mesa/src/gallium/docs/source/tgsi.rst

TGSI
====

TGSI, Tungsten Graphics Shader Instructions, is an intermediate language
for describing shaders. Since Gallium is inherently shaderful, shaders are
an important part of the API. TGSI is the only intermediate representation
used by all drivers.


TGSI Instruction Specification
==============================


1  Instruction Set Operations
=============================


1.1  GL_NV_vertex_program
-------------------------


1.1.1  ARL - Address Register Load

  dst.x = floor(src.x)
  dst.y = floor(src.y)
  dst.z = floor(src.z)
  dst.w = floor(src.w)


1.1.2  MOV - Move

  dst.x = src.x
  dst.y = src.y
  dst.z = src.z
  dst.w = src.w


1.1.3  LIT - Light Coefficients

  dst.x = 1.0
  dst.y = max(src.x, 0.0)
  dst.z = (src.x > 0.0) ? pow(max(src.y, 0.0), clamp(src.w, -128.0, 128.0)) : 0.0
  dst.w = 1.0


1.1.4  RCP - Reciprocal

  dst.x = 1.0 / src.x
  dst.y = 1.0 / src.x
  dst.z = 1.0 / src.x
  dst.w = 1.0 / src.x


1.1.5  RSQ - Reciprocal Square Root

  dst.x = 1.0 / sqrt(abs(src.x))
  dst.y = 1.0 / sqrt(abs(src.x))
  dst.z = 1.0 / sqrt(abs(src.x))
  dst.w = 1.0 / sqrt(abs(src.x))


1.1.6  EXP - Approximate Exponential Base 2

  dst.x = pow(2.0, floor(src.x))
  dst.y = src.x - floor(src.x)
  dst.z = pow(2.0, src.x)
  dst.w = 1.0


1.1.7  LOG - Approximate Logarithm Base 2

  dst.x = floor(lg2(abs(src.x)))
  dst.y = abs(src.x) / pow(2.0, floor(lg2(abs(src.x))))
  dst.z = lg2(abs(src.x))
  dst.w = 1.0


1.1.8  MUL - Multiply

  dst.x = src0.x * src1.x
  dst.y = src0.y * src1.y
  dst.z = src0.z * src1.z
  dst.w = src0.w * src1.w


1.1.9  ADD - Add

  dst.x = src0.x + src1.x
  dst.y = src0.y + src1.y
  dst.z = src0.z + src1.z
  dst.w = src0.w + src1.w


1.1.10  DP3 - 3-component Dot Product

  dst.x = src0.x * src1.x + src0.y * src1.y + src0.z * src1.z
  dst.y = src0.x * src1.x + src0.y * src1.y + src0.z * src1.z
  dst.z = src0.x * src1.x + src0.y * src1.y + src0.z * src1.z
  dst.w = src0.x * src1.x + src0.y * src1.y + src0.z * src1.z


1.1.11  DP4 - 4-component Dot Product

  dst.x = src0.x * src1.x + src0.y * src1.y + src0.z * src1.z + src0.w * src1.w
  dst.y = src0.x * src1.x + src0.y * src1.y + src0.z * src1.z + src0.w * src1.w
  dst.z = src0.x * src1.x + src0.y * src1.y + src0.z * src1.z + src0.w * src1.w
  dst.w = src0.x * src1.x + src0.y * src1.y + src0.z * src1.z + src0.w * src1.w


1.1.12  DST - Distance Vector

  dst.x = 1.0
  dst.y = src0.y * src1.y
  dst.z = src0.z
  dst.w = src1.w


1.1.13  MIN - Minimum

  dst.x = min(src0.x, src1.x)
  dst.y = min(src0.y, src1.y)
  dst.z = min(src0.z, src1.z)
  dst.w = min(src0.w, src1.w)


1.1.14  MAX - Maximum

  dst.x = max(src0.x, src1.x)
  dst.y = max(src0.y, src1.y)
  dst.z = max(src0.z, src1.z)
  dst.w = max(src0.w, src1.w)


