 4 3 2 1 0
opcode1 predicate
1 0 0 1 0 x
7 6 5 4 3 2 1 0
opcode2
x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x 0
7 6 5 4 3 2 1 0
? ? ? ? ? ? ? ?

Notes:

Having bit 0 in byte 2 set to 1 produces invalid instruction

Fields:

opcode2:

5 4 value
0 0 add.fx8
0 1 sub.fx8
1 0 min.fx8
1 1 max.fx8

predicate:

2 1 value
0 0
0 1 p0
1 0 p1
1 1 !p0

Examples:

add.fx8

sub.fx8

min.fx8

max.fx8

0x98000000 - 0xA0000000

Instructions: mad.u8

Encoding:

7 6 5 4 3 2 1 0
opcode1 predicate
1 0 0 1 1 x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
modifier
x x x x x x x
7 6 5 4 3 2 1 0
? ? ? ? ? ? ? ?

Fields:

modifier:

3 value
0
1 sat

predicate:

2 1 value
0 0
0 1 p0
1 0 p1
1 1 !p0

Examples:

mad.u8
mad.sat.u8

0xA0000000 - 0xA8000000

Instructions: mad

Encoding:

7 6 5 4 3 2 1 0
opcode1 predicate
1 0 1 0 0 x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
data_format modifier
x x x x x x
7 6 5 4 3 2 1 0
? ? ? ? ? ? ? ?

Fields:

data_format:

3 value
0 u16
1 i16

modifier:

2 value
0
1 sat

predicate:

2 1 value
0 0
0 1 p0
1 0 p1
1 1 !p0

Examples:

mad.u16
mad.u16.sat
mad.i16
mad.i16.sat

0xA8000000 - 0xB0000000

Instructions: mad

Encoding:

7 6 5 4 3 2 1 0
opcode1 predicate
1 0 1 0 1 x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
data_format modifier
x x x x x x
7 6 5 4 3 2 1 0
? ? ? ? ? ? ? ?

Fields:

data_format:

3 value
0 u32
1 i32

modifier:

2 value
0
1 sat

predicate:

2 1 value
0 0
0 1 p0
1 0 p1
1 1 !p0

Examples:

mad.u32
mad.u32.sat
mad.i32
mad.i32.sat

0xB0000000 - 0xB8000000

Instructions: this group only contains illegal instructions

Encoding:

7 6 5 4 3 2 1 0
opcode1
1 0 1 1 0 x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
? ? ? ? ? ? ? ?

0xB8000000 - 0xC0000000

Instructions: this group only contains illegal instructions

Encoding:

7 6 5 4 3 2 1 0
opcode1
1 0 1 1 1 x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
? ? ? ? ? ? ? ?

0xC0000000 - 0xC8000000

Instructions: this group only contains illegal instructions

Encoding:

7 6 5 4 3 2 1 0
opcode1
1 1 0 0 0 x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
? ? ? ? ? ? ? ?

0xC8000000 - 0xD0000000

Instructions: mad.u8

Encoding:

7 6 5 4 3 2 1 0
opcode1 predicate
1 1 0 0 1 x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
modifier
x x x x x x x
7 6 5 4 3 2 1 0
? ? ? ? ? ? ? ?

Fields:

modifier:

3 value
0
1 sat

predicate:

2 1 value
0 0
0 1 p0
1 0 p1
1 1 !p0

Examples:

mad.u8
mad.sat.u8

0xD0000000 - 0xD8000000

Instructions: mad

Encoding:

7 6 5 4 3 2 1 0
opcode1 predicate
1 1 0 1 0
7 6 5 4 3 2 1 0
modifier
x x 0 x x x x
7 6 5 4 3 2 1 0
data_format
x x x x x x x
7 6 5 4 3 2 1 0
? ? ? ? ? ? ? ?

Notes:

Having bit 5 in byte 1 set to 1 produces invalid instruction

Fields:

modifier:

4 value
0 s0
1 s1

data_format:

1 value
0 u32
1 i32

predicate:

2 1 0 value
0 0 0
0 0 1 p0
0 1 0 p1
0 1 1 p2
1 0 0 p3
1 0 1 !p0
1 1 0 !p1
1 1 1 Pn

Examples:

mad.u32.s0
mad.i32.s0

mad.u32.s1
mad.i32.s1

0xD8000000 - 0xE0000000

Instructions: this group only contains illegal instructions

Encoding:

7 6 5 4 3 2 1 0
opcode1
1 1 0 1 1 x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
x x x x x x x x
7 6 5 4 3 2 1 0
? ? ? ? ? ? ? ?

0xE0000000 - 0xE8000000

Instructions: tex

Encoding:

7 6 5 4 3 2 1 0
opcode1 predicate
1 1 1 0 0 0 0 0
7 6 5 4 3 2 1 0
modifier
x x x x x x x
7 6 5 4 3 2 1 0
data_format dim func
x x
7 6 5 4 3 2 1 0
? ? ? ? ? ? ? ?

Fields:

dim:

3 2 value
0 0 1D
0 1 2D
1 0 Cube
1 1 invalid

func:

1 0 value
0 0
0 1 Bias
1 0 Replace
1 1 Grad

modifier:

3 value
0
1 minp

data_format:

7 6 value
0 0
0 1 invalid
1 0 f16
1 1 f32

predicate:

2 1 0 value
0 0 0
0 0 1 p0
0 1 0 p1
0 1 1 p2
1 0 0 p3
1 0 1 !p0
1 1 0 !p1
1 1 1 Pn

Examples:

tex1D
tex1D.f16
tex1D.f32
tex1D.minp
tex1D.minp.f16
tex1D.minp.f32

tex1DBias
tex1DBias.f16
tex1DBias.f32
tex1DBias.minp
tex1DBias.minp.f16
tex1DBias.minp.f32

tex1DReplace
tex1DReplace.f16
tex1DReplace.f32
tex1DReplace.minp
tex1DReplace.minp.f16
tex1DReplace.minp.f32

tex1DGrad
tex1DGrad.f16
tex1DGrad.f32
tex1DGrad.minp
tex1DGrad.minp.f16
tex1DGrad.minp.f32

tex2D
tex2D.f16
tex2D.f32
tex2D.minp
tex2D.minp.f16
tex2D.minp.f32

tex2DBias
tex2DBias.f16
tex2DBias.f32
tex2DBias.minp
tex2DBias.minp.f16
tex2DBias.minp.f32

tex2DReplace
tex2DReplace.f16
tex2DReplace.f32
tex2DReplace.minp
tex2DReplace.minp.f16
tex2DReplace.minp.f32

tex2DGrad
tex2DGrad.f16
tex2DGrad.f32
tex2DGrad.minp
tex2DGrad.minp.f16
tex2DGrad.minp.f32

texCube
texCube.f16
texCube.f32
texCube.minp
texCube.minp.f16
texCube.minp.f32

texCubeBias
texCubeBias.f16
texCubeBias.f32
texCubeBias.minp
texCubeBias.minp.f16
texCubeBias.minp.f32

texCubeReplace
texCubeReplace.f16
texCubeReplace.f32
texCubeReplace.minp
texCubeReplace.minp.f16
texCubeReplace.minp.f32

texCubeGrad
texCubeGrad.f16
texCubeGrad.f32
texCubeGrad.minp
texCubeGrad.minp.f16
texCubeGrad.minp.f32

0xE8000000 - 0xF0000000

Instructions: lda32, ldl32, ldt32

Encoding:

7 6 5 4 3 2 1 0
opcode1 predicate
1 1 1 0 1
7 6 5 4 3 2 1 0
modifier
x x x x x x x
7 6 5 4 3 2 1 0
index opcode2
x x
7 6 5 4 3 2 1 0
? ? ? ? ? ? ? ?