1.1.15  SLT - Set On Less Than

  dst.x = (src0.x < src1.x) ? 1.0 : 0.0
  dst.y = (src0.y < src1.y) ? 1.0 : 0.0
  dst.z = (src0.z < src1.z) ? 1.0 : 0.0
  dst.w = (src0.w < src1.w) ? 1.0 : 0.0


1.1.16  SGE - Set On Greater Equal Than

  dst.x = (src0.x >= src1.x) ? 1.0 : 0.0
  dst.y = (src0.y >= src1.y) ? 1.0 : 0.0
  dst.z = (src0.z >= src1.z) ? 1.0 : 0.0
  dst.w = (src0.w >= src1.w) ? 1.0 : 0.0


1.1.17  MAD - Multiply And Add

  dst.x = src0.x * src1.x + src2.x
  dst.y = src0.y * src1.y + src2.y
  dst.z = src0.z * src1.z + src2.z
  dst.w = src0.w * src1.w + src2.w


1.2.1  SUB - Subtract

  dst.x = src0.x - src1.x
  dst.y = src0.y - src1.y
  dst.z = src0.z - src1.z
  dst.w = src0.w - src1.w


1.2.4  LRP - Linear Interpolate

  dst.x = src0.x * (src1.x - src2.x) + src2.x
  dst.y = src0.y * (src1.y - src2.y) + src2.y
  dst.z = src0.z * (src1.z - src2.z) + src2.z
  dst.w = src0.w * (src1.w - src2.w) + src2.w


1.2.5  CND - Condition

  dst.x = (src2.x > 0.5) ? src0.x : src1.x
  dst.y = (src2.y > 0.5) ? src0.y : src1.y
  dst.z = (src2.z > 0.5) ? src0.z : src1.z
  dst.w = (src2.w > 0.5) ? src0.w : src1.w


1.2.7  DP2A - 2-component Dot Product And Add

  dst.x = src0.x * src1.x + src0.y * src1.y + src2.x
  dst.y = src0.x * src1.x + src0.y * src1.y + src2.x
  dst.z = src0.x * src1.x + src0.y * src1.y + src2.x
  dst.w = src0.x * src1.x + src0.y * src1.y + src2.x


1.3.4  FRAC - Fraction

  dst.x = src.x - floor(src.x)
  dst.y = src.y - floor(src.y)
  dst.z = src.z - floor(src.z)
  dst.w = src.w - floor(src.w)


1.3.7  CLAMP - Clamp

  dst.x = clamp(src0.x, src1.x, src2.x)
  dst.y = clamp(src0.y, src1.y, src2.y)
  dst.z = clamp(src0.z, src1.z, src2.z)
  dst.w = clamp(src0.w, src1.w, src2.w)


1.3.8  FLR - Floor

  dst.x = floor(src.x)
  dst.y = floor(src.y)
  dst.z = floor(src.z)
  dst.w = floor(src.w)


1.3.9  ROUND - Round

  dst.x = round(src.x)
  dst.y = round(src.y)
  dst.z = round(src.z)
  dst.w = round(src.w)


1.3.10  EX2 - Exponential Base 2

  dst.x = pow(2.0, src.x)
  dst.y = pow(2.0, src.x)
  dst.z = pow(2.0, src.x)
  dst.w = pow(2.0, src.x)


1.3.11  LG2 - Logarithm Base 2

  dst.x = lg2(src.x)
  dst.y = lg2(src.x)
  dst.z = lg2(src.x)
  dst.w = lg2(src.x)


1.3.12  POW - Power

  dst.x = pow(src0.x, src1.x)
  dst.y = pow(src0.x, src1.x)
  dst.z = pow(src0.x, src1.x)
  dst.w = pow(src0.x, src1.x)

1.3.15  XPD - Cross Product

  dst.x = src0.y * src1.z - src1.y * src0.z
  dst.y = src0.z * src1.x - src1.z * src0.x
  dst.z = src0.x * src1.y - src1.x * src0.y
  dst.w = 1.0