Notes:

index is only applicable when fetch modifier is specified

Fields:

modifier:

5 value
0 fetch
1

index:

7 6 5 4 value
0 0 0 0 1
0 0 0 1 2
0 0 1 0 3
0 0 1 1 4
0 1 0 0 5
0 1 0 1 6
0 1 1 0 7
0 1 1 1 8
1 0 0 0 9
1 0 0 1 10
1 0 1 0 11
1 0 1 1 12
1 1 0 0 13
1 1 0 1 14
1 1 1 0 15
1 1 1 1 16

opcode2:

3 2 value
0 0 lda32
0 1 ldl32
1 0 ldt32
1 1 invalid

predicate:

2 1 0 value
0 0 0
0 0 1 p0
0 1 0 p1
0 1 1 p2
1 0 0 p3
1 0 1 !p0
1 1 0 !p1
1 1 1 Pn

Examples:

lda32
ldl32
ldt32

lda32.fetch1
lda32.fetch2
lda32.fetch3
lda32.fetch4
lda32.fetch5
lda32.fetch6
lda32.fetch7
lda32.fetch8
lda32.fetch9
lda32.fetch10
lda32.fetch11
lda32.fetch12
lda32.fetch13
lda32.fetch14
lda32.fetch15
lda32.fetch16

ldl32.fetch1
ldl32.fetch2
ldl32.fetch3
ldl32.fetch4
ldl32.fetch5
ldl32.fetch6
ldl32.fetch7
ldl32.fetch8
ldl32.fetch9
ldl32.fetch10
ldl32.fetch11
ldl32.fetch12
ldl32.fetch13
ldl32.fetch14
ldl32.fetch15
ldl32.fetch16

ldt32.fetch1
ldt32.fetch2
ldt32.fetch3
ldt32.fetch4
ldt32.fetch5
ldt32.fetch6
ldt32.fetch7
ldt32.fetch8
ldt32.fetch9
ldt32.fetch10
ldt32.fetch11
ldt32.fetch12
ldt32.fetch13
ldt32.fetch14
ldt32.fetch15
ldt32.fetch16

0xF0000000 - 0xF8000000

Instructions: sta32, stl32, stt32

Encoding:

7 6 5 4 3 2 1 0
opcode1 predicate
1 1 1 1 0
7 6 5 4 3 2 1 0
modifier
x x x x x x x
7 6 5 4 3 2 1 0
index opcode2
x x
7 6 5 4 3 2 1 0
? ? ? ? ? ? ? ?

Notes:

index is only applicable when fetch modifier is specified

Fields:

modifier:

5 value
0 fetch
1

index:

7 6 5 4 value
0 0 0 0 1
0 0 0 1 2
0 0 1 0 3
0 0 1 1 4
0 1 0 0 5
0 1 0 1 6
0 1 1 0 7
0 1 1 1 8
1 0 0 0 9
1 0 0 1 10
1 0 1 0 11
1 0 1 1 12
1 1 0 0 13
1 1 0 1 14
1 1 1 0 15
1 1 1 1 16

opcode2:

3 2 value
0 0 sta32
0 1 stl32
1 0 stt32
1 1 invalid

predicate:

2 1 0 value
0 0 0
0 0 1 p0
0 1 0 p1
0 1 1 p2
1 0 0 p3
1 0 1 !p0
1 1 0 !p1
1 1 1 Pn

Examples:

sta32
stl32
stt32

sta32.fetch1
sta32.fetch2
sta32.fetch3
sta32.fetch4
sta32.fetch5
sta32.fetch6
sta32.fetch7
sta32.fetch8
sta32.fetch9
sta32.fetch10
sta32.fetch11
sta32.fetch12
sta32.fetch13
sta32.fetch14
sta32.fetch15
sta32.fetch16

stl32.fetch1
stl32.fetch2
stl32.fetch3
stl32.fetch4
stl32.fetch5
stl32.fetch6
stl32.fetch7
stl32.fetch8
stl32.fetch9
stl32.fetch10
stl32.fetch11
stl32.fetch12
stl32.fetch13
stl32.fetch14
stl32.fetch15
stl32.fetch16

stt32.fetch1
stt32.fetch2
stt32.fetch3
stt32.fetch4
stt32.fetch5
stt32.fetch6
stt32.fetch7
stt32.fetch8
stt32.fetch9
stt32.fetch10
stt32.fetch11
stt32.fetch12
stt32.fetch13
stt32.fetch14
stt32.fetch15
stt32.fetch16

0xF8000000 - 0xFF000000

Notes:

this instruction group is much more complex than others so description is given in form of "glued" truth tables instead of independent truth tables.

predicate 000

Instructions:

Encoding:

7 6 5 4 3 2 1 0
opcode1 predicate
1 1 1 1 1 0 0 0
7 6 5 4 3 2 1 0
modifier1 opcode2
x x x x
7 6 5 4 3 2 1 0
opcode4 modifier2 opcode3
x x x x x
7 6 5 4 3 2 1 0
? ? ? ? ? ? ? ?