1.4.1  ABS - Absolute

  dst.x = abs(src.x)
  dst.y = abs(src.y)
  dst.z = abs(src.z)
  dst.w = abs(src.w)


1.4.2  RCC - Reciprocal Clamped

  dst.x = (1.0 / src.x) > 0.0 ? clamp(1.0 / src.x, 5.42101e-020, 1.884467e+019) : clamp(1.0 / src.x, -1.884467e+019, -5.42101e-020)
  dst.y = (1.0 / src.x) > 0.0 ? clamp(1.0 / src.x, 5.42101e-020, 1.884467e+019) : clamp(1.0 / src.x, -1.884467e+019, -5.42101e-020)
  dst.z = (1.0 / src.x) > 0.0 ? clamp(1.0 / src.x, 5.42101e-020, 1.884467e+019) : clamp(1.0 / src.x, -1.884467e+019, -5.42101e-020)
  dst.w = (1.0 / src.x) > 0.0 ? clamp(1.0 / src.x, 5.42101e-020, 1.884467e+019) : clamp(1.0 / src.x, -1.884467e+019, -5.42101e-020)


1.4.3  DPH - Homogeneous Dot Product

  dst.x = src0.x * src1.x + src0.y * src1.y + src0.z * src1.z + src1.w
  dst.y = src0.x * src1.x + src0.y * src1.y + src0.z * src1.z + src1.w
  dst.z = src0.x * src1.x + src0.y * src1.y + src0.z * src1.z + src1.w
  dst.w = src0.x * src1.x + src0.y * src1.y + src0.z * src1.z + src1.w


1.5.1  COS - Cosine

  dst.x = cos(src.x)
  dst.y = cos(src.x)
  dst.z = cos(src.x)
  dst.w = cos(src.w)


1.5.2  DDX - Derivative Relative To X

  dst.x = partialx(src.x)
  dst.y = partialx(src.y)
  dst.z = partialx(src.z)
  dst.w = partialx(src.w)


1.5.3  DDY - Derivative Relative To Y

  dst.x = partialy(src.x)
  dst.y = partialy(src.y)
  dst.z = partialy(src.z)
  dst.w = partialy(src.w)


1.5.7  KILP - Predicated Discard

  discard


1.5.10  PK2H - Pack Two 16-bit Floats

  TBD


1.5.11  PK2US - Pack Two Unsigned 16-bit Scalars

  TBD


1.5.12  PK4B - Pack Four Signed 8-bit Scalars

  TBD


1.5.13  PK4UB - Pack Four Unsigned 8-bit Scalars

  TBD


1.5.15  RFL - Reflection Vector

  dst.x = 2.0 * (src0.x * src1.x + src0.y * src1.y + src0.z * src1.z) / (src0.x * src0.x + src0.y * src0.y + src0.z * src0.z) * src0.x - src1.x
  dst.y = 2.0 * (src0.x * src1.x + src0.y * src1.y + src0.z * src1.z) / (src0.x * src0.x + src0.y * src0.y + src0.z * src0.z) * src0.y - src1.y
  dst.z = 2.0 * (src0.x * src1.x + src0.y * src1.y + src0.z * src1.z) / (src0.x * src0.x + src0.y * src0.y + src0.z * src0.z) * src0.z - src1.z
  dst.w = 1.0

 Considered for removal.


1.5.16  SEQ - Set On Equal

  dst.x = (src0.x == src1.x) ? 1.0 : 0.0
  dst.y = (src0.y == src1.y) ? 1.0 : 0.0
  dst.z = (src0.z == src1.z) ? 1.0 : 0.0
  dst.w = (src0.w == src1.w) ? 1.0 : 0.0


1.5.17  SFL - Set On False

  dst.x = 0.0
  dst.y = 0.0
  dst.z = 0.0
  dst.w = 0.0

 Considered for removal.

1.5.18  SGT - Set On Greater Than

  dst.x = (src0.x > src1.x) ? 1.0 : 0.0
  dst.y = (src0.y > src1.y) ? 1.0 : 0.0
  dst.z = (src0.z > src1.z) ? 1.0 : 0.0
  dst.w = (src0.w > src1.w) ? 1.0 : 0.0


1.5.19  SIN - Sine

  dst.x = sin(src.x)
  dst.y = sin(src.x)
  dst.z = sin(src.x)
  dst.w = sin(src.w)


1.5.20  SLE - Set On Less Equal Than

  dst.x = (src0.x <= src1.x) ? 1.0 : 0.0
  dst.y = (src0.y <= src1.y) ? 1.0 : 0.0
  dst.z = (src0.z <= src1.z) ? 1.0 : 0.0
  dst.w = (src0.w <= src1.w) ? 1.0 : 0.0


1.5.21  SNE - Set On Not Equal

  dst.x = (src0.x != src1.x) ? 1.0 : 0.0
  dst.y = (src0.y != src1.y) ? 1.0 : 0.0
  dst.z = (src0.z != src1.z) ? 1.0 : 0.0
  dst.w = (src0.w != src1.w) ? 1.0 : 0.0


1.5.22  STR - Set On True

  dst.x = 1.0
  dst.y = 1.0
  dst.z = 1.0
  dst.w = 1.0


1.5.23  TEX - Texture Lookup

  TBD


1.5.24  TXD - Texture Lookup with Derivatives

  TBD


1.5.25  TXP - Projective Texture Lookup

  TBD


1.5.26  UP2H - Unpack Two 16-Bit Floats

  TBD

  Considered for removal.

1.5.27  UP2US - Unpack Two Unsigned 16-Bit Scalars

  TBD

  Considered for removal.

1.5.28  UP4B - Unpack Four Signed 8-Bit Values

  TBD

  Considered for removal.

1.5.29  UP4UB - Unpack Four Unsigned 8-Bit Scalars

  TBD

  Considered for removal.

1.5.30  X2D - 2D Coordinate Transformation

  dst.x = src0.x + src1.x * src2.x + src1.y * src2.y
  dst.y = src0.y + src1.x * src2.z + src1.y * src2.w
  dst.z = src0.x + src1.x * src2.x + src1.y * src2.y
  dst.w = src0.y + src1.x * src2.z + src1.y * src2.w

  Considered for removal.


1.6  GL_NV_vertex_program2
--------------------------


1.6.1  ARA - Address Register Add

  TBD

  Considered for removal.

1.6.2  ARR - Address Register Load With Round

  dst.x = round(src.x)
  dst.y = round(src.y)
  dst.z = round(src.z)
  dst.w = round(src.w)


1.6.3  BRA - Branch

  pc = target

  Considered for removal.

1.6.4  CAL - Subroutine Call

  push(pc)
  pc = target


1.6.5  RET - Subroutine Call Return

  pc = pop()

  Potential restrictions:  
  * Only occurs at end of function.

1.6.6  SSG - Set Sign

  dst.x = (src.x > 0.0) ? 1.0 : (src.x < 0.0) ? -1.0 : 0.0
  dst.y = (src.y > 0.0) ? 1.0 : (src.y < 0.0) ? -1.0 : 0.0
  dst.z = (src.z > 0.0) ? 1.0 : (src.z < 0.0) ? -1.0 : 0.0
  dst.w = (src.w > 0.0) ? 1.0 : (src.w < 0.0) ? -1.0 : 0.0


1.8.1  CMP - Compare

  dst.x = (src0.x < 0.0) ? src1.x : src2.x
  dst.y = (src0.y < 0.0) ? src1.y : src2.y
  dst.z = (src0.z < 0.0) ? src1.z : src2.z
  dst.w = (src0.w < 0.0) ? src1.w : src2.w


1.8.2  KIL - Conditional Discard

  if (src.x < 0.0 || src.y < 0.0 || src.z < 0.0 || src.w < 0.0)
    discard
  endif


1.8.3  SCS - Sine Cosine

  dst.x = cos(src.x)
  dst.y = sin(src.x)
  dst.z = 0.0
  dst.y = 1.0


1.8.4  TXB - Texture Lookup With Bias

  TBD


1.9.1  NRM - 3-component Vector Normalise

  dst.x = src.x / (src.x * src.x + src.y * src.y + src.z * src.z)
  dst.y = src.y / (src.x * src.x + src.y * src.y + src.z * src.z)
  dst.z = src.z / (src.x * src.x + src.y * src.y + src.z * src.z)
  dst.w = 1.0


1.9.2  DIV - Divide

  dst.x = src0.x / src1.x
  dst.y = src0.y / src1.y
  dst.z = src0.z / src1.z
  dst.w = src0.w / src1.w


1.9.3  DP2 - 2-component Dot Product

  dst.x = src0.x * src1.x + src0.y * src1.y
  dst.y = src0.x * src1.x + src0.y * src1.y
  dst.z = src0.x * src1.x + src0.y * src1.y
  dst.w = src0.x * src1.x + src0.y * src1.y


1.9.5  TXL - Texture Lookup With LOD

  TBD


1.9.6  BRK - Break

  TBD


1.9.7  IF - If

  TBD


1.9.8  BGNFOR - Begin a For-Loop

  dst.x = floor(src.x)
  dst.y = floor(src.y)
  dst.z = floor(src.z)

  if (dst.y <= 0)
    pc = [matching ENDFOR] + 1
  endif

  Note: The destination must be a loop register.
        The source must be a constant register.

  Considered for cleanup / removal.


1.9.9  REP - Repeat

  TBD


1.9.10  ELSE - Else

  TBD


1.9.11  ENDIF - End If

  TBD


1.9.12  ENDFOR - End a For-Loop

  dst.x = dst.x + dst.z
  dst.y = dst.y - 1.0

  if (dst.y > 0)
    pc = [matching BGNFOR instruction] + 1
  endif

  Note: The destination must be a loop register.

  Considered for cleanup / removal.

1.9.13  ENDREP - End Repeat

  TBD


1.10.1  PUSHA - Push Address Register On Stack

  push(src.x)
  push(src.y)
  push(src.z)
  push(src.w)

  Considered for cleanup / removal.

1.10.2  POPA - Pop Address Register From Stack

  dst.w = pop()
  dst.z = pop()
  dst.y = pop()
  dst.x = pop()

  Considered for cleanup / removal.


1.11  GL_NV_gpu_program4
------------------------

Support for these opcodes indicated by a special pipe capability bit (TBD).

1.11.1  CEIL - Ceiling

  dst.x = ceil(src.x)
  dst.y = ceil(src.y)
  dst.z = ceil(src.z)
  dst.w = ceil(src.w)


1.11.2  I2F - Integer To Float

  dst.x = (float) src.x
  dst.y = (float) src.y
  dst.z = (float) src.z
  dst.w = (float) src.w


1.11.3  NOT - Bitwise Not

  dst.x = ~src.x
  dst.y = ~src.y
  dst.z = ~src.z
  dst.w = ~src.w


1.11.4  TRUNC - Truncate

  dst.x = trunc(src.x)
  dst.y = trunc(src.y)
  dst.z = trunc(src.z)
  dst.w = trunc(src.w)


1.11.5  SHL - Shift Left

  dst.x = src0.x << src1.x
  dst.y = src0.y << src1.x
  dst.z = src0.z << src1.x
  dst.w = src0.w << src1.x


1.11.6  SHR - Shift Right

  dst.x = src0.x >> src1.x
  dst.y = src0.y >> src1.x
  dst.z = src0.z >> src1.x
  dst.w = src0.w >> src1.x


1.11.7  AND - Bitwise And

  dst.x = src0.x & src1.x
  dst.y = src0.y & src1.y
  dst.z = src0.z & src1.z
  dst.w = src0.w & src1.w


1.11.8  OR - Bitwise Or

  dst.x = src0.x | src1.x
  dst.y = src0.y | src1.y
  dst.z = src0.z | src1.z
  dst.w = src0.w | src1.w


1.11.9  MOD - Modulus

  dst.x = src0.x % src1.x
  dst.y = src0.y % src1.y
  dst.z = src0.z % src1.z
  dst.w = src0.w % src1.w


1.11.10  XOR - Bitwise Xor

  dst.x = src0.x ^ src1.x
  dst.y = src0.y ^ src1.y
  dst.z = src0.z ^ src1.z
  dst.w = src0.w ^ src1.w


1.11.11  SAD - Sum Of Absolute Differences

  dst.x = abs(src0.x - src1.x) + src2.x
  dst.y = abs(src0.y - src1.y) + src2.y
  dst.z = abs(src0.z - src1.z) + src2.z
  dst.w = abs(src0.w - src1.w) + src2.w


1.11.12  TXF - Texel Fetch

  TBD


1.11.13  TXQ - Texture Size Query

  TBD


1.11.14  CONT - Continue

  TBD


1.12  GL_NV_geometry_program4
-----------------------------


1.12.1  EMIT - Emit

  TBD


1.12.2  ENDPRIM - End Primitive

  TBD


1.13  GLSL
----------


1.13.1  BGNLOOP - Begin a Loop

  TBD


1.13.2  BGNSUB - Begin Subroutine

  TBD


1.13.3  ENDLOOP - End a Loop

  TBD


1.13.4  ENDSUB - End Subroutine

  TBD



1.13.10  NOP - No Operation

  Do nothing.



1.16.7  NRM4 - 4-component Vector Normalise

  dst.x = src.x / (src.x * src.x + src.y * src.y + src.z * src.z + src.w * src.w)
  dst.y = src.y / (src.x * src.x + src.y * src.y + src.z * src.z + src.w * src.w)
  dst.z = src.z / (src.x * src.x + src.y * src.y + src.z * src.z + src.w * src.w)
  dst.w = src.w / (src.x * src.x + src.y * src.y + src.z * src.z + src.w * src.w)


1.17  ps_2_x
------------


1.17.2  CALLNZ - Subroutine Call If Not Zero

  TBD


1.17.3  IFC - If

  TBD


1.17.5  BREAKC - Break Conditional

  TBD


2  Explanation of symbols used
==============================


2.1  Functions
--------------


  abs(x)            Absolute value of x.
                    '|x|'
                    (x < 0.0) ? -x : x

  ceil(x)           Ceiling of x.

  clamp(x,y,z)      Clamp x between y and z.
                    (x < y) ? y : (x > z) ? z : x

  cos(x)            Cosine of x.

  floor(x)          Floor of x.

  lg2(x)            Logarithm base 2 of x.

  max(x,y)          Maximum of x and y.
                    (x > y) ? x : y

  min(x,y)          Minimum of x and y.
                    (x < y) ? x : y

  partialx(x)       Derivative of x relative to fragment's X.

  partialy(x)       Derivative of x relative to fragment's Y.

  pop()             Pop from stack.

  pow(x,y)          Raise x to power of y.

  push(x)           Push x on stack.

  round(x)          Round x.

  sin(x)            Sine of x.

  sqrt(x)           Square root of x.

  trunc(x)          Truncate x.


2.2  Keywords
-------------


  discard           Discard fragment.

  dst               First destination register.

  dst0              First destination register.

  pc                Program counter.

  src               First source register.

  src0              First source register.

  src1              Second source register.

  src2              Third source register.

  target            Label of target instruction.


3  Other tokens
===============


3.1  Declaration Semantic
-------------------------


  Follows Declaration token if Semantic bit is set.

  Since its purpose is to link a shader with other stages of the pipeline,
  it is valid to follow only those Declaration tokens that declare a register
  either in INPUT or OUTPUT file.

  SemanticName field contains the semantic name of the register being declared.
  There is no default value.

  SemanticIndex is an optional subscript that can be used to distinguish
  different register declarations with the same semantic name. The default value
  is 0.

  The meanings of the individual semantic names are explained in the following
  sections.


3.1.1  FACE

  Valid only in a fragment shader INPUT declaration.

  FACE.x is negative when the primitive is back facing. FACE.x is positive
  when the primitive is front facing.