Fields:

opcode2:

predicate modifier1 opcode3 opcode4 modifier2 5 4 3 value
0 0 0 0 0 x 0 0 0 0 ba
0 0 0 0 1 x 0 0 0 0 mov
0 0 0 0 0 x 1 0 0 0 ba.savelink
0 0 0 0 1 x 1 0 0 0 mov
0 0 0 0 0 x x 0 0 1 mov.f32
0 0 0 0 1 x x 0 0 1 mov
0 0 0 0 x x x 0 1 x mov.f32
0 0 0 0 x x x 1 0 x mov.f32
0 0 0 0 x 0 x 1 1 x pcoeff
0 0 0 0 x 1 x 1 1 x ptoff
predicate modifier1 opcode4 5 4 3 value
0 0 0 1 x 0 0 x invalid
0 0 0 1 x 0 1 x invalid
0 0 0 1 x 1 0 x mov.f32
0 0 0 1 0 1 1 x pcoeff
0 0 0 1 1 1 1 x ptoff

Examples:

predicate 001

Instructions:

Encoding

7 6 5 4 3 2 1 0
opcode1 predicate
1 1 1 1 1 0 0 1
7 6 5 4 3 2 1 0
modifier1 opcode2
x x x x
7 6 5 4 3 2 1 0
sub_pred
x x x x x x
modifier2 opcode3
x x x x x x
7 6 5 4 3 2 1 0
? ? ? ? ? ? ? ?

Notes:

predicate does not apply to all instructions

Fields

opcode2:

predicate modifier1 opcode3 modifier2 sub_pred 5 4 3 value
0 0 1 0 0 0 x x 0 0 0 p0 ba
0 0 1 0 1 0 x x 0 0 0 p0 mov
0 0 1 0 0 1 x x 0 0 0 p0 ba.savelink
0 0 1 0 1 1 x x 0 0 0 p0 mov
0 0 1 0 0 x x x 0 0 1 mov.f32
0 0 1 0 1 x x x 0 0 1 p0 mov
0 0 1 0 x x x x 0 1 x mov.f32
0 0 1 0 x x x x 1 0 x mov.f32
0 0 1 0 x x 0 0 1 1 x kill
0 0 1 0 x x 0 1 1 1 x !p0 kill
0 0 1 0 x x 1 0 1 1 x !p1 kill
0 0 1 0 x x 1 1 1 1 x p0 kill
predicate modifier1 sub_pred 5 4 3 value
0 0 1 1 x x 0 0 x invalid
0 0 1 1 x x 0 1 x invalid
0 0 1 1 x x 1 0 x mov.f32
0 0 1 1 0 0 1 1 x kill
0 0 1 1 0 1 1 1 x !p0 kill
0 0 1 1 1 0 1 1 x !p1 kill
0 0 1 1 1 1 1 1 x p0 kill

Examples:

predicate 010

Instructions:

Encoding:

7 6 5 4 3 2 1 0
opcode1 predicate
1 1 1 1 1 0 1 0
7 6 5 4 3 2 1 0
modifier1 opcode2 opcode4
x x
modifier1 opcode2
x x x
7 6 5 4 3 2 1 0
opcode5 opcode6
x x x
modifier2 opcode3
x x x x x x
7 6 5 4 3 2 1 0
? ? ? ? ? ? ? ?

Notes:

predicate does not apply to all instructions

Fields

opcode2

predicate modifier1 opcode3 modifier2 5 4 3 value
0 1 0 0 0 0 0 0 0 0 p1 ba
0 1 0 0 0 1 0 0 0 0 p1 mov
0 1 0 0 0 0 1 0 0 0 p1 ba.savelink
0 1 0 0 0 1 1 0 0 0 p1 mov
0 1 0 0 0 0 x 0 0 1 mov.f32
0 1 0 0 0 1 x 0 0 1 p1 mov
0 1 0 0 0 x x 0 1 x mov.f32
0 1 0 0 0 x x 1 0 x mov.f32
0 1 0 0 0 x x 1 1 x invalid
predicate modifier1 opcode4 opcode5 opcode6 5 4 3 value
0 1 0 0 1 0 x x x x x 0 0 x mov.f32
0 1 0 0 1 1 0 x 0 0 x 0 0 x mov.f32
0 1 0 0 1 1 0 x 0 1 0 0 0 x mov.f32
0 1 0 0 1 1 0 x 0 1 1 0 0 x invalid
0 1 0 0 1 1 0 x 1 0 x 0 0 x invalid
0 1 0 0 1 1 0 x 1 1 0 0 0 x invalid
0 1 0 0 1 1 0 x 1 1 1 0 0 x mov.f32
0 1 0 0 1 1 1 0 0 0 x 0 0 x mov.f32
0 1 0 0 1 1 1 0 0 1 0 0 0 x mov.f32
0 1 0 0 1 1 1 0 0 1 1 0 0 x invalid
0 1 0 0 1 1 1 0 1 0 x 0 0 x invalid
0 1 0 0 1 1 1 0 1 1 0 0 0 x invalid
0 1 0 0 1 1 1 0 1 1 1 0 0 x mov.f32
0 1 0 0 1 1 1 1 x x x 0 0 x invalid
0 1 0 0 1 x x x x x x 0 1 x invalid
0 1 0 0 1 x x x x x x 1 0 x mov.f32
0 1 0 0 1 x x x x x x 1 1 x invalid
predicate modifier1 opcode3 modifier2 5 4 3 value
0 1 0 1 0 0 0 0 0 0 p1 ba
0 1 0 1 0 1 0 0 0 0 p1 mov
0 1 0 1 0 0 1 0 0 0 p1 ba.savelink
0 1 0 1 0 1 1 0 0 0 p1 mov
0 1 0 1 0 0 x 0 0 1 mov.f32
0 1 0 1 0 1 x 0 0 1 p1 mov
0 1 0 1 0 x x 0 1 x mov.f32
0 1 0 1 0 x x 1 0 x mov.f32
0 1 0 1 0 x x 1 1 x invalid
predicate modifier1 5 4 3 value
0 1 0 1 1 0 0 x mov.f32
0 1 0 1 1 0 1 x invalid
0 1 0 1 1 1 0 x mov.f32
0 1 0 1 1 1 1 x invalid

Examples:

predicate 011

Instructions:

Encoding

7 6 5 4 3 2 1 0
opcode1 predicate
1 1 1 1 1 0 1 1
7 6 5 4 3 2 1 0
modifier1 opcode2
x x x x
7 6 5 4 3 2 1 0
sub_pred
x x x x x x
opcode4 opcode5 modifier2 opcode3
x x
7 6 5 4 3 2 1 0
? ? ? ? ? ? ? ?

Notes:

predicate does not apply to all instructions

Fields

opcode2

predicate modifier1 opcode3 modifier2 opcode4 opcode5 sub_pred 5 4 3 value
0 1 1 0 0 0 x x x x x x 0 0 0 p2 ba
0 1 1 0 1 0 x x x x x x 0 0 0 p2 mov
0 1 1 0 0 1 x x x x x x 0 0 0 p2 ba.savelink
0 1 1 0 1 1 x x x x x x 0 0 0 p2 mov
0 1 1 0 0 x x x x x x x 0 0 1 mov.f32
0 1 1 0 1 x x x x x x x 0 0 1 p2 mov
0 1 1 0 x x x x x x x x 0 1 x mov.f32
0 1 1 0 x x 0 0 0 x x x 1 0 x mov.f32
0 1 1 0 x x 0 0 1 x x x 1 0 x mov.f32
0 1 1 0 x x 0 1 0 x x x 1 0 x mov.f32
0 1 1 0 x x 0 1 1 0 x x 1 0 x mov.f32
0 1 1 0 x x 0 1 1 1 x x 1 0 x invalid
0 1 1 0 x x 1 0 0 x x x 1 0 x mov.f32
0 1 1 0 x x 1 0 1 x x x 1 0 x mov.f32
0 1 1 0 x x 1 1 0 x x x 1 0 x invalid
0 1 1 0 x x 1 1 1 x x x 1 0 x invalid
0 1 1 0 x x x x x x 0 0 1 1 x depthf
0 1 1 0 x x x x x x 0 1 1 1 x p0 depthf
0 1 1 0 x x x x x x 1 0 1 1 x p1 depthf
0 1 1 0 x x x x x x 1 1 1 1 x !p0 depthf
predicate modifier1 opcode4 opcode5 sub_pred 5 4 3 value
0 1 1 1 x x x x x x 0 0 x invalid
0 1 1 1 x x x x x x 0 1 x invalid
0 1 1 1 0 0 0 x x x 1 0 x mov.f32
0 1 1 1 0 0 1 x x x 1 0 x mov.f32
0 1 1 1 0 1 0 x x x 1 0 x mov.f32
0 1 1 1 0 1 1 0 x x 1 0 x mov.f32
0 1 1 1 0 1 1 1 x x 1 0 x invalid
0 1 1 1 1 0 0 x x x 1 0 x mov.f32
0 1 1 1 1 0 1 x x x 1 0 x mov.f32
0 1 1 1 1 1 0 x x x 1 0 x invalid
0 1 1 1 1 1 1 x x x 1 0 x invalid
0 1 1 1 x x x x 0 0 1 1 x depthf
0 1 1 1 x x x x 0 1 1 1 x p0 depthf
0 1 1 1 x x x x 1 0 1 1 x p1 depthf
0 1 1 1 x x x x 1 1 1 1 x !p0 depthf

Examples:

predicate 100

Instructions:

Encoding

7 6 5 4 3 2 1 0
opcode1 predicate
1 1 1 1 1 1 0 0
7 6 5 4 3 2 1 0
modifier1 opcode2
x x x x
7 6 5 4 3 2 1 0
sub_pred
x x x x x
modifier2 opcode3
x x x x x x
7 6 5 4 3 2 1 0
? ? ? ? ? ? ? ?

Notes:

predicate does not apply to all instructions

Fields

opcode2:

predicate modifier1 opcode3 modifier2 sub_pred 5 4 3 value
1 0 0 0 0 0 x x x 0 0 0 p3 ba
1 0 0 0 1 0 x x x 0 0 0 p3 mov
1 0 0 0 0 1 x x x 0 0 0 p3 ba.savelink
1 0 0 0 1 1 x x x 0 0 0 p3 mov
1 0 0 0 0 x x x x 0 0 1 mov.f32
1 0 0 0 1 x x x x 0 0 1 p3 mov
1 0 0 0 x x x x x 0 1 x mov.f32
1 0 0 0 x x 0 0 0 1 0 x mov.u32
1 0 0 0 x x 0 0 1 1 0 x p0 mov.u32
1 0 0 0 x x 0 1 0 1 0 x p1 mov.u32
1 0 0 0 x x 0 1 1 1 0 x p2 mov.u32
1 0 0 0 x x 1 0 0 1 0 x p3 mov.u32
1 0 0 0 x x 1 0 1 1 0 x !p0 mov.u32
1 0 0 0 x x 1 1 0 1 0 x !p1 mov.u32
1 0 0 0 x x 1 1 1 1 0 x Pn mov.u32
1 0 0 0 x x x x x 1 1 x invalid
predicate modifier1 sub_pred 5 4 3 value
1 0 0 1 x x x 0 0 x invalid
1 0 0 1 x x x 0 1 x invalid
1 0 0 1 0 0 0 1 0 x mov.u32
1 0 0 1 0 0 1 1 0 x p0 mov.u32
1 0 0 1 0 1 0 1 0 x p1 mov.u32
1 0 0 1 0 1 1 1 0 x p2 mov.u32
1 0 0 1 1 0 0 1 0 x p3 mov.u32
1 0 0 1 1 0 1 1 0 x !p0 mov.u32
1 0 0 1 1 1 0 1 0 x !p1 mov.u32
1 0 0 1 1 1 1 1 0 x Pn mov.u32
1 0 0 1 x x x 1 1 x invalid

Examples:

predicate 101

Instructions:

Encoding

7 6 5 4 3 2 1 0
opcode1 predicate
1 1 1 1 1 1 0 1
7 6 5 4 3 2 1 0
modifier1 opcode2
x x x x
7 6 5 4 3 2 1 0
modifier2 opcode3
x x x x x x
7 6 5 4 3 2 1 0
? ? ? ? ? ? ? ?

Notes:

predicate does not apply to all instructions

Fields

opcode2:

predicate modifier1 opcode3 modifier2 5 4 3 value
1 0 1 0 0 0 0 0 0 !p0 ba
1 0 1 0 1 0 0 0 0 !p0 mov
1 0 1 0 0 1 0 0 0 !p0 ba.savelink
1 0 1 0 1 1 0 0 0 !p0 mov
1 0 1 0 0 x 0 0 1 mov.f32
1 0 1 0 1 x 0 0 1 !p0 mov
1 0 1 0 x x 0 1 x mov.f32
1 0 1 0 x x 1 0 x mov.f32
1 0 1 0 x x 1 1 x invalid
predicate modifier1 5 4 3 value
1 0 1 1 0 0 x invalid
1 0 1 1 0 1 x invalid
1 0 1 1 1 0 x mov.f32
1 0 1 1 1 1 x invalid

Examples:

predicate 110

Instructions:

Encoding

7 6 5 4 3 2 1 0
opcode1 predicate
1 1 1 1 1 1 1 0
7 6 5 4 3 2 1 0
modifier1 opcode2
x x x x
7 6 5 4 3 2 1 0
modifier2 opcode3
x x x x x x
7 6 5 4 3 2 1 0
? ? ? ? ? ? ? ?

Notes:

predicate does not apply to all instructions

Fields

opcode2:

predicate modifier1 opcode3 modifier2 5 4 3 value
1 1 0 0 0 0 0 0 0 !p1 ba
1 1 0 0 1 0 0 0 0 !p1 mov
1 1 0 0 0 1 0 0 0 !p1 ba.savelink
1 1 0 0 1 1 0 0 0 !p1 mov
1 1 0 0 0 x 0 0 1 mov.f32
1 1 0 0 1 x 0 0 1 !p1 mov
1 1 0 0 x x 0 1 x invalid
1 1 0 0 x x 1 0 x mov.f32
1 1 0 0 x x 1 1 x invalid
predicate modifier1 5 4 3 value
1 1 0 1 0 0 x invalid
1 1 0 1 0 1 x invalid
1 1 0 1 1 0 x mov.f32
1 1 0 1 1 1 x invalid

Examples:

predicate 111

Instructions:

Encoding

7 6 5 4 3 2 1 0
opcode1 predicate
1 1 1 1 1 1 1 1
7 6 5 4 3 2 1 0
modifier1 opcode2
x x x x
7 6 5 4 3 2 1 0
modifier2 opcode3
x x x x x x
7 6 5 4 3 2 1 0
? ? ? ? ? ? ? ?

Notes:

predicate does not apply to all instructions

Fields

opcode2:

predicate modifier1 opcode3 modifier2 5 4 3 value
1 1 1 0 0 0 0 0 0 Pn ba
1 1 1 0 1 0 0 0 0 Pn mov
1 1 1 0 0 1 0 0 0 Pn ba.savelink
1 1 1 0 1 1 0 0 0 Pn mov
1 1 1 0 0 x 0 0 1 mov.f32
1 1 1 0 1 x 0 0 1 Pn mov
1 1 1 0 x x 0 1 x invalid
1 1 1 0 x x 1 0 x mov.f32
1 1 1 0 x x 1 1 x invalid
predicate modifier1 5 4 3 value
1 1 1 1 0 0 x invalid
1 1 1 1 0 1 x invalid
1 1 1 1 1 0 x mov.f32
1 1 1 1 1 1 x invalid

Examples: